At the end of October 2024, the Ecole Supérieure des Forces de Sécurité Intérieure (ESFSI) of the Home Ministry of Tunisia organised the first session of its fifth intensive training on early warning systems & indicators.
This session ran concurrently with a crisis management module, highlighting the interconnected nature of the two disciplines. First, if a warning system fails, a crisis ensues, requiring immediate crisis management decisions and actions. The job of the early warning module is thus to train senior officers to have to use as rarely as possible what is taught in the crisis management module, yet to be ready to do so. Second, as a crisis is managed through decisions and measures, it is crucial to anticipate potential future threats or hazards that may emerge as a consequence of the very management of the crisis. This includes grappling with the intricate domain of unintended consequences. Therefore, comprehending the concept and fundamentals of warning and effectively integrating warning systems and analyses into the crisis management process is essential.
For the October and November session of the training in early warning and indicators, Dr Hélène Lavoix trained senior officers in an intensive 35-hour programme focusing on fundamentals, processes, analysis and practice on issues of interest.
As always the many in-depth and extremely interesting discussions with the trainees and the executive management of the ESFSI, to say nothing of their amazing hospitality, transformed this week in a high level, high quality workshop.
The activity is supported by the European programme “CT-JUST” via Expertise France: “This multilateral and trans-regional programme aims to support regional stability by strengthening cross-border cooperation and the criminal justice system in the fight against terrorism and organised crime.” (News, “The EU-Just-CT project starts its activities in Morocco”, EU Neighbours South, June 2024).
On 30 September 2024, the U.S. Department of Energy (DOE) published the latest edition of the Pathways to Commercial Liftoff: Advanced Nuclear. Its aim is to contribute to accelerate advanced nuclear reactors’ commercial deployment in support of U.S. objectives in terms of nuclear energy, necessary notably to achieve its decarbonization goals. It follows on a first document published in March 2023.
Listen to the article as a deep dive conversation on our podcast, Foresight Frontlines – created with NotebookLM
The U.S. DOE document sets the civilian nuclear objectives the U.S. state would like to achieve, and presents arguments to convince American companies including bankers and other financial institutions that they should invest in nuclear energy.
What is the American nuclear renaissance all about? Why is it important knowing, in 2024, the U.S. is still the first civil nuclear power in the world? How does it fit in with the global return to nuclear power? What does it mean for the American National Interest? How does it compare with China’s surge in nuclear energy? Are the U.S. nuclear power objectives feasible?
First, we analyse the factors, including in terms of national security and international influence, that drive the American nuclear energy renaissance and the resulting objectives set, comparing them with China. Second, we look at the two scenarios the U.S. DOE designed to reach the objective in terms of nuclear energy capacity and envision a third worst-case scenario. Finally, we highlight the uncertainty and challenges the U.S. face in making the American nuclear renaissance a reality.
The drivers behind the American nuclear energy renaissance and objectives
We have three main factors or series of factors driving the American nuclear energy renaissance. Two of them are linked to international relations and American national security and, more classically, the last are related directly to the demand for energy constrained by decarbonisation.
Leading an international context favourable to nuclear energy
The renewed U.S. interest in nuclear energy takes place within a favourable global context, the return of nuclear power on the international stage.
The global nuclear renewal officially started in December 2023 at the COP 28 at Dubai with the pledge by 22 countries, supported by the nuclear industry, to treble nuclear energy by 2050 in the framework of the international efforts to reduce to zero GHG emissions for 2050 (see Helene Lavoix, “The Return of Nuclear Energy“, The Red Team Analysis Society, 26 March 2024). The pledge was jointly announced by both American Special Envoy John Kerry and French President Emmanuel Macron, the U.S. and France being then the first two world actors in terms of nuclear energy capacity. As far as the U.S. are concerned, this pledge followed on the first edition, in March 2023, of the Pathways to Commercial Liftoff: Advanced Nuclear, where the objective to treble U.S. nuclear capacity was set. The U.S., thus, did not only commit globally to the pledge to treble nuclear energy but also asserted their leading role by having taken this decision eight months earlier and now seeing the world follow them.
Then, in March 2024, 33 countries, including the U.S., attended the Nuclear Energy Summit, jointly hosted by the International Atomic Energy Agency (IAEA) and Belgium and signed the new “Nuclear Energy Declaration” reaffirming their strong commitment to nuclear energy (Ibid.).
On 19-20 September 2024 the commitment to nuclear energy was reasserted at the second high level conference of the Nuclear Energy Agency (NEA), “Roadmaps to New Nuclear 2024”, aimed at creating a network of government officials and industry leaders to “inform policy and investment decisions for new nuclear capacity and the long-term operation (LTO) of existing reactors” (NEA, Energy ministers and industry CEOs assemble to advance new nuclear deployment, 30 September 2024).
Considering the heavy investments needed for everything nuclear, financial support will be critical to see the global trebling of nuclear energy taking place. On 23 September 24 during the Climate Week, in New York City, at the “Financing the Tripling of Nuclear Energy – Leadership Event”, a group of 14 global financial institutions expressed their support for the pledge to triple nuclear energy capacity by 2050 (World Nuclear Association, “14 Major Global Banks and Financial Institutions express support to Triple Nuclear Energy by 2050“, 23 September 2024).
There is therefore a global determination to make the renaissance of nuclear energy possible on the terms decided at COP 28.
From the U.S. point of view, it is necessary the country remains at the forefront of this international effort and succeeds in carrying out its own nuclear renaissance. Indeed, the energy transition considering climate change, if current ways of life are to be preserved as much as possible, cannot take place without nuclear energy (see Lavoix, “The Return of Nuclear Energy“). Thus, as leading and “galvanizing the world” for a “clean energy transition” is part of the U.S. National Security Strategy., then the U.S. must lead the nuclear renaissance.
Moreover, the U.S. does not only perceives its international role as one of leadership, but being the world leader is both key for its own security and for the security of the world. This fundamental trend of American foreign policy was once more reasserted in the 2022 National Security Strategy, which stated, for example, that “around the world, the need for American leadership is as great as it has ever been” (President Biden, October 2022 National Security Strategy).
“Out-competing China and Constraining Russia”
Spearheading the renaissance in nuclear energy is even more important for the U.S. that nuclear power is also part and parcel of the battle that opposes the U.S. to China and Russia, and was defined in the 2022 National Security Strategy as “out-competing China and constraining Russia” (Ibid., pp. 23-27). The U.S. – and its allies – must combat what America perceives as the emergence of the enemy order – a new international order carried and shaped by China and Russia. Consequently, America’s enemies are identified; they are Russia and China, and they must fought (e.g. Hélène Lavoix “The American National Interest“, “The War between China and the U.S. – The Normative Dimension).
With an additional focus on the shares of reserves and resources of actors. This powerful tool allows you to visualise and analyse geopolitical influence and company exposure to geopolitics in the uranium mining sector. Its companion report includes 12 use cases and their analysis.
Indeed, to this end, on 17 April 2023, the United States, Canada, France, Japan, and the United Kingdom joined in the Sapporo 5 group, established for “Civil Nuclear Fuel Cooperation”. The members of the group, which will potentially be joined by friendly nations sharing the same vision, will collaborate strategically “on nuclear fuels to support the stable supply of fuels for the operating reactor fleets of today, enable the development and deployment of fuels for the advanced reactors of tomorrow, and achieve reduced dependence on Russian supply chains” (Department of Energy, “Statement on Civil Nuclear Fuel Cooperation Between the United States, Canada, France, Japan, and the United Kingdom“, 17 April 2023). One year later, Sapporo 5 underlined progress, notably regarding uranium enrichment, government-led investments and contracts awarded (Office of Nuclear Energy, “Sapporo 5 Leaders Make Significant Progress in Securing a Reliable Nuclear Fuel Supply Chain“, 18 April 2024).
Hence, nuclear energy, from its production to the entire nuclear fuel supply chain is now linked to U.S. role and influence in the world and to its combat against enemies and the emergence of an enemy order.
Demand for energy constrained by decarbonisation
Finally, the last driver or rather series of drivers for the American nuclear renaissance stems from the direct U.S. demand for nuclear energy, itself driven by the American demand for energy, and more particularly electricity, constrained by the need to decarbonise.
This is similar to what is happening worldwide, as seen in “Why nuclear energy increasingly matters” (in “The Return of Nuclear Energy”). However, in the U.S. case, we must also integrate the necessity to lead the global decarbonisation effort, as seen above, and the specificities of American energy demand.
Meeting energy demand is indispensable to make possible economic growth, especially in energy-hungry, notably electricity-hungry, sectors such as datacenters and artificial intelligence, which are key sectors for American economic expansion. Indeed, DOE, building upon in-depth research, highlights that U.S. electricity demand is likely to “more than double by 2050” (Pieter Gagnon, An Pham, Wesley Cole, et al. (2023), “2023 Standard Scenarios Report: A U.S. Electricity Sector Outlook“, Golden, CO: National Renewable Energy Laboratory; John D. Wilson and Zach Zimmerman, “The Era of Flat Power Demand is Over.”, Grid Strategies, Dec 2023; DOE, 2024 Pathways to Commercial Liftoff: Advanced Nuclear, p.9). It stresses the role of datacenters and artificial intelligence in this surge (2024 Pathways to Commercial Liftoff, p.8).
Meanwhile, in March 2024, Amazon (AWS) and Talen Energy Corp., owner of the 2.5-GW Susquehanna nuclear plant, entered into a deal according to the which Susquehanna plant will supply power to AWS over 10 years (Darrell Proctor, “AWS Acquiring Data Center Campus Powered by Nuclear Energy“, Power, 4 March 2024).
Similarly, Goldman Sachs Research estimates that the U.S. overall energy demand will rise 2.4% between 2022 and 2030, 0.9% stemming from datacenters (Goldman Sachs Insights, “AI is poised to drive 160% increase in data center power demand“, 14 May 2024). “Data centers will use 8% of US power by 2030, compared with 3% in 2022” (Ibid.). McKinsey sees this share to reach 11 to 12% by 2030, which would correspond to an additional 50 GW in energy (Alastair Green et al., “How data centers and the energy sector can sate AI’s hunger for power“, McKinsey 17 September 2024).
Considering the other factors, notably decarbonization imperatives and the need to reduce its cost, nuclear energy is a choice energy to meet this supplementary U.S. power demand, as stressed in DOE’s report (Pathways to Commercial Liftoff: Advanced Nuclear, pp. 9-11).
Besides the direct economic benefit, the energy demand factor is thrice important. First, ‘the expansion of America’s prosperity” is a stated part of the American National Interest (National Defense Strategy 2022 Factsheet). Hence being able to provide the right energy for economic development is fundamental for the U.S..
Second, as we saw, one aspect of the American national security strategy is to “out-compete China”. Economically and technologically, this can only be done with a lot of energy, and thus a lot of decarbonized energy. Furthermore, as we shall see below, China has already embarked on an ambitious civil nuclear programme.
In this race to out-compete China, the U.S. fears to be overtaken in this vital field. Some American future assessment of the PLA’s Strategic Support Force (PLASSF) highlight that “complacency could result in the U.S. falling behind China in intelligentized warfare by as early as 2027, with a likelihood of 93-99%” or by 2030 with the same probability (Col (s) Dorian Hatcher, “Intelligentization and the PLA’s Strategic Support Force“, the Army’s Mad Scientist Laboratory, 5 October 2023).
Thus, being able to serve the energy-hungry technologies, and more particularly those which are absolutely necessary to military intelligentization, becomes fundamental for defense and security, and, relatedly for influence.
These drivers and their interactions frame how the U.S. set its nuclear energy objective, as well as the ways and means that will be used to reach this aim.
Setting the nuclear energy objective
DOE, focusing mainly on energy demand in the context of decarbonisation, highlights the need for 200+ GW of new nuclear capacity in the U.S. by 2050 (2023 and 2024 Pathways to Commercial Liftoff, p.11).
The current U.S. nuclear net capacity is 96,952 GWe (IAEA/PRIS), resulting from “94 nuclear reactors operating at 54 sites”, providing approximately “20% of US electricity generation and almost half of domestic carbon-free electricity” (2024 Pathways to Commercial Liftoff, p. 21).
Thus, the objective is to treble the existing capacity by 2050 to reach approximately 300 GWe (Ibid., p.11), which has become the aim of the global framework for a nuclear renaissance.
Two or three scenarios?
The two scenarios of the U.S. Department of Energy
DOE evokes two scenarios to reach the objective of 300 GWe capacity by 2050 for nuclear energy (2024 Pathways to Commercial Liftoff, p. 39). Scenario 1, which is favoured, starts deployment in 2030, and considers that + 13 GWe per year are necessary. For this scenario to be possible, nuclear units must start being constructed in 2025 (Ibid.).
Scenario 2 starts deployment in 2035, and in that case + 20 GW added capacity per year will be necessary (Ibid.).
For a better understanding of what these scenarios entail and considering the U.S. willingness to “out-compete” China, we present below two charts, comparing the American objective with China’s surge* in terms of nuclear energy capacity. As far as China is concerned, we use the nuclear units under construction, planned and proposed, as known (World Nuclear Association, “Nuclear Power in China“, updated 13 August 2024, incl. IAEA/PRIS data). For planned and proposed units, building on latest known realisations, we attribute varying years for the start of construction, up to 2039, and then count 5 years until connection to the grid.
Comparison of U.S. and Chinese nuclear power capacity estimates and objectives to 2050
Interestingly, the American objective and the Chinese capacity being built lead to very similar results by the mid-2040s, despite China starting from a smaller nuclear capacity. A race could be taking place between China and the U.S.. However, despite the U.S. will to out-compete China, the objective set is more about being on a par with the Middle-Kingdom.
Keeping in mind that the dates given for Chinese planned and proposed nuclear units are tentative, if we look at the evolution of the American and Chinese positions in the development of nuclear energy capacity, and compare the two scenarios, in the first scenario the U.S. remains ahead of China almost for the whole period. The gap closes towards the end of the period, around 2043. In the second scenario, starting in 2031, China moves ahead of the U.S. and remains in the lead until 2039, when the U.S. finally catches up.
What if there were a third scenario?
There is also an unsaid scenario, which emerges if we only take into consideration nuclear units in construction, planned and proposed and not objectives. This is the approach we used in our article “The Future of Uranium Demand – China’s Surge“*. In this scenario, no new nuclear reactor would be constructed. This is a worst case scenario, where the U.S. would fail to generate enough private interest to trigger the massive investments demanded by nuclear energy.
In that case, by 2031, China would lead the world in terms of nuclear energy production. The U.S. would not catch up.
Uncertainty and Challenges
The uncertainty of “incentive-based objectives”
As explained in the DOE report, the U.S. sets objectives, establishes a common normative framework and common knowledge base and then develops incentives that should then incite the private sector to act in such a way that public objectives are met.
If we compare the U.S. approach through “incentive-based objectives” to the Chinese state-led planning, here is what we have in terms of estimates for nuclear energy capacity up to 2050.
In the American case, we have official objectives and yearly targets. However, so far, the reality of nuclear capacity remains quasi unchanged (see also below “An absence of first results when time is short“). The rise in nuclear capacity observed stems solely from the objectives. The level of uncertainty is high.
On the contrary, in the Chinese case, nuclear units being already in construction, it is almost certain that in 2030 China will have caught up with the U.S or be about to do so.
Furthermore, uncertainty is heightened in the U.S. case, because many challenges will need to be overcome.
An aged fleet of nuclear reactors
In September 2024, as seen the U.S. has 94 nuclear reactors operating. However, this fleet is old: “Over 90% of the 2024 US nuclear fleet was constructed in the 1970s and 1980s” (DOE, 2024 Pathways to Commercial Liftoff, p.23). The wave of sustained nuclear plants construction and connection to the grid ended in 1990, with only 5 nuclear plants built and connected to the grid since then (Ibid.).
U.S. DOE 2024 Pathways to Commercial Liftoff – Figure 20: Commercial nuclear capacity and number of reactors commissioned by year
Scenarios 1 and 2 therefore require the licenses of all US nuclear reactors to be renewed when necessary, so that existing reactors can continue to operate. Otherwise, American nuclear power would collapse.
By contrast, the Chinese nuclear fleet is younger. The oldest Chinese reactor, Qinshan 1, saw its construction start in 1985 (WNA, “Nuclear Power in China“, 13 August 2024). It was connected to the grid in 1991 (Ibid.). Two reactors were built at the end of the 1980s and connected during the early 1990s, seven were started at the end of the 1990s and connected during the first decade of the second millenium and all others, i.e. 46, are post-2000s (ibid.).
Furthermore, as the U.S. nuclear reactors’ fleet has aged, this means that the types of nuclear reactors in service also belong to older generations.
Advanced nuclear reactors include Generation III (Gen III), Generation III+ (Gen III+) and Generation IV (Gen IV) reactors (see for example, WNA, “Advanced Nuclear Power Reactors“, April 2021). Advanced nuclear reactors are safer, with a smaller footprint in terms of various materials – as well as space – used and waste produced, more efficient in terms of fuel and operation (Ibid.). They are also meant to have a lower cost of capital (Ibid.).
The Vogtle reactors – Vogtle-3 (1117 MWe) and Vogtle-4 (1117 MWe), the two latest U.S. reactors built, connected to the grid in November 23 and March 24, are Gen III+ reactors. They are the only ones of this kind in the country and the U.S. has no Gen III reactor. By comparison, in China, out of 56 operating reactors (IAEA/PRIS), 14 reactors are Gen III and 2 are Gen III+ (WNA, “Nuclear Power in China“, 13 August 2024). Furthermore, in December 2021, China also connected to the grid the first ever Gen IV small modular reactor.
As a result, in America, the resurrection of reactors that were retired early may help the growth of produced GW, but with a price to pay.
Resurrecting reactors may be done only at the margin, when safety is assured. Considering notably efficiency and safety, reopening old nuclear plants cannot replace building new advanced nuclear reactors.
The difficult construction of few advanced reactors and negative perceptions
U.S. perception of the construction of advanced reactors is informed by the construction and then connection to the grid of the Vogtle Units 3 and 4**, the only two successful available cases.
For these two units,”the original budget was ~$14B, while the final cost was approximately ~$32B” (DOE, 2024 Pathways to Commercial Liftoff, p.47). Vogtle Unit 3 was expected to start in 2016, and Unit 4 shortly afterwards (Nuclear Newswire, “Vogtle-4 startup delayed to Q2“, 5 February 2024). They started respectively in November 2023 and March 2024, thus with a 7 years delay.
Then, perceptions of the construction of Western advanced nuclear reactor may not ignore the EPR (initially European pressurized reactor, renamed Evolutionary power reactor) experiences.
France’s Gen III+ EPR at Flamanville (1600 MWe) was 12 years delayed (17 years instead of 54 months, i.e. 4.5 years, planned, construction started in December 2007). The initial cost was 3.3bn €, but finally amounted to 13.3bn € (Anthony Raimbault, “EPR de Flamanville : retour sur les nombreux déboires d’un interminable chantier“, France Bleu, 8 May 2024). The Olkiluoto-3 EPR in Finland was 13 years delayed and to the initial € 3.3 bn budget were added € 10 bn (Jean-Michel Bezat, “Nucléaire : l’Etat français aide Areva à solder le passif de l’EPR finlandais“, Le Monde, 8 juillet 2021). By contrast the two EPRs (Taishan 1 and 2) built in China also met delays, but only 8 years were needed between the start of the construction and first grid connection (WNA, “Nuclear Power in China“, 13 August 2024).
For the EPR, the final cost per MW (8.3 M € per MW) remains lower than for Vogtle units (14.3 M US$), but it is nonetheless higher than what was initially budgeted.
Hence, the perceptions of the construction of Gen III and Gen III+ nuclear plants for Americans and more largely Western actors, those that will also be involved into financing the American Nuclear Renaissance, include the risks of long delays and enormous budgetary drifts.
Because very few reactors were constructed, even though the analysis of the problems has been done, the solutions proposed have not been experimented. As a result, it cannot be demonstrated that the risks have truly been mitigated. For example, the DOE 2024 Pathways to Commercial Liftoff studies at length the difficulties met by Vogtle 3 and 4 and out of this understanding makes recommendations. Yet, so far, these remain recommendations on paper.
Investors and constructors must be convinced that these strategies and recommendations are the right one and sufficient to lead to shorter construction and deployment time and to the respect of initial cost.
Furthermore, because few advanced reactors were built, the full ecosystem that goes with the development of a thriving industrial activity, from workforce, to subcontractor, from foundry to the myriad of skills and competencies involved in the construction of advanced nuclear reactors, could not fully emerge and expand (DOE, 2024 Pathways to Commercial Liftoff, pp. 55-56). This could create unforeseen chokepoints and challenges that only increase uncertainty and the perception of a high-risk activity.
The quest for a new financing model
Considering the perceived high-risk and the heavy investment needs, the American and international private sector so far appears as hesitating.
This reluctance was highlighted in a Financial Times article reporting notably on bankers’ and Big Tech chief scientist officers’ and heads of energy’s discussions (Malcolm Moore and Lee Harris, “Is nuclear energy the zero-carbon answer to powering AI?“, 3 October 2024). For the interviewees, the main factors that could trigger the launch of the construction of nuclear plants were now positive – i.e. government commitments, financial commitments to support new nuclear construction, and demand for nuclear energy. Yet, so far, no one wants to put capital into the activity because of the high-risk perception in terms of years delay and billions over budget (Ibid.).
Within the American private-sector orientated paradigm, actors, including the U.S. state, will have to find a new model for financing nuclear energy and building new reactors if the targets are to be met. The new consortium-like approaches suggested by DOE, besides the array of various incentives in favour of nuclear energy, including loans, programs or tax credits, added to more authoritative actions directed abroad such as sanctions against Russia, could be a way forward or an element of the new model (DOE, 2024 Pathways to Commercial Liftoff, pp.40 and following).
Considering the very long timeline for everything related to nuclear power, if this new approach is not found, or is not efficient enough, then the third scenario or a variation on it may still take place. It is obvious that it is in the interest of the U.S. and of its companies not to see that happening. However, short-termism and the financialization of activities, alongside a quest for rapid growth and profits, may also be too powerful to favour the wise long-term investments, which are at the heart of the nuclear field (e.g. Thomas I. Palley, “Financialization: What It Is and Why It Matters“, Levy Economics Institute, Working Paper 525, 2007).
An absence of first results when time is short
Given the challenges involved, so far, most of the announcements made, such as the agreement between Microsoft and Constellation, have mainly concerned the purchase of energy, rather than the construction of nuclear reactors. Furthermore, they concerned old reactors and not new advanced ones.
Yet, to see the advanced nuclear “commercial liftoff in the US” taking place according to scenario 1, “first orders would need to be placed by ~2025” (DOE, 2024 Pathways to Commercial Liftoff, p.40).
As of 30 September 2024, there is no such order, i.e. “signed contracts, to construct new nuclear reactors in the US” (Ibid.). Only expressions of interests have been registered (Ibid.). Meanwhile, according to the counts of the World Nuclear Association, 13 reactors are being proposed, for a capacity of 0.11 GWe (WNA, “World Nuclear Power Reactors & Uranium Requirements“, 1st October 2024).
The year 2025 will be decisive.
A battle of ideologies
The capacity of the U.S. to see a new model for nuclear energy emerge could have tremendous impact beyond the nuclear field, .
Indeed, as seen above, the American nuclear energy renaissance is also enmeshed with the American National Security Strategy as it seeks to ensure the U.S. international order prevails over the Sino-Russian one. Thus, if the American model proved incapable of achieving the U.S. nuclear renaissance, then not only would the U.S. fail to meet its various objectives, and to remain the world primary player in nuclear energy, but the very ideological model it champions would be questioned. In addition, the U.S. would have to face cascading consequences for energy-intensive American technological development, for example in the field of artificial intelligence with effects on the military, which, in turn, would also negatively impact American influence in the world.
The challenge here is collective and involves the whole of American society. As the world tends to become bipolar again, it will also impact the U.S. allies.
The U.S. nuclear renaissance is thus at once key and challenging in many ways. It faces many hurdles such as an aged fleet of reactors and little experience of nuclear plant construction over the last decades. Meanwhile, the U.S. “incentive-based approach” as public policy model adapted to nuclear energy and to the scale of the effort envisioned has yet to prove itself.
Yet, American needs and objectives are formidable, as, according to DOE, the U.S. must build and connect to the grid twice as much nuclear capacity by 2050, i.e. in 25 years, than what it succeeded doing between 1965 and 2024 (shutdown capacity not taken into account), i.e. in almost 60 years.
Will power and creativity must never be underestimated, especially when intertwined with national and international security. It will be essential to watch closely what happens in the US nuclear energy sector over the next few years, and in particular over the next twelve months, as these will be critical.
Now, to these formidable tasks must be added another key element, the capacity to fuel the future fleet of reactors. The next article will focus on the uranium requirements of the U.S. nuclear renaissance.
Notes
*Compared with the article “The Future of Uranium Demand – China’s Surge“, here, we changed our way to estimate China’s future nuclear capacity. Notably we introduced estimated dates for the start of construction of reactors and the connection to the grid for the planned and proposed nuclear reactors.
**The Vogtle Units 3 and 4 are owned by Georgia Power (45.7%), Oglethorpe Power Corporation (30%), Municipal Electric Authority of Georgia (22.7%) and Dalton Utilities (1.6%).
However, reciprocally Hamas, the Gaza Islamic militant militia, and its allies are using generative AI at strategic ends. It does so to flood social networks with edited and extremely moving pictures, films as well as fictitious images of the Palestinian civil casualties (David Klepper, “Fake Babies, real horror: Deepfakes from Gaza war increase fears about AI’s power to mislead”, AP, 28 November 2023).
In other words, the Middle East is a major area for the emerging AI warfare on conventional battlefields. It is also true in the cognitive and performative war dimensions and in the AI technology race. This begs the question of the consequences of these new technologies on the evolution of warfare.
Reciprocally, the question arises of knowing if the militarization of AI may become a driver of new risks of uncontrolled escalation?
Israel and AI warfare
AI on the battlefield(s)
AI is ubiquitous in the wars and battles that Israel wages in Gaza and in Lebanon, while activating its multilayer air defense systems constituted by the Iron Dome, David’s frond and the Arrow. Indeed, the Israeli offensive is a mix of conventional urban warfare and of an intense bombing campaign.
Military AIs, known as “Gospel”, “Lavender” and “Where’s Daddy ?”, are producing target lists (Connor Echols, “Israel using secret AI tech to target Palestinians”, Responsible Statecraft, April 23, 2024). The rhythm of this target’s production is extremely high and can reach a hundred targets a day. Human validation being extremely quick, the AI generation of targets imposes a constant rhythm of bombing (Yuval Abraham, ““Lavender”: the AI machine directing the bombing spree in Gaza”, 972, April 3, 2024).
With an additional focus on the shares of reserves and resources of actors. This powerful tool allows you to visualise and analyse geopolitical influence and company exposure to geopolitics in the uranium mining sector. Its companion report includes 12 use cases and their analysis.
In the case of the “Gospel”, a machine learning algorithm estimates the probability of the presence of a Hamas fighter or official in a house or a building at certain times of day. “Lavender” and “Where’s daddy” estimate probabilities regarding the schedule of the movements of a Hamas member, at work or with his family. However, the level of errors of those AIs is as high as 10% (Connor Echols, ibid).
So, the targeting process defines the rhythm of the bombing process. This military version of the “fourth industrial revolution” induces the conversion of numerous civilians into “collateral damages”. The risk is even greater if they are part of the 10% “margin error” (Noah Sylvia, “The Israel’s Defense Force use of AI in Gaza: a case of misplaced purpose”, Royal United Services Institute, 4 July 2024 and Yuval Abraham, ibid ).
In other terms, AI appears as being an unquestionable “force multiplier”, because it bestows upon Tsahal a twin capability of precise targeting, in such quantities that the series of targeted bombing acquire a “mass destruction” quality.
However, if, in the Israeli case, AI is a conventional “force multiplier”, it appears that very advanced military technology can still be mitigated. Older and conventional forms of battlefield preparations, such as tunnels and combat uses of urban landscape, are still quite efficient (Carlo J.V Caro, « Unpacking the history of Urban warfare and its challenges in Gaza », Stimson Centre, 17 October, 2023, John Keegan A History of Warfare, 1993, Edward Luttwak, Strategy, the Logic of War and Peace, Harvard University Press, 2002).
New technologies and the shock of the old
That is why after 11 months of high intensity war, the Hamas militia retains a military capability in Gaza: its fighters have been using the immense labyrinth of underground tunnels as a disruptive battlefield. Those narrow structures are forcing Israeli’s units to loose their cohesion and their firepower.
As it happens, those bombings are also a driver of the “paradoxical logic of strategy”, that, for example, has the capability to turn the race towards victory into failure (Edward Luttwak, Strategy, the Logic of War and Peace, Harvard University Press, 2002). Indeed, the pointillistic mass destruction of the Gaza urban landscape transforms the city into an impassable maze (Jean-Michel Valantin, “The war in Gaza and China’s pivot to the Middle East”, The Red Team Analysis Society, November 22, 2023)..
Thus, it appears that the AI generated bombing rhythm and scale has unintended military consequences. Indeed, as seen in each and every urban theatre of operations, they reinforce the level of difficulty inherent to the military penetration of an urban landscape ( David Kilcullen, Out of the Mountains, The coming age of urban guerrilla, Hurst and Company, 2015).
Thus, the destroyed urban landscape becomes a driver, among others, of the lengthening of the war. This time factor plays in favour of the Hamas, especially through the AI-based strategy of performative and cognitive warfare of the Islamic militant group.
AI Warfare and the diffusion of (Hamas) AI technostrategies
If AI allows the Israeli military to benefit from a force multiplier on the physical battlefield, generative AI opens up the cyberspace to performative and cognitive warfare.
Performative/cognitive warfare
Indeed, since November 2023, there is a flood of edited videos surging on social medias depicting the terrible sufferings of the civilian population of Gaza. Those contents are duplicated from one platform to the other. It is the case, for example from the Chinese TikTok to the U.S. X/Twitter (Jean-Michel Valantin, “The war in Gaza and China’s pivot to the Middle East”,The Red Team Analysis Society, November 22, 2023 and Matthew Ford and Andrew Hoskins, Radical War, Data, Attention and Control in the 21st Century, Hurst Publishing, 2022).
As it happens, the bombings in Gaza also shock and mobilize Arab opinions as well as many people staggered by the dreary conditions of the civil population of Gaza. Specifically in the Palestinian case, those collective emotions mingle with the painful problem of the Palestinian issue, still “unresolved” after almost 75 years of conflict (Avi Shlaim, The Iron Wall, Israel and the Arab World, Penguin Books, 2014).
Those video streams feed collective reactions, such as the massive pro-Palestinian protests throughout Europe and the Middle East. All these reactions are interacting with the Hamas videos and expand its reach and its scale of hyper object. As a result, throughout 2023 and 2024, the more the Israeli bombing and attacks have created victims, the more they have reinforced the anti-Israeli protests (“Global protests in support of Palestinians, rallies for hostages trapped in Gaza”, Reuters, October 22, 2023).
This strategy then is prolonged by the flow of images, commentaries and interpretation of these online video streams at a global scale. Indeed, those video streams hybrid themselves with the explosive content of the political and affective collective memories of the Palestinian history “versus” the Israeli and Jewish history.
The information war strategy of the Hamas triggers an enormous and emotionally turbo-charged “conflict of interpretation” for these video streams, that infuses and immerses through constant dialectics the different levels of the political and military decision-making processes (Man, the State, and War: A Theoretical Analysis by Kenneth N. Waltz, New York, Columbia University: 1959).
Thus, in itself, this performative/political efficiency infuses the public opinions all around the world with the images of the Gaza war. Those images trigger very painful emotions in the population.
Generative AI and cognitive battlefield
In order to reinforce the impact of their performative strategy, the Hamas and its allies use generative AI. This innovative technology allows to produce fake pictures and victims, which are streamlined in the real videos. This editing approach is an « emotional force multiplier » (David Klepper, “Fake Babies, real horror: Deepfakes from Gaza war increase fears about AI’s power to mislead”, AP, 28 November 2023).
This strategy is tantamount to a cognitive warfare strategy. Indeed, the global diffusion of these videos on social medias is turning them into cognitive and emotional ammunitions. Furthermore, through the use of individual smartphones, those cognitive ammunitions impact millions of individual brains and psyches (Annamaria Sabû, Gabrielas Anca, “Using artificial intelligence tools for obtaining cognitive warfare advantage”, The Defence Horizon Journal, October 2023, 2023).
Thus, the very logic of social networks becomes a « cognitive force multiplier » that turns Hamas into a « performative and cognitive warfare great power ». With such tools, Hamas is now waging a cognitive war against Israel. This strategy is a new way to wage « political warfare » through the influence that digital and AI tools confer to its users.
Technology, violence and war
The consequences of the very rapid integration of these new technologies to the management of war need to be understood very quickly. Indeed, since the 19th century, the nexus of science/industry/military/war leads to large-scale transformations in the levels and scale of violence and intensity of war.
This discrepancy is inherent in the unforeseen interactions between mass armies and industrial destructive capabilities. This logic was magnified and pushed to the extreme during the Second World War, ending in the use of the very first nuclear bombings on Hiroshima and Nagasaki (Niall Ferguson, The Pity of War, Explaining World War I, Basic Books, 2000 and The War of the World, History’s Age of Hatred, Allen Lane, 2006).
These examples reveal the way new technologies may inject new levels of violence in war. Thus, they trigger accelerated rates of escalation, while inflicting very high levels of damages and mass destruction.
Nowadays, the military uses of AI seem to trigger the very same logic. This logic manifests through the rising conflict between Israel, the Lebanese Hezbollah militia and Iran.
AI and escalation: Hezbollah
While the Gaza war was dragged on from 2023 to 2024, the Lebanese Hezbollah militia started shifting from missiles and rockets to drones in its attacks on Israel.
Hezbollah and the AI arms race
Since June 2024, Hezbollah has launched hundreds of drone attack upon the Israeli territory. Hezbollah uses Ababil-B acts as a loitering munition. It is able to change trajectories, making it very difficult for the Israeli multi-layered air defence system of the Iron Dome and David’s sling to intercept them (Ari Cicurel, Yoni Tobin, “Hezbollah’s new drone threat to Israel”, The Jewish Institute for National Security of America / JINSA, July 2, 2024, Bassem Mroue, “The threat Israel didn’t foresee: Hezbollah’s drone power”, AP, August 9, 2024).
Hezbollah also uses the Iran-made Shaheed drones. Those are equipped with GPS and manoeuvrable capabilities. They embark missiles and launch them while already flying over Israel. Thus, the Israeli air force and air defence systems have to disrupt both drones and missiles (Cicurel and Tobin, ibid).
This doubling of Hezbollah’s capability to penetrate the Israeli’s air protection through the association of drones and missiles leads to a greater number of strikes. These new weapons also indicate that the Shiite militia enters in the age of AI war technologies. Thus, it is entering into an arms race with Israel.
It also mobilizes the Israeli AI-piloted air defence system, as well as fighter planes and the fleet of combat helicopters. This situation induces massive costs. For example, during the 13 to 14 April 2024 night, Iran, the Hezbollah’s patron state, launched a mammoth 300 missiles and drones strike against the Hebrew state.
Meanwhile, the Israeli counter strikes cost more than 1 billion dollars to Jerusalem.
In other words, a full and repetitive use of the air defense systems is a way to harm financially the technologically dominant Israeli system. As it happens, the Houthis develop an analogous strategy in the Red Sea when they attack the U.S., UK, French and Israeli navies (Jean-Michel Valantin, “Apocalypse in the Red Sea- Anthropocene wars 9”, The Red Team Analysis Society, 20 February, 2024).
Networks of pagers, swarms of bombs
Facing this new Hezbollah’s techno-strategic threat, the Israeli military escalates in another domain. This happens through the ultra-precise targeting of the Hezbollah’s command structure. This leads to a series of impressively targeted strikes. The more notable are the simultaneous detonations of the 2100 pagers of the militia’s commanders, maiming or killing their holders (Jonathan Saul, Steven Scheer and Ari Rabinovitch, “Hezbollah’s pager attack puts spotlight on cyber warfare unit 8200”, Reuters, September 20, 2024).
Then, on 27 September 24, the Israeli air force launched several “bunker buster” bombs on Lebanon. Those weapons were piloted by JDAM / “smart bomb” devices. The attack killed Hassan Nasrallah, the political leader of Hezbollah in his underground headquarters in Beirut (Emmanuel Fabian, “Israel confirms bunker-buster bombs used in attack on Nasrallah”, The Times of Israel, 29 September 2024).
The JDAM system (“Joint direct attack munition”) is a guidance system that combines inertial guidance and Global Positioning System and that is compatible with multiple bombing systems (“Joint Direct Attack Munitions”, Military.com). This system turns the bomb into an autonomous inertial weapon system. It can correct its own trajectory with a 5 meters accuracy.
The last generation of JDAM systems integrate AI reinforcement. In 2023, experiments with this innovative evolution of JDAM were aiming at making the bombs work together as a swarm, dubbed “the Golden Horde” (Joseph Trevithic, “Jet-powered JDAM aims to turn bombs into cruise missiles”, The War Zone, 24 October, 2023).
Hypersonic missiles in the Sky
On 30 September, in retaliation to this strike, Iran, the patron state of Lebanese Hezbollah, launched a 180 missiles strong salvo against Israel. This is the second missile strike against Israel since April 2024.
As it happens, this Iranian strike raises a strategic issue: if the Islamic Republic succeeds in achieving its nuclear program, there are reasons to believe that the hybridation of nuclear bombs and hypersonic missiles will not be far away. What will happen then to the regional and international balance of power? So, this technological and strategic race may also become, if it is not already the case, a driver of strategic escalation across the Middle East.
Who will dominate dominant technologies ?
In other terms, the militarization of AI confers a relative advantage to the Israeli military. However, this technological and military wave spreads rapidly across the Middle East. Israel is a world leader in terms of AI development as well as in innovative weaponry. However, the extension of warfare and of theatres of operations in the AI field is driving changes in warfare. Among those, the emergence of cognitive warfare as well as of new generations of offensive weapons have great impact.
This new technological wave drives the development of new forms of strategies playing alternatively or simultaneously in different domains. The military AI wave is starting to have a conventional as well as performative and cognitive “force multiplier” effect in a region already saturated with strategic tensions and conflicts.
It remains to be seen how this technological / strategic evolution is going to impact the rising conflicts, and if governments are going to be able to contain the new forces they unleashed into old conflicts.
Or not.
In which case, technology-driven escalation in the Middle East is going to have to be assessed.
On 28 August 2023, on the Island of Guam, Kathleen Hicks, Deputy Secretary of Defense, gave a speech at a military and media crowd about the need for the U.S. to mass produce drones in order to outsize the Chinese People’s Liberation Army (Deputy Secretary Kathleen Hicks Keynote Adress: “The Urgency to innovate”, US Department of Defense, 28 August, 2023).
The mission of this centre will be to protect NATO and NATO countries against cyberattacks, especially Russian and Chinese. It is worth noting that, during the NATO summit, the Belarus and Chinese militaries were carrying out joint maneuvers in Belarus, close to the Polish border (“China, Belarus start joint miliary drills near Polish border”, Reuters, July 9, 2024.
From 9 to 11 July 2024, the 75-years NATO Summit took place in Washington D.C. Among the various and momentous conclusions of this international, high level, political and military gathering, it was announced that the organization was creating a cyber-defense centre (Brandi Vincent, “NATO seeks to confront the growing pressure of “hybrid war””, DefenseScoop, 16 July 2024).
Meanwhile, Chinese companies are selling drones all around the world, especially to U.S ambiguous allies or U.S. adversaries. Among these are Nigeria, General Haftar’s forces in Libya, Saudi Arabia, Serbia… (Guy Martin, “Italy intercepts Chinese UAVs being smuggled to Libya”, Defence Web, 4 July, 2024 and Jean-Michel Valantin, Jean-Michel Valantin, “China, Saudi Arabia and the Arab AI Rise”, The Red Team Analysis Society, January 31, 2023 and “China, Serbia, AI and the Pincer Movement on Europe”, The Red Team Analysis Society, April 2, 2023).
As it happens, the artificial intelligence (AI) field absorbs the whole drones-robotic field. So, the few U.S. and Chinese examples we have just mentioned are part of the AI militarization dynamic. They are features common to the two great powers.
Reciprocally, it appears that the mounting tensions between the U.S .and China are one of the drivers of the militarisation of AI / intelligentization of the military.
The question therefore arises as to whether the constant strengthening of this technological and strategic trend is leading the United States and China towards direct confrontation.
We also have to ask if and how the militarization of AI would, or will, influence the potentially coming confrontation.
Common ground : intelligentization of the military, preparing for war
In defense of a “good” Skynet
Since 2017, the Pentagon has militarized what Hélène Lavoix defines as “AI power” (“When Artificial Intelligence will Power Geopolitics – Presenting AI”, The Red Team Analysis Society, November 27, 2017). This dynamic follows two tracks. One track is the integration of AI capabilities to weapons systems, to the command and control systems as well as to the “observation / orientation / decision / action” loop (“OODA loop”).
The other track is the building of a mammoth AI infrastructure, called “Joint All-Domain Command and Control” (JADC2) (Sean Carberry, “Special report : Joint All Domain Command and Control”, a journey, not a destination”, National Defense, 07/10/2023). This giant network of AI networks encompasses and integrates the whole of the U.S. military. This means that the Pentagon and the U.S. Army, U.S. Air Force, U.S .Navy and U.S. Space force are part of a single and common architecture (Shelley K. Mesh, “Air Force to host JADC2 industry day with Army, Navy”, Inside Defense, December 21, 2023.
Massive increases of the military budget support this endeavour. For example, the budget of the Department of Defense was of USD 771 billion in 2023, and USD 842 billion in 2024. The Department of Defense (DoD) devotes most of this staggering increase to the integration of innovative technologies, chiefly AI, while facing the costs of supporting Ukraine against Russia. Meanwhile, the Ukraine war becomes a giant laboratory for experimenting military applications of AI (Michael Klare, “Spurring an endless arms race, the Pentagon girds for mid-century wars”, Tomdispatch, April 16, 2023).
The advent of Replicator
It is in this strategic and industrial contact that, on 28 August 2023, Kathleen Hick, assistant-secretary of defense, announced the launch of the “replicator” project. This project aims at offsetting the strategic advantage of the Chinese military “mass” (Deputy Secretary Kathleen Hicks Keynote Adress: “The Urgency to innovate”” US Department of Defense, 28 August, 2023).
To achieve this objective, the project of the assistant-secretary is to mobilize the U.S. production capabilities, in order to produce a “mass” of military drones that will be superior to the whole of the Chinese military. It is important to note that the assistant-secretary was talking on the Island of Guam, which is a major U.S. Navy base in the Pacific. So, the geography of the declaration of assistant-secretary of defense is anchored in a military infrastructure that would, or will be, on the front line of naval operations against China.
The Ukraine Experiment
Since May 2024 the Ukrainian start up Swarmer has developed autonomous combat swarms. As soon as they are produced, the Ukrainian military projects them on the battlefield. However, these new generation of AI piloted drones are also “only” the new batch of robots being sent in the Ukraine cauldron (Max Hunder, “Ukraine rushes to create AI-enabled war drones”, Reuters, 18 July, 2024 and Jean-Michel Valantin, “AI at War (1) – Ukraine”, The Red Team Analysis Society, April 3, 2024).
Given the importance of the U.S. military and of the GAFAM / “Google /Apple / Facebook /Amazon / Microsoft presence in Ukraine, one can safely hypothesize that the Ukrainian and U.S. military and industries widely share the feedback of this “new way of war” (Vera Bergengruen, “How Tech giants helped turn Ukraine into a giant AI war lab”, Time Magazine, February 8, 2024).
Finally, through different channels, the DoD develops a dense relationship with the Silicon Valley AI giants. As we have seen in AI-at War (1) – Ukraine, the U.S. military, notably through the national geo-spatial intelligence agency, already works with Oracle, Palantir and Amazon in order to implement Project Maven (Courtney Albon, “Geospatial-intelligence agency making strides on Project Maven AI”, C4ISR, May 22, 2023, Saleha Mohsin,“Inside project Maven, The US military’s AI Project”, Bloomberg, 29 February 2024).
The latter aims at using the imagery capabilities of the drone fleet in order to feed an electronic and interactive global military mapping of the earth. In the same time, Open AI works with the DoD in order to develop a military-only Chat GPT version. This project will help the military to sort through the gigantic and perpetually growing mass of data and information that the DoD collects through its global network of land, air, space and marine sensors (Jon Harper, “Microsoft deploys GPT-4 Large Language Model for Pentagon use in top-secret cloud”, DefenseScoop, 7 May 2024) .
In parallel as well as in confronting the US militarization of AI, the Chinese People’s Liberation Army is “intelligentizing” itself on a massive scale. As in the U.S., the different armed branches of the PLA integrate AI power. This process runs through the entire organization of the PLA (Nigel Inkster, The Great decoupling: China, America and the struggle for technological supremacy, Hurst, 2021). As the intelligentization of the Chinese OODA Loop is underway since 2018, other programs are implementing AI in order to pilot strategic weapons systems such as hyper sonic missiles (Jean-Michel Valantin, “Militarizing Artificial Intelligence – China (2)”, The Red Team Analysis Society, May 22, 2018 and Christopher Mc Fadden, « China uses cheap AI chip to control hypersonic weapons, boosting range », Interesting Engineering, 17 April, 2024).
Towards a Chinese Skynet ?
As it happens, the Chinese military is also developing its own AI architecture that should integrate the whole of its different services. This Multi-domain precision warfare (“MDPW”) could appear as a mirror construct of the US JADC2. However, there is a fundamental offensive component at the core of this Chinese intelligentization concept. As it happens, the MDPW aims at breaking the information flows that link the different U.S. armed branches through the JADC2 architecture (Kris Osborne, “China’s new multi domain precision warfare” operational concept”, RealClear Defense, 26 October, 2023).
Both the U.S. and Chinese AI militarization processes are nothing but a new and profound military revolution. Those new ways and means of battle and war management necessitate intense and constant training in order to achieve the proper integration of deployment and combat AI architectures. However, if the U.S. and Chinese military both share this goal, they have to face very different predicaments.
AI or Cultural (U.S.) Revolution ?
For the U.S. military, the main challenge is the implementation of the AI architecture by each of the armed branch. The AI capabilities of each service needs to be fully integrated to the overall DoD JADC2 architecture. Thus, each armed branch, and each element of them, would become an element of a common military. This may turn the mammoth U.S. military apparatus into a giant, fully coordinated modular weapons system (Michael Klare, “AI versus AI, and human extinction as collateral damage”, Tomdispatch, 11 July, 2023).
The sheer technical complexity of this endeavour is reinforced by the very history of the U.S. military. As it happens, its four components, the U.S. Army, the U.S. Navy, the U.S. Air Force and the U.S. space force (2019), have evolved in largely autonomous ways. Their coordination has always been a major problem for the national U.S. command and control (Alfred W. McCoy, In the Shadows of the American Century, The Rise and Decline of US Global Power, Haymarket Books, 2017).
Revolution is not enough
The Revolution in Military Affairs of the 1990s has been an important medium of inter-services jointness. It was the result of the common integration of space power and of information technologies. However, the four military branches are still largely autonomous kingdoms (Alfred W. McCoy, ibid).
Adding another layer of difficulties, the deployment of AI systems comes as a profound cultural shock. For example, in the U.S. Air Force, simulations of air combats opposing AI pilots and human pilots took place in 2021. After two dozen fictive fights, the AI had reached clear combat dominance ((Kenneth Payne, I, Warbot, the dawn of artificially intelligent conflict, London, Hurst, 2021 and Stephen Losey, « US Air Force stages dog-fights with AI-flown fighter jets », Defense News, 19 avril 2024).
Then, in 2023, training took place at “Top Gun” air school, opposing human pilots to AI piloted aircrafts. The fact that the U.S. Air Force did not disclose the results entails several hypotheses. One of those is that the AI pilots outperformed the human ones. The U.S. Air Force is deeply centred on the pilot community (Losey, ibid). So, those results may trigger a profound technological as well as social revolution.
These U.S. achievements are part of the wide arc of the U.S. military experience with AI and drones. As it happens, the U.S. military capital of war experience is ever growing. It is the consequence of its many involvements in numerous theatres of operations: among others, in Afghanistan, Irak, Syria, Jordan, Yemen, Colombia, the Sahel region… (Roberto J. Gonzalez, War Virtually, the Quest to Automate Conflict, Militarize Data, and Predict the Future, Oakland, University of California Press, 2022).
The PLA’s Race
On the Chinese side, the People’s Liberation Army must overcome its own double challenge. The first challenge is modernization through the triple dynamic of mechanization/ informatization/ intelligentization. This process started at the beginning of the 2010s and necessitates a massive political, scientific, technological, and industrial mobilization. In order to guarantee its sustainability, this modernization dynamic is anchored in the civil-military fusion policy (Nigel Inkster, ibid).
The Central Military Commission of the Chinese Communist Party implements this policy. The latter runs through the military-civil relationship established between the People’s Liberation Army (PLA) and civil research-development laboratories and industrial companies (Elsa B. Kania in “The PLA’s trajectory from informatized to “intelligentized” warfare”, The Bridge, June 8, 2017).
This militarization of AI leads us to wonder about the consequences of this process. Indeed, this dynamic affects the tactical and operational levels. But is is also true in terms of strategy, i.e. the level where political, economic and strategic interests intersect. That is why the CCP’s Central Military Commission equips itself with AI, in order to create an information loop with the People’s Liberation Army AI networks (Nigel Inkster, ibid).
Boots on the ground(s) or no boots on the ground(s)
The other challenge met by the intelligentization of the PLA is the lack of warfare experience. The U.S. military has been very active since 1945. There were massive force projections for the Korean war (1950-1952), then for the Vietnam war (1963-1975). Then followed the Gulf War (1990-1991), the Afghan War (2001-2021), the Iraq War (2003-2010). Those latter were part of the “Global War on Terror (2001-2021)”. There were also numerous U.S. deployments in numerous “small wars in faraway places” in Latin America, Asia, the Middle East and Africa…. (Michael Burleigh, Small wars, Faraway places : global insurrection and the making of the Modern World, 1945-1965, and Alfred W. Mc Coy, ibid).
As it happens, the Chinese PLA did not accumulate such a capital of war experience. Its last two major military involvement were the Korean War and the sorry affair that was the Sino-Vietnamese War of 1979 (David Kilcullen, The Dragons and the Snakes, How the Rest Learned to fight the West, Hurst, 2020).
In order to offset this training disadvantage, the PLA projects drones units on active theatres of operations. For example, in Libya, Chinese drone units support the coalition gathered by general Haftar (Jon Mitchell & Hélène Lavoix, « The Libya Series », The Red Team Analysis Society). Thus, they expand the (AI) fire power of the coalition and inflict significant losses to the Islamist coalition.
Turkey supports the latter through the projection of Daesh mercenaries as well as of Turkish Bayraktar drone units (Alex Gatopoulos, “”Largest drone war in the world”: how air power saved Tripoli”, Al Jazeera, 28 May, 2020, Dale Aruf, « China’s tech outreach in the Middle East and North Africa », The Diplomat, 17 novembre 2022 and « Italia seizes chinese-made military drones destined for Libya », Reuters, July 2, 2024). (As it happens, Ukraine buys the same Turkish drones and uses those against the Russian forces (Agnes Helou, “With Turkish drones in the headlines, what happened to Bayraktar TB2?, Breaking Defense, October 6, 2023 ).
The Great (drones) Flood
As it happens, China sells drones, especially of the CASC “Rainbow” and Wing Loong series, all over the Middle East and the Persian Gulf. That is the case, for example in Iraq, Yemen, the United Arab Emirates, Saudi Arabia, Egypt, Nigeria (Dale Aruf, ibi and Kris Osborn, ibid). It is also the case in Serbia. So, when those drones are used in combat situation, their performance inform the Chinese political and military authorities about their tactical and strategic usefulness (Jean-Michel Valantin, “China, Serbia, AI and the Pincer movement on Europe”, The Red Team Analysis Society, April 2, 2023).
One should also highlight that the Chinese drones are largely sold to adversaries of the U.S. Reciprocally, the U.S. current interventions in the Middle East target allies and supporters of China. In other words, it appears that the fast intelligentization of both U.S. and Chinese militaries are driven by as well as driving the rising tensions between the two great powers.
Indeed, it appears that AI is militarized along the same principles as former scientific, technological revolutions. It is true for the bow, the saddle with stirrups, wall building, the wheel, wind sailing, cast iron, black powder. After the industrial revolution, it was also true, among others, of the steam engine and of chemistry (John Keegan A History of Warfare, 1993).
During the 20th century, there was a massive militarization of internal combustion engines, electricity, electronic communications, aviation, nuclear power, computers and space travel. This trend is especially true for the first centres of development of those technologies, from which they spread. On a regular basis, scientific and military authorities work together at militarizing new technologies. Indeed, they appear as a technological and strategic high ground (Ian Morris, War, what is it good for ? : Conflict and progress of civilization from primates to robots, Farrar, Strauss and Giroux, 2014).
This advantage, be it tactical or strategic triggers the “paradoxical logic of strategy” (Edward Luttwak, Strategy, the Logic of War and Peace, Harvard Unoversity Press, 2002). As it happens, adversaries of the technological dominant power emulate that very advantage. So, the dissemination of the technological wave and its acquisition by competitors turns them into potential competitors.
Thus, the technological-military advantage tends to weaken through its very development. However, this competitive dissemination is also the driver of technological and military escalation. Indeed, each competitor tries to dominate with the new military use, or threat of use, of the newly militarized technology. In the context of the rising U.S.-China competition, it is this very strategic logic that is a driver of the militarization of AI in both great powers.
Nowadays, this logic of strategic escalation is at work through both the process of the U.S. and of the Chinese militarization of AI. It now remains to be seen if the mix of AI power and of the China-U.S. rivalries emerges in other strategic domains as, for, example, agriculture?
Article updated to include October 2024 and early November 2024 events.
On 19 June 2024, Niger has revoked French nuclear state company Orano’s mining permit for the Imouraren mine. This means that Orano – and thus France – loses 47% of its uranium reserves, when an era of renewal for nuclear energy starts globally and when France plans to add between three pairs of EPR2 reactors, up to 14 and even 20 new EPRs to its current nuclear park (see below). What happened and what is at stake for France?
France still is the second power worldwide in terms of nuclear energy generating capacity (Helene Lavoix, The Future of Uranium Demand – China’s Surge, The Red Team Analysis Society, 22 April 2024). Orano still is the third company globally for the nuclear fuel cycle (Helene Lavoix, Revisiting Uranium Supply Security – 1, The Red Team Analysis Society, 21 May 2024). As a result, France plays a leading role in the current global evolution towards the renewal of nuclear energy (Helene Lavoix, “The Return of Nuclear Energy“, The Red Team Analysis Society, 26 March 2024). Furthermore, nuclear energy is vital for the country, with nuclear power accounting for 62.6% of electricity in 2022 (IAEA-PRIS – 28/04/2024; Helene Lavoix, Revisiting Uranium Supply Security – 1, The Red Team Analysis Society, 21 May 2024).
However, uranium is needed to power nuclear plants. Despite appearances, France was well positioned in terms of uranium supply thanks to Orano’s overseas mining permits (Lavoix, Revisiting Uranium Supply Security – 1). Yet, as a result, French uranium security supply is also more fragile than it would be if uranium mines were located on its territory (Ibid.). Considering the specificity of France’s uranium supply, geopolitics and influence are key to secure that supply, as illustrates the catastrophe Orano now faces in Niger.
First, we address the situation in Niger and explain the threat now materialised surrounding Orano’s mining permit for Imouraren. As Niger’s Imouraren mine represents up to 47% of Orano’s uranium reserves and its loss will likely degrade France’s position globally as well as Orano’s, second, we focus on the stakes for France in terms of uranium supply and and underline the role that geopolitics plays now and in the future for uranium and thus nuclear energy.
Losing Orano’s Imouraren mining project in Niger
France loses its hold in the difficult political and geopolitical Nigerien context
Mining operations in Niger take place in a complex context.
The 26 July 2023 coup in Niger had multiple impacts (e.g. Gilles Yabi, “The Niger Coup’s Outsized Global Impact“, Carnegie Endowment for Peace, 31 July 2023). Notably, it soured greatly diplomatic relations between France and Niger. The French Embassy in Niger was closed on 2 January 2024 (French Ministry of Foreign Affairs). All French troops had to leave the country the end of December 2023, after France was similarly asked to leave Mali then Burkina Faso (France 24, “Last French troops leave Niger, ending decade of Sahel missions“, 22 December 2023).
The eviction of France from Sahel results from difficulties with the peace-keeping and stabilising missions there, then used by Russia against France as a consequence of France’s decision to side with the U.S. in Ukraine (e.g. Aja Melville, “Russia Exploits Western Vacuum in Africa’s Sahel Region,” Defense and Security Monitor, 22 April 2024). Russia retaliated strategically in a flanking manoeuvre, hitting France by further degrading its influence and replacing it in Sahel (e.g. Ibid., Fatou Elise Ba, “L’aide publique et humanitaire de la France n’est plus la bienvenue au Mali“, IRIS, 16 February 2023)
Thus, Russia’s influence is on the rise, as elsewhere in the region (e.g. Olumba E. Ezenwa and John Sunday Ojo, “Russia has tightened its hold over the Sahel region – and now it’s looking to Africa’s west coast“, The Conversation, 29 April 2024). For example, on 26 March 2024, the Russian President Vladimir Putin and the President of the National Council for the Safeguard of the Homeland of the Republic of Niger Abdourahamane Tchiani “expressed determination to step up political dialogue and develop mutually beneficial cooperation in various spheres” (Kremlin website). Then, in April 2024 military advisors from the ex-Wagner group renamed African Corps arrived in Niger’s capital Niamey (AFP, “Russian military instructors, air defence system arrive in Niger amid deepening ties“, France 24, 12 April 2024).
After a day of silence, on 20 June, Orano issued a press communiqué stating that Nigerien authorities had decided “to withdraw its licence to operate the deposit from its subsidiary Imouraren SA”.
According to the communiqué Orano is ready to continue the discussion as well as to press the matter “before the competent national or international judicial bodies”.
What is at stake for France?
Current supply of uranium is not at stake as production has not yet started at the Imouraren mine.
The concern is for the future and depends on the potential of the mine. Obviously, the larger the uranium mine and the higher its yield, the higher the stakes.
Imouraren and uranium reserves for France
Now, Imouraren, discovered in 1966 by France, is not a small mine nor does it represent small reserves and resources for Orano and thus for France (for explanations on reserves and resources, Lavoix, Revisiting Uranium Supply Security – 1; Nuclear Energy Agency (NEA)/International Atomic Energy Agency (IAEA), Uranium 2022: Resources, Production and Demand (Red Book), OECD Publishing, Paris, 2023, p. 387). Quite the contrary.
Imouraren holds 145.712 tonnes of Uranium in reserves, out of which 95.527 were Orano’s share (2023 Orano Annual report, pp. 35-36). “Production was expected to be 5.000 tU/yr for 35 years” (NEA/IAEA Red Book 2022 p.388). For the sake of comparison, France yearly requirements in 2024 were 8.232 tU (WNA, “World Nuclear Power Reactors & Uranium Requirements“, May 2024). Imouraren could thus have covered alone 60,7% of France 2024 Uranium requirements. Exploiting Imouraren would thus have greatly facilitated projects to increase the number of nuclear reactors in France, as well as ensured trade revenues for Orano, considering other mines. Together, this would have secured Orano and France’s influence in the field.
In terms of reserves, as shown on the series of charts below, the mine of Imouraren in Niger represents 24% of the total uranium reserves and resources in the ground plus inferred resources of Orano, i.e. the largest segment (in tonnes of Uranium, i.e. considering the varying yield of each mine). The share of Imouraren is larger if we only consider total reserves and resources in the ground, i.e. 32 %, and even larger if we only take into account reserves, i.e. 47%.
In other words, the further away we are in time, the more potential uranium supply outside Niger Orano holds. This is a tribute to the company’s exploration and diversification effort. Nonetheless, we should not forget the uncertainty weighing on Orano’s Mongolian supply since February 2024 (Lavoix, “The Return of Nuclear Energy“).
However, as far as medium term uranium supply is concerned, losing the mine of Imouraren creates a security challenge.
A threat on medium term supply considering plans for nuclear energy
Indeed, changes in a mining portfolio introduce an uncertainty for the future that is all the more important that exploration to find mines, then feasibility studies then plans for operation, before mining and milling can start, are of the long period, as shows the figure below.
Currently, in line with the current objectives of renewal of nuclear energy, France plans to construct and connect to the grid between 6, 14 and even 20 new nuclear power plants.
Finally, in the 3rd Programmation pluriannuelle de l’énergie (PPE), opened for concertation in November 2024, France plans to launch three pairs of EPR2 reactors with decision to invest to be taken by EDF by 2026 , to which should be added support to the development of Small Modular Reactors (SMR) (SFEN, “PPE 3 et SNBC 3 : neuf orientations pour le nucléaire français“, 4 November 2024). Meanwhile, the lifespan of existing reactors will be extended beyond 50 or even 60 years (Ibid.)
Thus, supplementary uranium requirements will be needed, starting approximately in 2035 (if the time needed to build and connect an EPR is approximately 9 years – see Towards a U.S. Nuclear Renaissance?), for yet unknown quantities. This means that by 2029-2030 latest, corresponding decisions to mine must be taken so that new production can start and be delivered in 2035. If ever EPRs were built faster then supplementary uranium requirements would occur earlier.
If France already has mining sites that are ready or quasi ready for the stage corresponding to the decision to mine, then all is well. Imouraren corresponds to this case. Indeed, Orano planned to start a pilot programme there in 2024, with decision to invest in 2028 “if feasibility is confirmed” (World Nuclear News, “Preparatory activities begin at Imouraren“, 17 June 2024).
If no mining site of this type is available, then Orano must rely on sites that are at the stage of in-depth exploration and studies. Because only four to five years are left until 2029/2030, then that means that in-depth exploration, which may last approximately 10 years, must have already started. However, compared with a site where in-depth exploration has already taken place, in that case, uncertainty is stronger.
Furthermore, mining permits will need to be requested and obtained in four to five years time, which heightens uncertainty stemming from geopolitics.
For example, Orano may very well see in-depth exploration yielding excellent results, but in an area where Russian influence is very strong and could be even stronger in four to five years. In such a case, the current international French position regarding Russia, if it remains as such in the future, strongly lowers the probabilities of Orano obtaining any mining permit.
Alternatively, accounting for strong competition over uranium resources, Orano may also lose mining permits to allies, which will have no qualm about thinking about their national interest first. For example, considering American needs in terms of uranium and the U.S. small size of overseas reserves, as the U.S. ranks globally 10 in terms of reserves and resources, it would not be surprising to see the U.S. using strong if not violent methods to secure uranium in the future (Lavoix, The Future of Uranium Demand and Revisiting Uranium Supply Security – 1). One should remember the American attitude when the stakes were face masks during the COVID 19 pandemic or the way the U.S. stole from France the Australian contract for submarines (e.g. Ouest France “Coronavirus. En Chine, une cargaison de masques destinés à la France détournée par des Américains“, 2 April 2020; Helene Lavoix, “The American National Interest“, 22 June 2022).
Similar geopolitical insecurity weighs on exploration licenses to be obtained, as well as on those already granted, as the loss of Imouraren shows.
The threat in a global perspective
Considering the current and future appetite for uranium, what does mean the loss of Imouraren for France in a global perspective?
Taking into account the international context in the Sahel and more particularly in Niger, in the framework of the war in Ukraine, we make the hypothesis that the Imouraren mine permit will be granted to Russian Uranium One through Rosatom in the same conditions as what existed for Orano (same shares – for Uranium One, see Lavoix Revisiting Uranium Supply Security – 1). This hypothesis is increasingly likely considering the 8 November statements of the Nigerien Mining Minister Ousmane Abarchias, according to which “Niger is actively seeking to attract Russian investment in uranium and other natural resources” (RFI, “Niger embraces Russia for uranium production leaving France out in the cold“, 13 November 2024).
We use as basis the graph we developed previously to revisit the security of uranium supply by integrating overseas efforts by mining companies. The data in the graphs comes from extensive research carried out for The World of Uranium: Mines, States, and Companies – Database and Interactive Graph. The initial graph is presented on the left-hand side. We then show the same graph without Imouraren for France, and without the Madaouela deposit for Canada (Canadian GoviEx Uranium Inc. having lost its mining permit on 4 July 2024). In a third graph on the right-hand side, we attributed as scenario the two deposits to Russia.
Comparing the graphs, France falls from 8th to 11th place, behind China, the U.S. and Brazil, and the EU from 7th to 9th place, whilst Russia takes 3rd place, before Kazakhstan, with a large increase in its overseas reserves and resources. Japan goes from rank 14 to rank 15, as two Japanese companies hold stake in French Orano.
Not only is the security of uranium supply degraded, but France’s weight in the world in terms of uranium supply is lessened. Considering future uranium demand this will have a negative commercial impact while diminishing the global influence of both Orano and France.
Knowing that supplying France nuclear power plants is a very important stake considering the French share of nuclear electricity, and that the global development of nuclear energy, notably stemming from China, will intensify global competition, the loss of Imouraren is very bad news.
Indeed, as an update (7 November 2024), since the article was initially written, matters got worse between Orano and Niger. Following growing difficulties impacting Somair (63.4% held by Orano), including an inability to export production, and debts of Nigerien Sopamin (holding 36.6% of Somair) owed to Somair, Orano decided “to suspend its [Somair] activities, as an interim measure, as of the end of October” 2024 (News, “Niger: growing financial difficulties will force SOMAÏR to suspend operations“, Orano, 23 October 2024). In response, ignoring its own responsibility and decisions, Niger, through Sopamin attacked Orano’s decision, reportedly having been neither consulted nor informed. Niger offered to buy 210 t uranium out of the 1000 t held by Somair to help the company to continue its activity. Meanwhile, Niger repeatedly accuses France to carry out overt and covert operations against Niger (e.g. La Tribune, La France déstabilise-t-elle le Niger ?, 3 August 2024; Mathieu Olivier, “La DGSE française dans la tourmente après les accusations du Niger“, Jeune Afrique, 6 November 2024).
The tug of war between France, Orano and Niger goes on. As a result, it is now 58% of overseas French reserves of uranium (44 % of reserves and resources) that have disappeared or are disappearing compared with December 2023. Meanwhile, closing Somair means that the 2000 t U a year that were produced will not be anymore, i.e. for France’s share 1268 t U. This represents approximately 15.4% of France’s yearly requirements in uranium.
In the future, it appears as essential that international relations and foreign policy decisions be taken while also considering nuclear and uranium security, including on the longer term. This means that anticipation becomes even more important than what was already highlighted in the 2023 Rapport de la commission d’enquête of the National Assembly.
Meanwhile, the impact of Orano’s operations on domestic politics in country where it operates, as well as on regional and global geopolitics, must also be integrated into assessments, planning and policy. For example, known adverse dynamics such as the “resource curse” should always be assessed and integrated into analysis to make sure the right strategy to secure permits and operations is designed and implemented (e.g. Mähler, Annegret (2010), Nigeria: A Prime Example of the Resource Curse? Revisiting the Oil-Violence Link in the Niger Delta, GIGA Working Papers, No. 120, German Institute of Global and Area Studies (GIGA), Hamburg).
Our point is not that Orano causes or has caused such adverse dynamics, but that the possibility it happens should be envisioned and taken into account if the risk exist.
Potentially, policies to avert such negative consequences could also be developed, if adequate. Furthermore, such an approach could also reassure host governments and become an argument for influence.
The resource curse theory suggests that countries rich in natural resources, particularly oil and minerals, often experience slower economic growth, weaker democratic institutions, and greater political instability compared to resource-poor nations. This paradox arises from the disruption of fundamental political, societal and economic dynamics that typically balance governance.
One critical breakdown is in the “taxes-for-security exchange” dynamic. In all countries, ideally, political authorities rely on taxation to finance their activities, creating a direct relationship with citizens: governments provide security, public goods, and services in exchange for tax revenues. This dependence fosters accountability, as citizens demand responsible governance in return for their contributions. When the relationship functions, notably, legitimacy is strengthened. These dynamics appear to be naturally at work in resource-poor nations.
In resource-rich nations, however, governments often finance themselves through resource rents (profits from resource extraction), reducing their reliance on taxation. This weakens the social contract between rulers and the governed: the rulers have no interest in providing security to their citizens as their resources do not come from taxes and, as a result, the ruled live in insecurity and are less empowered to hold political authorities accountable. Political authorities become increasingly illegitimate domestically but remain in power out of the support of those benefiting from the resources, most often external to the polity.
Finally the new insecurity on French uranium supply stemming from the loss of Imouraren in Niger, and the tug of war surrounding Somair, highlights a difficult question: Is it secure for France to have a foreign policy that is not one of independence and non-alignment (e.g. Pascal Boniface, “Why the Legacy of De Gaulle and Mitterand Still Matters for the French Public Opinion“, IRIS, 15 March 2021; The Non-Aligned Movement“, Wikipedia).
France’s national interest, with uranium supply at the top of the agenda considering the importance of nuclear energy for the country and the world, should be sought first, before any other point. This may be what the November 2023 and November 2024 reciprocal state visits between France and Kazakhstan, a major supplier and partner in terms of uranium for France, indicate (Elysée, Visite d’État de son Excellence Kassym-Jomart Tokaïev, Président de la République du Kazakhstan, République Française, 5 November 2024; “Kazakhstan’s Tokayev in France: It’s All About Nuclear Energy“, The Times of Central Asia, 6 November 2024). The price to pay for another type of foreign policy could be very high indeed in terms of influence and power and ultimately in terms of uranium thus energy and finally access to electricity in the country.
Conclusion
The importance of politics and geopolitics for uranium supply is once more demonstrated in Niger.
Despite appearance the development of nuclear energy cannot remain the sole preserve of R&D, engineering and industrial planning. Succeeding in securing uranium supply when this security depends on overseas resources will demand to attribute the highest priority to the understanding of international relations and geopolitics and to the design and implementation of strategy. Influence and power, diplomatic acumen, and skilful, bold and original international actions, will become even more essential as the volatile international context unfolds and as the international impact of the renewal of nuclear energy spreads.
In December 2023, twenty two governments and the nuclear industry decided to treble nuclear energy by 2050. If we are to meet this objective, then the corresponding uranium supply must be adequate.
We thus need to assess the present and future supply of each country, in the light of its current and planned future needs, in a political and geopolitical environment which is not anymore peaceful, as has been known since the end of the Cold War, but, on the contrary, increasingly tense and fraught with hostility.
In this article, we focus on assessing the potential supply available per country. First, we present a classical approach which is related to the producer versus consumer countries’ vision. To better understand the tensions that may be generated between countries, we add a third variable to this classical approach indicating the importance of uranium supply for each country.
Second, we explain, that, to understand what may happen in terms of supply of uranium, notably as far as politics and geopolitics are concerned, we need to consider the actors involved in uranium supply, i.e. not only countries but also and foremost uranium mining companies. We thus present the mining actors.
Finally, building upon the two first parts, we obtain a revisited outlook for uranium reserves and resources per country, including also uranium overseas holdings. This perspective offers a better comprehension of uranium supply security. It allows for better strategy and planning, including in terms of foreign relations, influence and future feedback impacts on domestic politics. Those should be of concern to all actors involved in the nuclear industry field.
A classical vision of uranium supply
When assessing the security of supply of uranium for a country, as resulting from the classical producers versus consumers model, we look at reserves and resources of uranium per country. Those are estimated according to the quantity of uranium in the mines that can be recovered in line with the price of uranium, and to the precision and certainty of knowledge one has about the deposit of uranium (e.g. Nuclear Energy Agency (NEA)/International Atomic Energy Agency (IAEA), Uranium 2022: Resources, Production and Demand (Red Book), OECD Publishing, Paris, 2023; for the producer versus consumer model, see, for example, pp. 99-136).
Figure 2.4. “Uranium production and reactor-related requirements for major producing and consuming countries (data as of 1 January 2021)”, Nuclear Energy Agency (NEA)/International Atomic Energy Agency (IAEA), Uranium 2022: Resources, Production and Demand (Red Book), OECD Publishing, Paris, 2023, p. 117.
Using the official international reference for uranium production, Uranium 2022: Resources, Production and Demand (Red Book) by the NEA/IAEA, this gives us, for the resources which are most certain, those called “Reasonably Assured Resources (RAR)” (see glossary below), at the highest price range, the following chart:
Calculated for 2023 from the figures given in the NEA/IAEA, Uranium 2022 (2023) corresponding to inventories for 1st January 2021.
Then, those resources are then compared with countries’ yearly uranium requirements. As a result, some states are perceived as current and future importers, while others are exporters.
For example, Australia does not use nuclear energy, indeed the latter is legally prohibited despite regular debates on the issue (Commonwealth Scientific and Industrial Research Organisation – CSIRO, “The question of nuclear in Australia’s energy sector“, 20 Dec 2023). Yet, the country produces uranium and has huge reserves, the first in the world. Hence, Australia was the second largest world exporter in 2020 and the fourth in 2021 and 2022 (2022 Red Book, pp. 77; WNA, “World Uranium Mining Production“, 16 May 2024). It is also very likely it will be a future very large net exporter, possibly the largest.
At the other end of the spectrum, France does not have any resources of uranium left on its territory. Yet, it is part of the major nuclear energy producers, indeed it is currently the 2nd in the world. In the future, according to our base case scenario it should move to the 3rd then 4th place (see Helene Lavoix, The Future of Uranium Demand – China’s Surge, The Red Team Analysis Society, 22 April 2024). As a result, currently, producing nuclear energy in France would require an estimated 8232 tU per year (WNA, Nuclear Fuel Report 2023, September 2023). In the classical view, France is thus a current and future net consumer of uranium. The only way forward to improve the situation would be technical, for example with fuel recycling.
If we are concerned about security, then we can improve this approach by looking at the importance of nuclear energy for a country. The most interesting indicator here is the share of nuclear energy in the electricity production of a country. Indeed, for example, if nuclear energy represents 1% of the electricity production of a country, then not much is at stake here. The higher the nuclear share of electricity generation, the higher the stake for all issues related to nuclear energy. For 2022, the shares of nuclear energy in electricity generation for the world are shown in the chart below (source IAEA-PRIS – 28/04/2024).
Thus, in 2022, France, had the world highest nuclear share of electricity generation, i.e. 62,6% (IAEA-PRIS – 28/04/2024), while being, according to classical analysis, a net current and future consumer of uranium. Uranium and more largely the whole nuclear industry will thus be highly sensitive issues in terms of security.
If we apply this approach to the world countries, we have the following two charts. The first uses a linear scale for the axes, and the second a logarithmic scale:
Nuclear Energy in the International Order: Uranium, Stakes and Security (2022/23) – Sources: NEA/IAEA Red Book 2022, WNA, IAEA-PRIS – 28/04/2024 – Own calculation for 2023.
Nuclear Energy in the International Order: Uranium, Stakes and Security (2022/23) – Sources: NEA/IAEA Red Book 2022, WNA, IAEA-PRIS – 28/04/2024 – Logarithmic scale – Own calculation for 2023.
The first chart highlights varying situations. The U.S. is the largest consumer with little reserves but a relatively high stake in nuclear energy, compared with China, which is in a similar situation but with a lower current stake in nuclear energy. This means that China’s position is stronger. Russia, the EU without France, France and Korea constitute a second group, with Russia and the EU without France far better positioned in terms of reserves. Canada has a balanced and secure position, and Australia is possibly unconcerned despite its huge reserves.
Interestingly, three groups of countries clearly appear when we use a logarithmic scale. First, we have consumer countries for which nuclear energy represents a high or relatively high stake at the bottom of the chart. Then we have supplier countries for which nuclear energy is not a stake – of course not considering the importance of uranium in terms of trade – at the top left corner of the chart. Finally, we have countries for which nuclear energy is a stake but with a relatively secure position in the top right quarter of the chart.
We note that the EU without France appears to have a better and more secure position than France, and is on a par with Russia. If Russia’s reserves are higher, nuclear energy is also more important for Russia.
However interesting, it would be misleading to stop here. Indeed, such a classical approach does not account for the way uranium is supplied. It does not consider the actors involved in mining.(1)
The unique world of those who mine uranium
Uranium is supplied to the world through mining and milling, which is done by companies. A few major mining companies dominate the world, alongside large entities part of very large nuclear groups and, finally, smaller mining companies.
These companies are either state-owned or private. Most often they will work through the creation of joint-ventures with other companies, one of them bringing in the mines, which belongs to the territory of its state, the other its know-how and technology in terms of exploration, mining, milling and sometimes also other steps of the fuel cycle (see H Lavoix, “Uranium and the Renewal of Nuclear Energy“, The Red Team Analysis Society, 9 April 2024).
As a result, mining companies own mines or part of them for the duration of the corresponding mining permit, and thus the uranium reserves and resources corresponding to these mines.
If we look at the consequences for a country, we can consider that the supply of uranium, including reserves, may be territorial or extraterritorial. It is territorial if the mines are located on its own territory. This is the classical and obvious understanding. However, it can also be extraterritorial if a company of the nationality of the country owns mining permits outside the country. The stronger the power of the said country over the company, the stronger the fact that uranium may be considered as a captive extraterritorial resource, by opposition to a resource available to all through market dynamics.
Three types of uranium mining companies
We have three types of mining companies.
First, we have very large ones, with western-style corporate structure.
Then, we have “smaller” mining companies compared with the previous type, but which are part of very large nuclear conglomerates, reminding somehow of the old Kombinat (Комбинат) model. Furthermore, the new Kombinats also include the use of financial incentives and cooperation packages in their operations. This is more or less the format for Russia and China.
On another note, the French company Orano is a state-owned company with many activities related to the whole nuclear fuel cycle, and with privileged links with other state-owned companies such as EDF (electricity provider) and Framatome (design and provision of equipment, services and fuel for nuclear power plants – 80,5% belongs to EDF), to say nothing of the Commissariat à l’énergie atomique et aux énergies alternatives (CEA) (French state-owned organisation for research and innovation notably in energy). Thus it could be perceived as a middle way between or a synthesis of a Western corporation and a Kombinat. The recent purchase of shares of Westinghouse by Cameco (see below) highlights the interest of and in this approach.
Together, the companies belonging to these two categories – the western-style corporate structure and the Kombinat – are the major corporate actors for uranium mining. In the world, we only have seven of those.
Finally, we have much smaller companies, usually labelled as “junior mining companies”, often centred around a mine or project. Junior companies can also be held by bigger corporations, which, potentially could give them the power to develop. They may also become stakes in friendly or hostile takeovers.
It is difficult to rank mining companies as they have different activities and publish different data. Nonetheless, if we take uranium mining revenue for 2023 as main indicator, then the largest uranium mining company is Kazakh Kazatomprom, followed by Canadian Cameco then French Orano.
The revenues related to uranium mining from Russia or from China appear as being far smaller, yet important. However, where figures are available, they are probably not comparable. Then, we mention Uzbekistan although there is no specific data for uranium mining revenues.
If we look at uranium production per company and country in 2022 (see WNA, “World Uranium Mining Production“, 16 May 2024), we have the same ranking for the three largest mining companies, followed by China’s CGN – 4th – and CNNC – 7th, by Russia’s Uranium One – 5th – and ARMZ – 9th, by Uzbek Navoi – 6th, Australian BHP – 8th, and American General Atomics/Quasar – 10th. Quasar’s production represents 15% of Kazatomprom’s.
The largest uranium mining companies, “Western-style”
Kazatomprom (Kazakhstan)
National Atomic Company (NAC) Kazatomprom, created in 1997, is the national company of Kazakhstan, responsible for everything related to the nuclear industry, as well as rare metals. In 2018, a strategy of privatisation of NAC Kazatomprom was launched. In 2024, Kazakhstan’s National Wealth Fund, Samruk-Kazyna holds 75% of Kazatomprom, the remaining shares being traded on the London Stock Exchange and the Astana International Stock Exchange. Kazatomprom covers the whole nuclear fuel cycle through joint ventures with other companies.
In 2022, uranium mining represented 85% of the revenues of the company and in 2023 82% (2023 annual report p.47). Kazatomprom, in 2022, represented 22% of the mining market, with 11.373t U3O8 produced and, in 2023, 20% of the mining market with 11.169t U3O8 produced (Ibid. pp. 7-10). Its 2022 revenues amounted to KZT (Kazakhstani Tenge) 1.001.171 million ( USD 2.248,16 million ; EUR 2.109,35 million) and in 2023 KZT 1.434.635 million (USD 3.233,24 million ; EUR 3.001,77 million)(Ibid.).
Cameco (Canada / Saskatchewan)
Cameco is a privately-owned Canadian company. More exactly, it is a “privately-owned” company from Saskatchewan, with land-holdings and exploration permits located in majority in Northern Saskatchewan for their Canadian part. When Cameco was created in 1988, a special type of shares, “share B” were issued, “assigned $1 of share capital, [which] entitles the shareholder to vote separately as a class in respect of any proposal to locate the head office of Cameco to a place not in the province of Saskatchewan” (p. 147). This shows the very strong link between Cameco and the province of Saskatchewan, even though it is indeed privately-owned.
Furthermore, Crown Investments Corporation is the holding company used by the Government of Saskatchewan to manage its financial and commercial Crown Corporations as well as its minority holdings in private-sector ventures. Crown Investments Corporation holds 0,15% of the capital of Cameco. Meanwhile, Cameco’s President and Chief Executive Officer, Tim S. Gitzel, comes from the university of Saskatchewan (and has also held direction positions with Orano).
Cameco’s activity covers the entire front-end of the nuclear fuel cycle, from exploration, mining and milling to fuel manufacturing through conversion and is part to the development of laser enrichment (not yet commercialised).
Its clients are nuclear utilities in 15 countries. Cameco represents 16% of the world production of uranium (total sales commitments of over 205 million pounds of U3O8) and has 21% of the world primary conversion facilities (total sales commitments to supply over 75 million kilograms of UF6). Furthermore, in November 2023, it completed the acquisition of 49% of Westinghouse. Its 2023 turnover (revenues in Canadian terms) was CAN$ 2.588 million (approx. US$ 1.887 million ; € 1.770 million), the produce of mining and milling representing 84,5% of their expected revenue for 2024 (Cameco 2023 Annual Report).
Orano (France)
Orano is a French state-owned company, created in 2017 out of the restructuration of defunct Areva, the latter resulting from the 2001 merger of Framatome, Cogema and Technicatome, all in turn stemming from French choices in terms of nuclear energy after World War II. Orano is active at all stages of the nuclear fuel cycle – front end, including enrichment, and back-end operations such as reprocessing and recycling as well as mines decommissioning – and in nuclear materials transport and logistics. The French state holds 90% of Orano, alongside Japan Nuclear Fuel Limited and and Mitsubishi Heavy Industries which holds 5% each (2023 Annual report, p.246).
Orano’s 2023 turnover (revenues) was EUR 4.775 million (USD 5.088 million). The mining sector represented 27,62% of the turnover (EUR 1.319 million; USD 1.405,55 million).
Navoi Mining and Metallurgical Company (Uzbekistan)
Navoi Mining and Metallurgical Company is the state company of Uzbekistan handling all mining and metallurgical matters. It focuses especially on gold but expressed a willingness to increasingly develop uranium mining (website).
It was incorporated as a joint stock venture in 2021, as part of an effort to reform the state enterprise.
For the year 2022, its revenue (all activities) were USD 5.095 million.
Rosatom revenues reached USD 27.300 million in 2023 (Tass).
Rosatom holds 100% of the voting shares of Joint-Stock Company Atomic Energy Power Corporation (JSC Atomenergoprom). JSC Atomenergoprom holds shares in 222 companies. It covers the whole cycle of nuclear production from mining up to electricity generation. According to its financial statements, in 2022, it ranked second in terms or uranium production with 14% of the market. In 2022, its total revenue reached RUB 1396,5 bn (“equivalent to” USD 19979,77 million at average exchange rate for 2022 1 USD = 69.8957 RUB), and the mining revenue, including but not limited to uranium, was RUB 24,7 bn (“equivalent to” USD 353,38 million at average exchange rate for 2022), out of which RUB 8,9 bn (“equivalent to” USD 127,33 million) were sold to “external customers (p. 59 and 17).
Its main mining companies are JSC AtomRedMetZoloto (ARMZ), directly held at 84,52%(the remaining shares belonging to Rosatom and TVEL JSC) and Uranium One Group. ARMZ represents mainly the domestic mining “division” and all Russian uranium producers are part of ARMZ (Interfax, “Rosatom plans to start commercial mining of uranium in Tanzania in several years“, 22 Nov 2022). In 2022, the revenues of ARMZ labelled as “the mining division of Rosatom” reached RUB 24,7 bn (“equivalent to” USD 353,38 million at average exchange rate for 2022 1 USD = 69.8957 RUB). Uranium One Group is “responsible for … uranium production outside the Russian Federation and is the world’s fourth-largest uranium producer” (Website). Uranium One Inc, originally Canadian, is an indirect subsidiary through Uranium One Group. In 2019 (latest financial statements available), Uranium One Inc. revenues were USD 394 million. So far, it operates mainly in Kazakhstan.
It is the only company supplying domestic uranium (WNA, “China’s Nuclear Fuel Cycle“, 25 April 2024). It operates the mines in China through its subsidiary China Uranium Corporation Limited (CUC or CNUC, also Sino-U), which is also responsible to develop projects overseas (Ibid., CNNC Int Ltd “Corporate Information“).
CUC notably holds as wholly-owned subsidiary CNNC Overseas Uranium Holding Limited (“CNNC Overseas”), which in turn holds 66,72% of CNNC Int Ltd (Ibid.). The latter owns as an indirect wholly-owned subsidiary formerly Canadian Western Prospector Group Ltd. Western Prospector’s projects (uranium and coal) are located in Mongolia (Ibid.). CNNC Overseas transferred to CNNC Ltd notably the mines of Somina (Azelik Mines) in Niger. CNNC LtD is exchanged on the Hong Kong Stock exchange (Ibid). CNNC Int Ltd also acts as a trader in Uranium including for CNUC.(2) In 2022 CNNC Int Ltd revenues reached HK$ 567, 9 million (USD 72,61 million)
China General Nuclear Power Corporation (CGN) under the direction of State-Owned Assets Supervision and Administration Commission (SASAC) of the State Council of China owns China General Nuclear Power Co (CGNP). The latter is the Chinese platform for nuclear power generation. It owns CGN Mining Co Ltd (CGNM), which acquired CGN Global Uranium Ltd (CGNGU) in 2019 and also holds 100 % of CGNM UK Ltd. CGNGU trades CGN’s uranium resources on the international market. CGNM UK Ltd through a joint venture with Kazatomprom set the Mining Company Ortalyk LLP, founded in 2011, which holds the permits and exploits mines at the Central Mynkuduk and Zhalpak fields in Kazakhstan.
For 2022, the Group China General Nuclear Power Co recorded revenue of approximately RMB 82.822 million (USD 11.431 million). In December 2023, CGN Mining Co Ltd revenue was HKD 2.210 million (USD 282 million, Euro 263 million).
Besides holding and exploiting mines, China also complements its supply through purchases of uranium. For example, “in May 2014 China’s CGN agreed to buy $800 million of uranium through to 2021” to Uzbekistan (WNA, “Uranium in Uzbekistan,” 2 April 2024). Reportedly according to Chinese customs, Uzbekistan was “second only to Kazakhstan as a uranium supplier to the country” (Ibid.). In 2018, Orano was also an important supplier of natural uranium for the CGNP (Orano China website).
We should also mention a company such as Beijing Zhongxing Joy Investment Co., Ltd (ZXXJOY invest), located in Beijing, which is specialised on international mining projects including uranium mining but not solely focused on that mineral. ZXJOY invest is linked to ZTE, a partially state-owned telecommunication company (Management of ZXJOY invest; Raphaël Rossignol, “Uranium Nigerien, Le Coup De Maître De La Russie“, Forbes, March 2024). It notably participates in uranium mining projects in Niger (mine of Arlit – see below, third part) and Zimbabwe.
A new Klondike rush? Other uranium companies and projects and Junior companies
Australian companies are smaller and operate mainly in Australia or in Namibia. We have notably BHP Group Limited, a multinational mining and metals company exploiting, among other, uranium as a by-product of copper in Australia.
Paladin Energy is an Australian company in the process of restarting the Langer Heinrich mine in Namibia. The latter should start production during the first quarter 2024. Paladin Energy had no revenue in 2022 and 2023 (see 2023 financial statements p.71). Australian Bannerman Energy develops the Etango Project in Namibia, and, as a result, does not earn any significant revenue besides interests (2023 financial statements).
We also find two smaller Canadian companies, Global Atomic Corporation – Canada (GAC) and GoviEx, active in Niger. In 2023, GAC had a revenue of CAN $ 0,689 million (USD 0,5 million) and GoviEx had not yet engaged in commercial production, being still focused on exploration and projects’ developments (financial statements for 2022, p.13 ).
Companies seem to be perceived as junior when they are at the stage of uranium exploration and are smaller.
A new outlook for uranium potential supply
As a result, if we want to assess the supply situation for a country, we need not only to look at countries but also at national and foreign companies that hold reserves and resources, according to their joint ventures and mining permits, on a territory.
If we consider uranium resources according to reserves’ and resources’ holders, national or foreign, we obtain a vision of uranium resources per country, as shown in the charts below, which is different compared with the classical outlook we saw previously.
Methodology, sources and discrepancies
The overall uranium reserves and resources for a country will be composed of:
the reserves and resources on the country’s territory held
either by the state or national companies,
or by foreign companies. This share of reserves and resources actually cannot be used by the state except if contracts are terminated in one way or another.
the reserves and resources held by national companies abroad. In that case, the status and links of the national company operating abroad to the state will strengthen the capacity of the state to use the reserves held abroad and thus the security of supply. However, this type of supply is obviously less secure as those held by a state on its own territory, because contracts can be broken, expropriation can take place, etc. Nonetheless, they are reserves and resources available for supply.
For Kazatomprom, Cameco and Orano, as well as for corresponding countries where they operate, we used proven and probable reserves and measured and indicated resources (see glossary), then inferred resources as given in their respective 2023 annual reports.
However, we should note that the treatment of ore reserves and resources vary according to auditing companies. For example, when CRIRSCO (see Glossary) specifies that ore reserves should not be included in resources, a company such as SRK consulting, auditing mines for Kazatomprom, on the contrary highlights that “SRK’s audited Mineral Resource statements are reported inclusive of those Mineral Resources converted to Ore Reserves. The audited Ore Reserve is therefore a subset of the Mineral Resource and should not therefore be considered as additional to this” (SRK Consulting (UK) Limited, 2020 auditing report, p.23). On the contrary, Cameco follows CRIRSCO guidelines and reserves are reported besides resources (2023 Annual report, p. 104). Orano, for its part, only mentions its follows CRIRSCO in terms of reporting, thus logically excluding reserves from resources (2023 Annual report, p.34).
Furthermore the way future uranium prices are assessed strongly influences reserves and resources estimates, to say nothing of anticipating future exchange rates. For example, SRK Consulting for Kazatomprom estimates future yearly prices with precision for its assessment of reserves and resources. For its part, the NEA/IAEA presents resources according to price range.
Hence we have discrepancies stemming from multiple factors when seeking to assess the future of supply per country.
We exemplify this difference in the chart below comparing data provided by Kazatomprom and SRK Consulting (UK) Limited, for reserves and resources at the end of 2020 (pp. 23, 24, 29), and data from the NEA/IAEA 2022 Red Book, which correspond to the same period.
Uranium Reserves and Resources discrepancy according to assessment method The case of Kazakhstan – 2020
The differences between the Kazakh estimates and the international agencies’ assessment are wide and vary from minus 19.800 tU when comparing Kazatomprom 2020 to NEA/IAEA Recoverable resources <USD 80/kgU (less Uranium for the NEA/IAEA), to 239.000 tU, when comparing Kazatomprom 2020 to NEA/IAEA in situ(3) resources <USD 260/kgU (less uranium for Kazatomprom).
In the worst case, the difference corresponds approximately to 13 years of 2024 estimated uranium requirements for the U.S., 18 years for China, and 29 years for France (see, for the yearly estimates, Helene Lavoix, “The Future of Uranium Demand – China’s Surge“, The Red Team Analysis Society, 22 April 2024).
Considering the complexity of the methodologies at work to estimate each type of reserves and resources, and differing types of reporting, it is impossible to reconcile easily et perfectly all statistics.(4)
Glossary
The classification of uranium resources varies according to actors.
For the NEA/IAEA:
“Conventional resources, as well as unconventional resources when sufficient data are available, are further divided according to different confidence levels of occurrence into four categories:
Reasonably assured resources (RAR)
Inferred resources (IR)
Prognosticated resources (PR)
Speculative resources (SR)”
The correspondance between systems varying according to countries is as follows, according to the NEA/IAEA, “Figure A3.1. Approximate correlation of terms used in major resources classification systems”, Uranium 2022: Resources, Production and Demand, OECD 2023.
Identified resources
Undiscovered resources
NEA/IAEA
Reasonably assured
Inferred
Prognosticated
Speculative
Australia
Measured
Indicated
Inferred
Undiscovered
Canada (NRCan)
Measured
Indicated
Inferred
Prognosticated
Speculative
United States (DOE, USGS)
Reasonably assured
Inferred
Undiscovered
Russia, Kazakhstan, Ukraine, Uzbekistan
A+B+C1
C2
C2+P1
P1
P2 / P3
NEA/IAEA, Uranium 2022: Resources, Production and Demand, OECD 2023, pp. 537-538
For companies, such as Kazatomprom, Cameco and Orano, for example, the Committee for Mineral Reserves International Reporting Standards (CRIRSCO) establishes out of worldwide best practices and recommends the reporting international standard for estimates of mineral resources and calculations of mining reserves. Reserves and resources are explained in detail in the International Reporting Template (latest edition 2019):
Reserves: “A Mineral Reserve is the economically mineable part of a Measured and/or Indicated Mineral Resource…. Studies to Pre-Feasibility or Feasibility level, as appropriate, will have been carried out prior to determination of the Mineral Reserves.” (p. 25).
Probable reserves: “A Probable Mineral Reserve is the economically mineable part of an Indicated, and in some circumstances, a Measured Mineral Resource. The confidence in the Modifying Factors applying to a Probable Mineral Reserve is lower than that applying to a Proved Mineral Reserve.” (p. 26).
Proved reserves: “A Proved Mineral Reserve is the economically mineable part of a Measured Mineral Resource. A Proved Mineral Reserve implies a high degree of confidence in the Modifying Factors.” (p. 26).
Resources (not aggregated with reserves): “A Mineral Resource is a concentration or occurrence of solid material of economic interest in or on the Earth’s crust in such form, grade or quality and quantity that there are reasonable prospects for eventual economic extraction. The location, quantity, grade or quality, continuity and other geological characteristics of a Mineral Resource are known, estimated or interpreted from specific geological evidence and knowledge, including sampling. Mineral Resources are subdivided, in order of increasing geological confidence into Inferred, Indicated and Measured categories.” (p. 19).
Measured resources: “quantity, grade or quality, densities, shape, and physical characteristics are estimated with confidence sufficient to allow the application of Modifying Factors to support detailed mine planning and final evaluation of the economic viability of the deposit…” (p. 21).
Indicated resources: “quantity, grade or quality, densities, shape and physical characteristics are estimated with sufficient confidence to allow the application of Modifying Factors in sufficient detail to support mine planning and evaluation of the economic viability of the deposit…” (p. 21).
Inferred resources: “quantity and grade or quality are estimated on the basis of limited geological evidence and sampling…” (p. 20).
As a result, in the chart below, when available, we used corporate data as provided in the latest available comprehensive public report at the time of writing – most of the time yearly statements for 2023. When no other source was available, we used RAR, as given in the 2022 Red Book and by the WNA. We built the Reserves and Measured & Indicated Resources (R&R) per country from the bottom up, starting from mines and companies, as researched for our The World of Uranium: Mines, States, and Companies – Database and Interactive Graph.
A different outlook for uranium supply
In the chart below we look at reserves and resources according to their holders. The chart shows the reserves and measured and indicated resources (R&R) as they were before June and July developments in Niger (see Niger: a New Severe Threat for the Future of France’s Nuclear Energy?).
Revisiting Uranium Reserves and Resources : reserves, RAR and M&I resources
If we look at availability of supply through this prism, then the ranking changes for many countries, compared with the classical approach to reserves and resources.
Australia still comes first. The reserves and resources Australian companies hold abroad more than compensate for those foreigners hold on its territory. However, the very high level of Australian R&R should be considered with caution. Indeed, one single mine (the Olympic Dam mine), belonging to BHP group, holds 1.970.000 tU of R&R according to the company (uranium is produced as a byproduct of copper, the exact R&R are 1280600 in OC Sulphide and 689400 in UG Sulphide, BHP Group annual report 2023). This represents 86% of the known uranium R&R for geographical Australia. Furthermore at least one other large Australian deposit is currently unavailable (see forthcoming report). Hence the rich reserves and resources of Australian uranium could be partly a mirage.
Australia is followed by Canada with its active foreign operations, then by Kazakhstan, and its joint-ventures policy.
Canada, considering the extensive experience of Canadian mining companies on the one hand and of dealing with foreign companies mining in Canada could be in a far better position than Australia for the future. This may progressively change if Australia decided to change its nuclear policy, the debate on the matter being re-opened as elections loom (John Boyd, “Australia Debates Going Nuclear : A politician’s vow to build seven nuclear plants prompts fierce dispute“, 3 July 2024, IEEE Spectrum)
As a result, in geopolitical terms, uranium reserves and resources would be far less equally shared throughout the world than thought. For example, if the uranium-poor U.S. (see below) were thinking about relying on its close allies Australia and Canada for uranium supply, it could find that supply is far less readily available than hoped. It may take time to bring Australia resources to become exploitable. Meanwhile, Canadian influence in the world would also have the potential to be greatly enhanced, with consequences on Northern America.
Kazhakstan ranks third in terms of reserves and resources, while also being one of the main world producer. However, as Meirzhan Yusupov, Chairman of the Board at Kazatomprom, highlighted during an interview with the Financial Times (FT) the logistical problems stemming from the war in Ukraine and the closing of routes through Russia could favour Eastern markets (e.g. “Kazatomprom: Kazakhstan faces uranium supply constraints to West amid Ukrainian conflict“, 11/09/2024, Daryo). As a result, Kazhakstan’s position as a global supplier would be fragilised. Reciprocally, should transportation problems last, the supply of Western Uranium buyers could become complex. Similarly, those overseas reserves located East from Russia could become less readily available.
Without taking into account inferred resources, we then have Russia, with close to 40% of its reserves and resources coming from overseas mining. Unfortunately, the annual reports of the Russian mining divisions and holdings do not provide precise estimates for domestic reserves and resources. The resources used are those provided by international bodies. It seems, nonetheless, that Russia is well endowed with domestic uranium, which strengthens its security.
We may expect, considering the ever heightening international tension, as well as the May 2024 American sanctions banning imports of Russian uranium products, that Russia may increase its operations abroad, would it be only to deny or complicate uranium supply to the U.S. and its allies (“Congress Passes Legislation to Ban Imports of Russian Uranium“, Morgan Lewis, 13 May 2024). Russia could also seek to act on uranium long-term prices, making sure they are at a level that benefits Russia and its allies, while disrupting others’ strategies. The strong statements Russian and Chinese Presidents issued during the mid-May 2024 state visit of Russian President Putin in China, specifically mentioning energy cooperation, that “extends beyond hydrocarbons to encompass the peaceful use of nuclear energy” are another important signal enhancing the probability to see geopolitical tensions impacting and even shaping uranium mining (Website of the President of Russia, “Media statement following Russia-China talks“, 16 May 2024; Bernard Orr, Guy Faulconbridge and Andrew Osborn, “Putin and Xi pledge a new era and condemn the United States“, Reuters, 17 May 2024). Further research and analysis, as well as scenarios are more than warranted.
China then ranks 5, without inferred resources. Considering the planned development of its nuclear energy production over the next decades and the related considerable increase in its yearly uranium requirements, will these resources be enough in terms of supply (see The Future of Uranium Demand – China’s Surge)? China has been active in developing overseas mining as seen, and we can expect it will further strengthen these efforts. What will be the consequences globally? China also has a policy to purchase uranium through long term contracts. Thus, will these purchases, alongside development of overseas mining, be able to continue and increase without denying supplies to other countries? Here again, further research and keeping the issue closely under watch is warranted.
We then have Niger. Mines in Niger have been mainly developed by France, Canada, and China. As a further sign of the importance of strengthening uranium supply, on 13 May 2024, Niger’s government announced the decision to reopen the mine of Azelik held by the joint venture Somina, itself owned at 37.2% by CNUC (China) and at 24.8% by ZXJOY invest (China) and closed since 2014 (e.g. Le Monde, “Au Niger, une entreprise chinoise va reprendre l’extraction d’uranium après dix ans d’interruption“, 14 May 2024). Beforehand, on 10 May 2024, ZXJOY invest had met with Niger Ambassador highlighting “future opportunities for investors between China and Niger” (ZXJOY CEO Met with Niger Ambassador, website). That decision had been prepared in June 2023 through an agreement between CNUC and Niger’s government planning for the reopening of the mine (Ibid.).
Furthermore, as events and developments in 2024 show, Niger is part of powerful political and geopolitical forces, which interact with the politics and geopolitics of uranium supply (see Niger: a New Severe Threat for the Future of France’s Nuclear Energy?). The coup in Niger has upset the previous state of play, as seen in the Nigerien decision to end military cooperation with the U.S. following American reaction to Nigerien desire to sell uranium to Iran (Le Monde, “Au Niger, la question de l’uranium à l’origine de la discorde avec les Etats-Unis, selon le premier ministre“, 14 May 2024). Thus, considering reserves and resources there will imperatively demand to count with those forces and act accordingly.
The EU then ranks seven, thanks to France and Orano’s mining expertise and portfolio overseas.
France, for its part ranks 8th and the EU without France 16th. France’s position, compared with the classical vision of reserves and resources is thus considerably changed, moving from an apparent absence in the world of suppliers to a rather strong place, even though overseas reserves and resources are less secure than those held on one’s territory, as the situation in Niger shows (Niger: a New Severe Threat…). This should lead to a foreign policy and strategy considering the necessity to both secure these key supplies and develop them.
Meanwhile, the untapped European resources should also be kept in mind. Europe should, in the light of the aim to triple nuclear energy production by 2050, starts developing its mines, all the more so considering the long timeline from exploration to production. In any case, ranking 7 worldwide for Europe further legitimates the March 2024 creation of the EU Nuclear Alliance (Declaration of the EU Nuclear Alliance, meeting of March 4th, 2024). Europe here could play a strong card in terms not only of energy security but also of international influence. Thanks to uranium, it could notably find back a leverage with the U.S., which could help the old continent win back its independence.
We then have Namibia with the policy to let foreigners develop Namibian mines. The mines in Namibia are operated mainly by China and Australian companies.
Also noteworthy, the U.S. only ranks 10. Not only its efforts to secure supply abroad are sparse, but part of its own uranium mines are operated by foreigners, mainly Canadian (note that Rosatom’s holdings of American mines were sold to Texas-based Uranium Energy Corp in Nov 2021, “UEC to buy Uranium One’s US uranium assets“, World Nuclear News, 9 Nov 2021).
Considering the U.S. current and future needs, we may wonder if the present apparent absence of interest and efforts overseas is strategically coherent. As highlighted above, hoping for Australian and Canadian supply may not be that secure. Furthermore, the cooperation between Russia and China in the peaceful use of nuclear energy, in the framework of China’s increasing needs in uranium, may strongly impact uranium availability.
To conclude, if we use the revisited perspective on uranium reserves and resources in the light of current uranium requirements and stakes regarding electricity production stemming from nuclear energy, we obtain the charts on the right hand side column below. For the sake of comparison we give the classical approach on the left hand side column.
Different perceptions of Uranium stakes and security in the world
The most staggering changes concern France, and of course, consequently, the EU with France, as well as Japan, thanks to its joint-ventures in Kazakhstan and to Japanese companies share in French Orano. We can see that the security of uranium supply for these three state and quasi-state is much stronger than initially thought. All move into the group of state actors with both important stakes in nuclear energy and a relatively balanced security in terms of supply and requirements.
The revisited approach reveals an improved situation for Russia and Canada, which already benefited from a secure and balance outlook. China’s situation also appears as better than thought.
By contrast, relatively, the U.S, appears as lagging behind the others.
Now, in this article, we have only looked at reserves and resources. Moving from reserves to production should add another layer of complexity to the issue.
Notes
(1) Similar approaches should also be developed for each stage of the fuel cycle to have an exhaustive vision of the field and its security.
(2) Following various circulars and frameworks signed in 2022, the activities of CNNC Group Ltd are defined as follows:
“The Group agreed to
i) act as the prioritised supplier of CNUC Group for its short term demand for natural uranium products and the regional sole supplier of CNUC Group for its medium-to-long-term demand for natural uranium products; and
ii) act as the exclusive authorised distributor for the sale and distribution of uranium products produced by the Rössing uranium mine (being indirectly owned by CNUC as to approximately 68.62%), for on-sale to third party customers in all countries and regions around the world except the PRC.”
(3) According to the NEA/IAEA, “in situ resources are referring to the estimated amount of uranium in the ground” before considering way to recover resources (pp. 10, 17). The NEA/IAEA then applies recovery factor to get the recovered resources (Ibid.). Here, in the case of Kazakhstan, the factor applied is 88,38% and 88,18% to go from in situ resources to recoverable ones.
(4) The more recent German BGR Energiestudie 2023 (Feb 2024) does not either allow reconciling data easily if we take Kazakhstan as example.
The world is poised to make the effort to treble its nuclear energy capacity by 2050. Even though uranium reserves are meant to be abundant and widespread worldwide, the necessity of producing uranium from mines, with long timelines from exploration to extraction and milling, added to the fate of geography and to a volatile national, international and geopolitical context, let us expect that politics and geopolitics could soon become very important factors for uranium supply, thus to see demand met, and, as a result, for nuclear energy production (see Helene Lavoix, “Uranium and the Renewal of Nuclear Energy“, The Red Team Analysis Society, 9 April 2024).
To be able to assess further what could happen in the future, we need to go beyond, or rather beneath, the global level.
In this article, we look at uranium demand at country level, which is greatly determined by operating and planned nuclear plants. Thus, first we establish a base case scenario for the future of nuclear energy capacities, upon which the 2023 decision to treble nuclear energy and related policies will apply. We follow the evolution per country and notably China’s surge, which displaces America’s lead. We then highlight a direct geopolitical consequence of the multiplication of nuclear reactors on territories, as nuclear plants become essential elements in the theatres of war. Finally, we turn to the need for uranium per country.
Present and future nuclear energy capacity across the world
The demand for uranium obviously depends first on the nuclear energy produced by a country, which, in turn, depends on the operating nuclear reactors (Nuclear Energy Agency (NEA)/International Atomic Energy Agency (IAEA), Uranium 2022: Resources, Production and Demand (Red Book), OECD Publishing, Paris, 2023).
If the world intends trebling the capacity to generate nuclear energy then we need to find out, per country, how many nuclear plants exist, how many are already planned and how many more need to be added. The existing and planned nuclear capacity on the one hand, the capabilities that need to be added to meet the trebling objective, on the other, will then determinate scenarios for the coming demand for uranium at country level.
However, as most decisions regarding nuclear capacities, thus related plans, were taken before the December 2023 decision for the renewal of nuclear energy, existing programs and projects will likely change. To consider this possibility, what we assess here, in terms of nuclear capacities, is a base case scenario.
From the American lead to the Chinese supremacy?
Nuclear energy production in 2024
In April 2024, the global nuclear energy production reached 375,57 GWe net (IAEA – PRIS, 14/04/2024). Compared with the assessment made by the NEA/IAEA for the start of 2021 with a net energy generating capacity of 393 GWe, we would thus have a decrease of 4,43% (Uranium 2022, p. 12). Considering the IAEA – PRIS, 14/04/2024 statistics, we would have an increase of 1,23% compared with 2022 from 370,99 GWe and of 2,39% compared with 2021 (year-end) from 366,79 GWe.
The nuclear energy generating capacity per country is shown on the chart below:
Countries’ share of nuclear generating capacity 2024 in the world Source: IAEA PRIS 14/04/2024
The largest producers of nuclear energy are, by order of importance, the United States, followed by France, China, Russia, the Republic of Korea, Canada and Ukraine, as shown on the pie chart. All together they represent 80% of the world production.
From present to future nuclear energy capacities
Considering the time needed to build a nuclear plant, as well as the stringent regulations surrounding the nuclear industry, we have a pretty good idea of tomorrow’s nuclear generating capacity landscape for classical plants, i.e. excluding small modular reactors (SMR) and advanced modular reactors (AMR).
On top of current operating plants, we need to look at reactors in construction (known until 2030), then at those planned (up to 15 years in the future), and finally to those proposed (not yet planned, with an uncertain timeline) (World Nuclear Association, “Plans For New Reactors Worldwide“, April 2024).
Nonetheless, we should add to the base case scenario a variation according to the number of reactors that could be shutdown, or on the contrary, that could be subjected to extended operations. In 2023, the World Nuclear Association (WNA) estimated that “Upwards of 140 reactors could be subject to extended operation in the period to 2040” (Global Scenarios for Demand and Supply Availability 2023-2040, 21st edition, Sept 2023). Otherwise, it appraised in its 2023 reference scenario that 66 reactors would close by 2040 (WNA, Notes in “World Nuclear Power Reactors & Uranium Requirements“, April 2024).
By 2030 China overtakes France in nuclear energy production
If we add to the current capacity the reactors in construction we obtain the nuclear capacity for the year 2030
For the base case scenario, if we look at the reactors under construction, as shown on the chart below, we obtain an idea of the maximum (i.e. assuming there is no reactor shutdown) nuclear capacity countries should reach by 2030 (WNA, “Plans For New Reactors Worldwide“, April 2024, and IAEA PRIS 14/04/2024 “Under Construction”).
Estimated max nuclear generating capacity by 2030 per country – Source: IAEA PRIS 14/04/2024 and WNA
Estimated countries’ share of max. nuclear generating capacity by 2030 in the world – Source: IAEA PRIS 14/04/2024 and WNA
In 2030, in share of the world production, if the United States still lead, China overtakes France. Then follow Russia, the Republic of Korea, Ukraine, Japan, Canada and India, which enters the group of the largest nuclear energy producers. All together, these nine countries represent 80 % of the world nuclear energy production.
By 2039 China leads the world in nuclear energy production
We can then add the “classical” nuclear plants that are planned, i.e. according to the WNA’s taxonomy, those plants for which “approvals, funding or commitment [is] in place, mostly expected to be in operation within the next 15 years” (WNA, “Plans For New Reactors Worldwide“, April 2024).
Thus by 2039, we can estimate to have a maximum nuclear generating capacity per country (without SMR and AMR) as shown on the chart below.
Estimated Max nuclear generating capacity by 2039 per country – Source: IAEA PRIS 14/04/2024 and WNA
Estimated countries’ share of max. nuclear generating capacity by 2039 in the world – Source: IAEA PRIS 14/04/2024 and WNA
In 2039, in share of the world production, China now leads the world followed by the United States, France, Russia, the Republic of Korea, India, Ukraine, Japan, and Canada. All together, these nine countries represent 81 % of the world nuclear energy production.
From 2040 onwards China’s nuclear energy production dwarves other countries
Finally, we can add to the nuclear energy capacity the proposed nuclear plants, which correspond, according to the WNA to “Specific programme or site proposals” but for which the timing is very uncertain (Ibid.). We can assume they will start operating in more than 15 years, thus earliest in 2040.
We see here efforts made by most country, led especially by China, with 186,4 GWe proposed, followed in a lesser way by Russia with 37,7 GWe, and India with 32 GWe, as seen in the chart below. The nuclear capacity China proposes to build represents half of the 2024 nuclear capacity for the whole world.
If no further efforts at planning for new reactors and at proposing programmes are made, within a quarter of a century, the United States will have completely lost its dominating place and will be far behind China. France, similarly seems to be plagued by an inability to plan and propose ahead, falling from the second to the fourth place in terms of nuclear energy capacity, and from representing 16% of the world nuclear energy capacity to 8%.
Estimated Max nuclear generating capacity after 2040 per country – Source: IAEA PRIS 14/04/2024 and WNA
Estimated countries’ share of max. nuclear generating capacity after 2040 in the world – Source: IAEA PRIS 14/04/2024 and WNA
After 2040, always looking at maximum capacity for the base case scenario, and without taking into consideration SMRs and AMRs, in share of the world production, China now leads by far followed by the United States, Russia, France, India, the Republic of Korea, Japan and Ukraine. Canada is not part anymore of the main nuclear energy producers. All together, these eight countries represent 80 % of the world nuclear energy production.
Yet, more must be done
Nonetheless, if we do not take into account SMRs and AMRs, despite China’s sustained efforts to develop its nuclear energy generating capacity, which implies the Middle Kingdom potentially represents one quarter of the world after 2040, so far, globally, we are short of the objectives set to treble nuclear capacity.
We pay the absence of long term planning we saw in the previous article and the consequent deficiency in terms of reactors under construction.
Apart for China, this lack of anticipation has not yet been corrected and is still operating at the stage of planification of reactors. We may assume that the new pro-nuclear global policies will amend this approach. However, considering the long timeline necessary between the decision to build nuclear generators and their commercial connection to the grid – approximately 15 years – firm and committed decisions will have to be taken by the end of 2024, latest 2025, if we want to see the objectives to treble world capacity by 2050 met.
Estimated maximum nuclear generating capacity and trebling objectives by 2050 – (without SMR and AMR)
A widespread use of SMRs and AMRs could help close the gap. It could also help papering over difficulties at anticipation, rather than addressing the problem. However, the SMR and AMR approaches are still novel, with more than 80 different designs for SMRs and we have little real life experience for their use, advantages and drawbacks (IAEA, Joanne Liou, “What are Small Modular Reactors (SMRs)?“, 13 September 2023; Charles Cuvelliez, “Nucléaire : pourquoi tant d’attirance pour les SMR ?“, La Tribune, 28 May 2023) . Detailed scenarios must be done before their deployment. Obviously, human beings cannot escape the imperatives of anticipation and planning, especially in terms of governance and in the nuclear field.
What about security and warfare?
A direct geopolitical consequence of the trebling of nuclear capacities lies in the multiplication of nuclear generators on a territory. What will that imply in terms of future war strategy and tactics?
Indeed, any civilian nuclear installation may become weaponised by belligerants, from hindering energy supply to blackmailing opponents (Przybylak, “Nuclear power plants in war zones…), through forbidding carpet-bombing, or counter-attacking and sacrificing population in exchange for casualties inflicted to an occupation force, etc.
Imagine how the maps below could look like with the increase in nuclear capacity already planned we saw, to which would be added what is necessary to fill the gap to meet the trebling objectives.
We should also take into account that the very size and technology of SMR should allow to locate them underground or underwater (World Nuclear Association, “Small Nuclear Power Reactors“, February 2024). In that case, the way to attack or protect those underground or underwater facilities, alongside related types of potential damages, will have to be considered. For example, although the WNA highlights that underwater facilities are safer from “man-made … hazards” (ibid.), divers teams or submersibles may also conduct attacking operations. The sabotage of the Nord Stream pipeline should serve as a lesson (e.g. UN Briefings, SC/15351, “Briefers Urge Security Council to Independently Investigate 2022 Nord Stream Pipeline Incident…“, 11 July 2023). The deleterious and wide-ranging impacts of such attacks would also need to be taken into account.
Design and deployment criteria for SMRs, with their various security stakes, will certainly be included in state’s defensive and attacking doctrines.
In terms of both defense and attack, according to the aims of the attacker, the evolution towards more nuclear plants will demand careful planning. Scenarios, using red teaming notably – i.e. understanding the enemy’s ideologies and beliefs, aims, resources, strategy etc. – will imperatively need to be crafted, to ensure security.
Estimated uranium requirements per country
Now we have a base case scenario for the future nuclear energy capacities per country, what does that imply in terms of demand for uranium?
A country’s demand for reactor-related uranium for a year is, as seen, first determined by the number of operating nuclear plants for that country. It is called “uranium requirements” and is measured in tonnes of uranium per year: tU/y (NEA/IAEA, Red Book 2022, p. 111).
However, uranium requirements are also sensitive mainly to four factors depending on the type of generator and the way it is operated: fuel cycle length or lifetime fuel cycle, uranium enrichment level and strategies of optimisation (level of tails assays chosen in the enrichment phase), discharge burn-up and capacity (or load) factors (NEA/IAEA, Red Book 2022, pp. 111-112).
As a result, the statistics given for uranium requirements are about purchase or acquisition of uranium and not consumption, which is adjusted by operators according to needs and context (Ibid.).
Here again, we shall create a base case scenario. This will allow for further detailed scenarios considering these factors and operators adjustments. We rely first on the assessments and hypotheses made by the NEA/IAEA in the Red Book 2022, i.e. “160 tU/GWe/yr, under the new assumption of a tails assay of 0.25% over the lifetime of the reactor”, knowing that, before the Fukushima accident, the Red Book used 175 tU/GWe/yr, with a tails assay of 0.30% (pp. 111-112, table p.100). Then we use the WNA latest data (published April 2024).
For the base case scenario, we use the NEA/IAEA and WNA assumptions for future nuclear reactors. Variations considering technological evolution should then be added to the base case scenario.
Glossary
Load Factor: “also called Capacity Factor, for a given period, is the ratio of the energy which the power reactor unit has produced over that period divided by the energy it would have produced at its reference power capacity over that period” (IAEA/Power Reactor Information System’s Glossary).
Lifecycle of nuclear fuel: It depends on the type of reactor. “In a pressurized water reactor, it lasts about three to seven years, depending on the fuel and its location in the reactor core”. See, for example, IAEA “Lifecycle of Nuclear Fuel” (pdf).
The discharge burn-up of nuclear fuel is usually defined as the thermal energy output during the lifetime of the fuel divided by the initial mass of heavy metal (denoted HMi). (See p. 14, NEA, “Very High Burn-ups in Light Water Reactors“, 2019).
First we should note that each country displays different “efficiency” in terms of GWe produced per tU, which also changes according to years and source of data, as shown on the chart below.
The world average for this “efficiency” seems however to remain almost constant (0,0064 for 2022 WNA data; 0,0063 for 2020/21 Red Book 2022 data). For want of another more reliable way, we use the latest “efficiency” (2022 WNA) also for the future. This “efficiency” will vary, notably considering technological evolution and types of reactors and should also lead to construct further scenarios. For those countries without nuclear energy in 2022, we use the world average efficiency, i.e. 0,0064.*
The results obtained for the base case scenario are rough estimates of future uranium requirements per country. They are indications of future trends, which will then change according to the various efforts made by each country to fill the gap to the trebling objective by 2050.
As we did for the estimated nuclear energy capacity, the next charts show the estimated yearly uranium requirements per country up until 2030, until 2039 and then after 2040 with uncertain timeline.
Estimated yearly uranium requirements up until 2030, up until 2039 and then after 2040, per country with total world.
By 2030, China will have been catching up with the U.S. as largest purchaser of uranium. Global demand will have increased but without fundamentally altering the ranking of the largest purchasers.
However, by 2039, China will have overtaken the U.S. by far, to say nothing of other countries. It will absorb 31% of the total world uranium requirements.
Considering the long timeframes necessary to bring about new production of uranium, as seen in the previous article, it is necessary that suppliers as well as other “consumer” countries start considering the large increase of China’s requirements and include it in their strategies.
After 2040, China’s uranium requirements will dwarf other countries’ including those of the U.S.. They could represent 3,7 times those of America. China could absorb 44% of the world uranium requirements. Furthermore, the order of uranium purchasers changes. China and the U.S. are followed by Russia then India. France is only at the 5th position, when it was third, just after the U.S., previously.
Whatever its ranking in terms of uranium requirements, for a country, being able to acquire uranium will be fundamental. Indeed not only the costs to build nuclear plants are heavy, but also such investments mean dependency on electricity and nuclear-generated power grows. Thus, the strong increase in China’s requirements, if not planned also with others in mind, could give rise to bitter competition for uranium.
All actors will need to take those trends into account.
Considering this context, what is the outlook for the supply of uranium? This is what we shall see with the next article.
Notes
*For the 2020 and 2021 data of the NEA/IAEA “Red Book”, note that some fleet of nuclear reactors use mixed oxide (MOX) fuel while others don’t. MOX fuel is constituted by plutonium, from reprocessed nuclear fuel or from weapons-grade plutonium, mixed with either natural uranium, reprocessed uranium or depleted uranium. At the start of 2021, the countries using MOX fuel are France (23 reactors), India (one reactor), and the Netherlands (one reactor) (NEA/IAEA, Red Book 2022, pp.123-124).
As MOX fuel is not counted as uranium requirement in the NEA/IAEA statistics, we may assume that the high efficiency of energy production per tU shown by France compared with other countries in the Red Book 2022 stems from the use of MOX fuel (Ibid & p.100).
The statistics of the WNA (April 2024) specify that uranium requirements are for 2024 (title of the column), but give as source: “World Nuclear Association, The Nuclear Fuel Report (published September 2023, reference scenario forecast) – for uranium requirements”, which would imply that the uranium requirements given are for 2022.
The December 2023 “Declaration to Triple Nuclear Energy by 2050”, signed by 22 states, officially cemented the beginning of the renewal of nuclear energy (see Helene Lavoix, “The Return of Nuclear Energy“, The Red Team Analysis Society, 26 March 2024). Then, on 21 March 2024, 33 governments and international agencies reasserted their commitment with the first Nuclear Energy Summit. The nuclear industry endorsed the two declarations and their objectives.
However, as we started outlining previously with a case study focused on the Franco-Mongolian deal (Ibid.), this new era will also come with new geopolitical challenges and tensions, as states seek to reduce the potential for insecurity linked to nuclear energy.
This article continues exploring what the renewal of nuclear energy entails for the future. First, we highlight the need to look at the whole nuclear fuel cycle, while stressing that anticipation and long-term planning are key, if we want to succeed in trebling nuclear energy by 2050. Second, starting with the beginning of the cycle, mining and milling uranium, we focus on the various types of uranium reserves and evaluate their availability considering objectives.
Finally, we move from reserves to uranium production and highlight a growing risk of undersupply, considering potential future demand, that will need to be overcome. This coming quest for uranium security will be intertwined with political and geopolitical issues, while itself becoming a geopolitical stake, in an escalating feedback loop.
The nuclear fuel cycle and long-term planning
A very large part of the world is thus committed to try trebling nuclear power generation by 2050, i.e. in 26 years.
This implies meeting many challenges, which go beyond the fundamental, but not sufficient, “cost, performance, safety and waste management” efforts highlighted by international agencies (International Energy Agency – IEA, Nuclear Power and Secure Energy Transitions, 2022; Nuclear Energy Agency – NEA, Meeting Climate Change Targets: The Role of Nuclear Energy, OECD Publishing, 2022, Paris, pp.39-46).
Actually, if we want to understand what it means to treble the nuclear energy capacity, then we need to look at what is called the nuclear energy or nuclear fuel cycle (see diagram below). Trebling our capacity to produce nuclear energy does not only mean “simply” trebling the energy generated by nuclear plants. It will also demand that the whole nuclear cycle allows for this major increase.
The Nuclear Fuel Cycle – diagram from “Nuclear Fuel Cycle Overview” – World Nuclear Association – April 2021 (Between each industrial step the chemical symbol for the type of uranium obtained is given – e.g. U3O8 = triuranium octoxide. U3O8 is a compound of uranium, solid that is transported from the mill to the conversion unit under the form of “yellowcake”).
Each step will bring its own challenges (for a detailed explanation of the industrial process, read “Nuclear Fuel Cycle – Overview” – World Nuclear Association – April 2021).
The issue is even more complex as changes taking place at one step of the process will reverberate on other steps. For example, projects that include recycling nuclear fuel as well as operate in a fully closed fuel cycle, for instance fast neutron reactors, could alter the fuel cycle by lowering the need for uranium (e.g. Lucy Ashton, “When nuclear waste is an asset, not a burden“, IAEA, September 2023; Orano, “Traitement & recyclage des combustibles usés : ce qu’il faut retenir“).
Furthermore, changes at each step, including building a new nuclear plant -apart for Small Modular Reactors (SMR) – for example, belong to the long term.
For instance, China’s Shidaowan high temperature gas-cooled reactor (HTGR) nuclear power plant, the world first generation-IV nuclear power plant, went officially into commercial operations in December 2023. Its construction started in 2012 and it began generating power in December 2021 (Xinhua, “World’s 1st 4th-generation nuclear power plant goes into commercial operation in China“, Global Times, 7 December 2023). It therefore took 12 years from the start of construction to final launch. The timeline is even longer if we consider research and development, as, for example, for Generation-IV reactors “several innovative concepts … have been under development for decades” (NEA, Meeting Climate Change Targets, p.28).
Hence, possible futures for each step will need to be foreseen, while impacts on every other step for each scenario will have to be evaluated.
Anticipation and long-term planning are of the essence for the nuclear industry.
For instance, over the last decades, a focus on a temporarily depressed nuclear market, the absence of consideration of geopolitical security stakes, short-termism and financialization, added to adverse public opinion and lack of political courage among others, all favouring an inability to anticipate and thus plan ahead, have increasingly plagued many countries and led them to fall in terms of nuclear energy behind other states with a less myopic worldview (e.g. in the case of France, Assemblée nationale, Rapport de la commission d’enquête visant à établir les raisons de la perte de souveraineté et d’indépendance énergétique de la France, 30 mars 2023, pp. 20-26, 268-309).
For example, the IEA highlights that “investment in nuclear power in advanced economies has stalled over the last two decades”(Nuclear Power and Secure Energy Transitions, June 2022, p.16). Now these countries will have to catch up.
Globally, as a result, to reach the new objectives for the nuclear energy capacity, the world must now overcome a “global installed nuclear capacity gap (2020-2050)” (NEA, Meeting Climate Change Targets…, p.39).
“Global installed nuclear capacity gap (2020-2050)” figure 23 from NEA, Meeting Climate Change Targets: The Role of Nuclear Energy, OECD Publishing, 2022, Paris, p.39
This example highlights how dangerous the absence of long-term vision is and how difficult to overcome it is in the case of nuclear energy.
Furthermore, however huge the task the NEA highlights, this gap is “only” about the “power generation” phase of the cycle (Ibid. pp. 38-39).
Power generation is indeed critical as it is the driver for the whole chain of processes.
We thus need to pay heed to the recommandations the IEA and the NEA crafted to allow the nuclear to play its full role in achieving net zero by 2050 (IEA) by trebling nuclear power production by 2050 (NEA):
Acting now (NEA)
Understanding and reducing costs (NEA) and Make electricity markets value dispatchable low emissions capacity (IEA)
Improving deployment timelines (NEA)
Accelerate the development and deployment of small modular reactors (IEA)
Extend plant lifetimes (IEA)
Financing and investing, with “the right policy frameworks” (NEA) and Create financing frameworks to support new reactors (IEA)
Make long-term support [by governments] contingent on the industry delivering safe projects on time and on budget (IEA)
Building public confidence (NEA)
Promote efficient and effective safety regulation (IEA)
Implement solutions for nuclear waste disposal (notably involving citizens) (IEA)
Breaking the silence on nuclear energy, ensuring full representation in policy discussions about clean energy and climate change (NEA)
NEA, Meeting Climate Change Targets…, pp.39-46 and IEA, Nuclear Power and Secure Energy Transitions, p. 12.
Yet, we must also consider the remaining part of the nuclear fuel cycle to avoid disappointments and unintended consequences, bearing in mind the importance of feedback loops between the different steps of the nuclear fuel cycle, of timelines and obviously of anticipation, without forgetting the political and geopolitical context and stakes.
We focus here on the first part of the cycle, uranium mining and milling, seen from a geopolitical and international security perspective.
Uranium reserves
If nuclear energy must triple by 2050, then the supply of fuel, i.e. uranium, needed for the plants must also increase. The first question is thus to know if there is enough uranium available to meet this objective. We must thus look at uranium reserves.
According to the international official reference estimates, the “Red Book”, a joint publication of the NEA and International Atomic Energy Agency (IAEA), there is sufficient uranium to meet current and long term needs, including those implied by new developments:
“Identified recoverable resources, including reasonably assured resources and inferred resources (at a cost <USD 260/kgU, equivalent to USD 100/lb U3O8) are sufficient for more than 130 years, considering the uranium requirements 0f the year 2020.”
NEA/IAEA Uranium 2022: Resources, Production and Demand (Red Book), pp. 14-15.
The previous edition of the “Red Book” (published every two years) estimated that uranium recoverable resources (at a cost <USD 260/kgU, equivalent to USD 100/lb U3O8) were sufficient for over 135 years for the uranium requirements of 2019 (NEA/IAEA Uranium 2020: Resources, Production and Demand ). The decrease from 2020 to 2022 stemmed from “mine depletions, …downgrading of resources, …reassessment of recoverability factors” (NEA/IAEA Uranium 2022, pp.19-20).
Let us look more in detail at the estimates of uranium supply for future needs, considering current political and industrial willingness to triple nuclear capacity by 2050.
First we shall assess estimates of uranium requirements considering objectives, then we shall look at the reserves of uranium given these estimated future uranium requirements.
Having been published before the plans to triple nuclear energy by 2050, the “Red Book” estimated in 2022 that a high demand case would correspond to a nuclear net generating capacity of 677 GWe in 2040, i.e. an increase of around 70% compared with 2020 capacity (NEA/IAEA, Uranium 2022, p. 12). The “Red Book” estimated in that case that “world annual reactor-related uranium requirements (excluding the use of mixed oxide fuels, which is marginal)” are “projected to rise to 108.200 tU/y by 2040” (Ibid. – note that here, ideally, different scenarios should be made according to variations for each step of the cycle, for example the different types of generators that will be built. In the framework of this article we shall rely on the NEA/IAEA estimates).
The NEA, for its part estimated that “the average IPCC 1.5°C scenario requires nuclear energy to reach 1.160 GWe (gigawatts electrical) by 2050 (NEA, Meeting Climate Change Targets, p. 33). We know that the 2021 net energy generating capacity was of 393 GWe (Gigawatt electrical) requiring about 60.100 tU/y (tonnes of uranium per year) (NEA/IAEA, Uranium 2022, p. 12). Thus the new target corresponds to almost tripling the current world nuclear capacity by 2050. It was the objective endorsed by twenty two countries and the nuclear industry at the COP 28 in December 2023 (see Lavoix, “The Return of Nuclear Energy“). It is thus this objective and not the “high demand case” scenario of the 2022 “Red Book” that we must consider and for which we need to evaluate uranium requirements.
If we use the ratio of uranium related requirements per GWe of the 2022 “Red Book”, and the yearly progression in capacity of the IEA Net Zero by 2050 revised for the World Outlook 2023, and apply them to the officially endorsed objectives of the NEA, we obtain the following table for the yearly needs in tU/y.
Now, in the next table, we estimate the reserves of uranium available, starting from the figures given in the 2022 Red Book.
Available resources vary according to price – the higher the price, the more reserve available. Thus, to be able to estimate the reserves of uranium available for the increase in nuclear capacities, we need first to assess the future price of uranium.
The price range used is as follows:
/ KgU
<US$ 40,00
<US$ 80,00
<US$ 130,00
<US$ 260,00
/lbs U3O8
<US$ 15,00
<US$ 30,00
<US$ 50,00
<US$ 100,00
Range of price of Uranium for reserves from NEA/IEAE Uranium 2022: Resources, Production and Demand
In January 2024, for the first time since April-July 2007, spot uranium prices rose above USD 100,00/lbs U3O8. Long term price traded at USD 72. On 29 February, a pound of U3O8 traded USD 95 on the spot market and long term contract traded USD 75 (Cameco using month-end prices by UxC and TradeTech ). On 31 March 2024, a pound of U3O8 traded USD 87,75 , the the long-term contract traded USD 77,5.
Uranium Price from January 2020 to March 2024 and price range for uranium reserves estimates
We see a slight decrease over the last three months for spot prices, bt these prices concern only “15% to 25% of all annual uranium transactions” (NEA/IAEA, Uranium 2022, p.128). On the other hand, long-term contracts are steadily rising. Furthermore, for both spot and long term contracts, prices have risen over the last five years. Finally and most importantly we must take into account the officially planned development of nuclear capabilities. Thus, in market conditions, it is very likely that long-term prices will rise above USD 100/lbs U3O8 as the trebling objective starts truly being implemented. We make here the hypothesis that this will be true from 2030 onwards.
If ever the imperatives of energy production were high enough in terms of national security, un the future, uranium could become a nationalised resource. Then market price would become irrelevant. In that case, the reserves at the highest cost would most likely represent a reality in terms of quantity.
As a result, we consider here reserves available at the highest cost, i.e. USD 260/kgU, equivalent to USD 100/lbs U3O8.
We take into account the various types of resources as categorised in the Red Book: “Identified recoverable resources, including reasonably assured resources” [corresponding approximately to decisions to mine] “and inferred resources” [corresponding to decisions to carry out in-depth studies], and finally “undiscovered resources” [expected to exist based on geological knowledge] (NEA/IAEA, Uranium 2022, p. 17) . The various types of reserves and the stages of uranium exploration, mining and process are portrayed on the timeline below.
We obtain the following table with years of remaining resources. For example, the production used to calculate the number of years of uranium left for the year 2035 is the production of 2035.
IRR = Identified recoverable resources (RAR + inferred resources) – RAR = reasonably assured resources – UR = Undiscovered resources -Source: 2022 Red Book. Estimates of number of years: own estimates – Reserves are those available at a cost <USD 260/kgU, equivalent to USD 100/lbs U3O8 – For the years 2022 to 2029, the reserves to consider could be those available at a cost <USD 130/kgU, equivalent to USD 50/lbs U3O8. The figures for this decade would be lower but nonetheless globally sufficient (IRR = 6.029.145 in 2022; RAR 3.814.500) – For the estimated identified recoverable resources, no new discovery was added, the estimated reserves given in the Red Book 2022 were diminished by the estimated uranium required for the period. The years of “reserve” correspond to an estimated uranium requirement for the year of the column.
According to the table above, assuming the intermediary objectives until 2050 are reached, then, indeed, up until 2050 there is enough reserve of uranium of the “reasonably assured resources” type. However, in 2050, there will only be 5 years left of this type, 17 years of the type “inferred resources” and 31 years of the type “undiscovered resources”.
Meanwhile, the map of identified recoverable conventional uranium resources – RAR + Inferred R (for a lower price, i.e. <US$ 50,00/lbs U3O8 or <US$ 130,00/kgU), as drawn by the NEA/IAEA is as follows:
Source: Figure 1-1, NEA/IAEA, Uranium 2022, p. 18
Even though the NEA/IAEA highlights the “widespread” distribution of uranium resources, the map let us anticipate uranium will be increasingly part of future geopolitical stakes.
For now, considering estimated available uranium resources, the question is not so much to know if there are enough reserves globally in the world, but if current and planned uranium production is sufficient to meet the rise of nuclear capacities or, alternatively if the production can increase quickly enough to meet that increase.
Increasing uranium production to meet objectives
What is the status of potential uranium production?
The NEA/IAEA estimates production capability, using a mix of countries’ projections of production capability from 2025 through 2040 with their own evaluations when a country did not communicate information (for the whole paragraph, NEA/IAEA, Uranium 2022, pp.89-91). They use two metrics. First, we have the most certain uranium production projection, i.e. those resulting from “existing and committed production centres”, labelled A-II. Then, we have larger but less certain production projections, i.e. those stemming from “existing, committed, planned and prospective production centres”, labelled B-II. B-II thus includes A-II. The results are reproduced in the first line of the next table.
We then compare these estimates to the uranium requirement to reach the trebling goal we calculated earlier and estimate as a result if the world produces or not enough uranium.
Table of the total projected production of uranium for the years 2025-2040 (source: Uranium 2022: Resources, Production and Demand, pp.89-91), Estimated reactor-related uranium requirement (own estimates) and difference between projected production and what would be necessary.
The estimates are made for a price inferior to USD 130/KgU (i.e. less than USD 50/lbs U3O8), knowing that, in 2024, we are above those prices, as seen. Indeed, as for reserves, the lower the price the more likely mines or part of them will be closed, hence a lowered production. Reversely, the higher the price, the more likely a mine will produce at full capacity. Considering the rising price for uranium, which is likely to be sustained as we treble nuclear capacities, it is possible that the potential for uranium production to 2040 be higher. The supplementary possible production is however impossible to evaluate without further detailed information per mine. We can expect that the next edition of the “Red Book” will include these projections.
For now, considering the large undersupply of uranium for each milestone year of the scenario – up for 2040 almost 1,5 times the 2022 production, assuming each year we manage to catch up with the previous year’s deficit, it is obvious that a major effort must be made in terms of development of mines and processing capabilities.
We thus meet a global problem: to increase in a timely manner uranium production.
The fatality of geography for uranium production
Then, in 2020 and 2021, uranium was produced only in 17 countries “with total global production amounting to 47 342 tU in 2020 and 47 472 tU in 2021” (NEA/IAEA, Uranium 2022, p. 116). In 2022, the World Nuclear Association estimated that the world production reached 49.355 tU (“World Uranium Mining Production“, Updated August 2023).
Only six countries (Kazakhstan, Namibia, Canada, Australia, Uzbekistan and Russia) accounted for 88% of the production and 10 countries (the former plus Niger, China, India and Ukraine) for 99% (NEA/IAEA, Uranium 2022).
Thus, as for the reserves, the projected production is unequally shared among countries as the four interactive maps below show. A large part of Africa, Central America, Europe, the Near East, South East Asia and some parts of South America have no or hardly any production.
Uranium projected production in tU/y
(Data from Table 1.23. World production capability to 2040, estimates B-II in Uranium 2022: Resources, Production and Demand, p. 90)
2025
2030
2035
2040
Again, this unequal spread of production let us expect future geopolitical competition for uranium production.
Increasing uranium production, timeliness and geopolitics
Being able to have enough uranium in the future when needed will greatly depend, on the one hand, upon the demand, of course, and, on the other, upon the production capacity of the mines being currently exploited, of their remaining lifespan, as well as on the status of current exploration, added to the delay existing between successful exploration and full capacity production in terms of mining and milling. And here, for now, we make abstraction of the remaining part of the nuclear cycle as well as of transportation.
For example, as shown in the timeline below, according to Orano, one of the leading international groups in the nuclear energy sector, between 15 and 25 years are needed between the discovery of a potentially useful deposit of uranium and the start of mining and milling operations (pp. 6-7), assuming there is no unforeseen event of a geopolitical type for instance. Then, a mine will be exploited for 15 to 20 years, followed by a period of 10 years and more to remediate the mining and milling site, possibly converting it, while constantly monitoring it (Ibid.).
Timeline – Uranium exploration, mining and process – redesigned from 2023 Orano’s mining activities, pp. 6-7 with approximately corresponding reserves as explained in NEA/IAEA Uranium 2022, p. 17. (Back to reserves.)
In other terms, assuming we need to add to our resources some totally new uranium site, the discovery of an uranium site taking place at the start of 2024 would correspond to a production starting between 2039 and 2049. Hence if we want to be able to treble our energy production by 2050, all necessary supplementary discoveries of sites must have taken place by 2025 if we want to be absolutely certain to produce enough uranium for 2050.
We know from our analysis of uranium reserves that globally we may use mainly RAR reserves to increase the supply of uranium. We may thus, again globally, focus on these RAR reserves.
In that case, because 5 years are needed between the “decision to mine” and the actual production of a mine, then we shall need to make sure that all “feasibility studies and decisions to mine” are taken 5 years before the need for uranium occurs. It means that to meet 2030 objectives, all “decisions to mine” will need to have been firmly taken by 2025. Not only “planned and prospective production centres” will need to be fully operational, but some further 1646 t will have to be produced from somewhere, either from existing sites where production capabilities will be increased, or from new mines included within the RAR reserves. The challenge will increase each year, with new “decisions to mine” having been taken by 2030 for, at best, 34 768 tU/y.
Practically, the “decisions to mine” translate into having mining permits with the country where the mine is located, then making the last studies and then constructing the industrial facilities, if needed.
Thus, by 2030, globally, the producing mines and milling capacities will need to represent 2,82 times those of 2022.
Obviously, here geopolitics will play an important role as sour relations with a country, competitive or adverse external influence may derail any project. Similarly the security situation within a country will also be key as instability up to civil war, organised crime, terrorist and guerrillas activities will have the power to question mining permits – for example in the case of a coup – or to strongly hinder if not stop the completion of final surveys and the construction of industrial facilities. Obviously, the challenges will continue throughout the production period.
These difficulties are not new but as instability spreads in the world and as international tensions heightens, then the political and geopolitical risks to uranium production will increase. Furthermore, because of the trebling objective for nuclear energy, the stakes linked to uranium production will be higher. As a result, the threats to uranium production will intensify.
Thus, if, globally, we know that we have enough RAR reserves, and it seems we need not worry about availability of supply, this security is partly an illusion.
Apart in statistics, in the real world there is no such thing as globally available uranium resources where each company and country can take anytime whatever amount of supply it needs. Geological surveys, industrial logic, timeline and competition, domestic instability, national interest and international tensions, all taking place under the rising stress brought about by climate change must imperatively be taken into account.
Furthermore, as highlighted above, the shorter timeline between the start of a SMR and its completion may create new challenges for the whole industry as 2 to 3,5 years (the time to build a SMR) is far below the 5 years necessary between the decision to mine is taken and the start of production.
The real question is thus to rise uranium production according to the trebling objectives in a way that allows each nuclear plant, whatever its type, to function and produce energy, while considering resources, mining and milling timeline and process alongside domestic security situation, national interest and geopolitics, while climatic conditions change and become more extreme.
Countries that plan to increase nuclear power capabilities must, in the same time, secure, in a timely manner, their supply of uranium, either through their national nuclear company or through nuclear companies of other nationalities. Similarly, companies will need to plan strategically ahead.
Indeed, as seen above globally, at country level, the energy security stakes linked to uranium will increase for a country as a threat to its uranium supply will mean its electricity production may become dangerously destabilised. The danger will intensify with the rise of electrification advocated (NZE).
To anticipate further the new world that will be born as we seek to fuel the renewal of nuclear energy, we need to find ways to move beyond a global approach, adapted to the nuclear specificities.
Combat robots, AI targeted air strikes, new generation cyber and information warfare, AI generated deep-fakes, unmanned air and sea military vehicles, and smart artillery are being currently massively projected on the theatres of operations of Ukraine, Gaza and the Red Sea.
As it happens, those new weapons systems are different iterations of the projection of AI power on the different dimensions of those battlefields.
According to Hélène Lavoix, we define “AI power” as the intersection between AI and the different conceptions and ways to exercise power (“When Artificial Intelligence will Power Geopolitics – Presenting AI”, The Red Team Analysis Society, November 27, 2017). As we saw in previous articles, in the military field, AI power may translate itself as an enhancement of firepower, as well as of the “observe, orient, decide and act”, i.e. the “OODA” loop (Jean-Michel Valantin, “Militarizing Artificial Intelligence – China (2)”, The Red Team Analysis Society, May 22, 2018).
If we look at the war in Ukraine, AI power, as we shall see, appears to be a multidimensional technological form of continuation of politics and war. This begs the question of the political and strategic meaning of AI (fire) power and of AI powered operational and tactical enhancement. We should also wonder about the counter-measures to these capabilities, and how they mitigate or reinforce the “battlefield AI power”.
Furthermore, the projection of AI power on the current battlefields by several of the belligerent parties is also a way to experiment with those new capabilities in real war and combat conditions. As Carl von Clausewitz establishes, “war is nothing but the continuation of politics with other means” (Carl von Clausewitz, On War, 1832, Penguin Classics, London).
Hence, in this new series, we are going to study how, in war situations, the projection of AI power on the theatre of operations and on the battlefield is the “continuation of politics”. Reciprocally, we are also going to see how, in these situations, AI power merges with politics, and how it creates continuities between the politics of war and warfare operations and tactics.
Applying this approach to Ukraine, this article thus looks at the way the integration of AI power to the politics and war continuum currently operates. First, this article describes how the big U.S. AI companies project themselves in Ukraine. It also analyses the consequences of this technological projection on the Ukrainian state and military. Then, we shall study the way Russia reacts by deploying its own AI-powered capabilities on the battlefield. Finally, we shall analyse the political and strategic consequences that AI power entails.
Since 2022 and 2023, the war in Ukraine has become an attractor and an “in vivo” laboratory for those new capabilities in a war environment (Vera Bergengruen, “How Tech giants helped turn Ukraine into a giant AI war lab”, Time Magazine, February 8, 2024).
As it happens, from 26 February 2022 onwards, i.e. two days after the start of the war, the tech billionaire Elon Musk provided the embattled country with internet service. He achieved this through the connection of portable relay antennas with his Starlink satellites constellation (Ronan Farrow, “Elon Musk’s shadow rule”, The New-Yorker, 21 August 2023.
Soon after that, Starlink became an essential support for the Ukrainian military communications. Thus, the Starlink constellation became a major support of the Observe-Orient-Decide-Act (OODA) loop (Farrow, ibid). One must keep in mind that the OODA loop is a central tool of modern warfar. This process allowss the coordination between command and control authorities, the flows of information and the decision making process with digital tools while abattle is going on.
These command levels are those involved in the management of military operations through the function of coordination of the units, actions, weapons systems, jointness and command during warfare operations. The “intelligentization” (i.e. the addition of artificial intelligence to a system) of C2 and C4ISR and C5ISR (C5 standing for cyber-defense) levels is meant to enhance the rapidity of treatment of data and information flows produced by the deployment of troops, weapons systems, and platforms to such a speed that human capabilities are largely exceeded (Kania, ibid).
The Ukrainian state in the Amazon cloud
In the same time, during the first hours of the war, on 24 February 22, Amazon started to help the Ukrainian government. The giant company offered to secure the Ukrainian state’s administrative and economic data. Then, Amazon uploaded 10 million gigabytes of data in its cloud.
Thus, the Ukrainian administrative system remains operational, and secure from any conventional strike (Tangalakis-Rippert, ibid)).
Palantir as state resiliency solution
Starting in June 2022, the AI company Palantir offered its services to the Kiev government. Palantir, originally partly funded by CIA’s capital venture service, is a data analytics software company. It has been known as the “Google for spies” since its creation in 2004 (Vera Bergengruen,ibid).
It is one of the most advanced AI U.S. companies, that uses machine learning capabilities (knowing that machine learning includes deep learning, see, Hélène Lavoix, “When Artificial Intelligence will Power Geopolitics – Presenting AI“, The Red Team Analysis Society, 27 Nov 2017). Palantir’s specialty is in collecting and treating data for security and Defense organizations.
In Ukraine, Palantir has rapidly become a vital infrastructure. Its deep learning and machine learning capabilities ensure the administrative continuity of the Ukrainian civil service and government, from economics to education and, obviously, the military (Vera bergengruen, ibid). On the military side, Palantir is part of a multi-actors very innovative and territory-wide data extraction system.
It seems that one of the factors for the important Russian losses, at least in military ordnance, results from NATO’s Palantir supported targeting, through the Palantir’s Metaconstellation tool (Margarita Konaev, “Tomorrow’s technology in today’s war: the use of AI and autonomous technologies in the war in Ukraine and implications fr strategic stability“, CNA, September 2023). According to a CNA report, this AI powered tool analyses data produced by sensors on the Ukrainian theatre of operation. Those data flows are completed by satellite observation and electronic interceptions. Then, Palantir identifies targets and produces firepower solutions.
At the start of the war, the Ukrainian ministry of technology modified an app used to download pictures. This app was widely distributed to the civilian population as well as to the troops. Then, Ukrainians started using it to photograph Russian troops or military ordnance and to transmit it with geo-coordinates.
Deep learning software treat those pictures and coordinates in real time. The results are compared with satellite imagery. Targeting units receive then the information, in order to launch an artillery or a drone strike. AI tools power the whole process in order to create an ultra-rapid OODA loop (Robin Fontes and Dr Jorrit Kamminga, “Ukraine a living lab for AI warfare”, National Defense, 3/24/2023).
This loop integrates the air and space U.S. and NATO capabilities, which support the Ukraine war effort. In other words, the U.S. puts its air and space power at work in order to turbocharge the Ukrainian military and the Palantir’s Metaconstellation systems supports the integration of US space power and AI power in order to greatly accelerate the OODA loop.
Indeed, since the 2003 U.S. invasion of Iraq, the U.S. have become a major air and space power. The famous “shock and awe” strategy, devised in order to attain “rapid dominance” in Iraq in 2003 was based on the coordination of space, electronic, and air capabilities with ground forces (Oliver Burkeman, “Shock tactics”, The Guardian, 25 march 2003). It created a “dome” of planes, missiles, artillery, and information, which was used to destroy the Iraqi troops and to protect U.S. forces, while the latter were quickly advancing on Baghdad (Gordon and Trainor, The Endgame, the inside story of the struggle for Iraq, from George W. Bush to Barack Obama, 2012).
In other terms, the satellite systems, the air force and the ground forces were immersed in multi-layer information feeds, which helped coordinating the strikes against the Iraqi forces, and thus reduced greatly any direct contact between the U.S. forces and the Iraqi forces during the invasion phase. Then, the concepts of air, space and cyber dominance were distributed in the different arms of the U.S military.
Since 2003, those concepts and capabilities have become major tenets of all military and warfare operations, from secret and tactical operations to major deployment, as those in the Red Sea, and in Ukraine.
In this context, it is interesting to note that in 2017 Alphabet Inc., the parent company of Google, partnered with the U.S. military. This partnership was known as Project Maven. Contested by Google employees, it was abandoned in 2018 by Google. However, the Department of Defense kept it alive without Google.
The Agency develops Maven Project with a number of of partners, among them Palantir, in order to turn the project into an AI machine learning combat situational awareness tool, able to intervene on the battlefield (Steven Musil, “Palantir extends US Defense contract that prompted protest at Google”, CNet, Sept. 29; 2022).
It is now in its development phase. The goal of this high interest project is to develop unmanned systems AI powered systems able take pictures, data, imagery and full motion captures and to process those in order to identify targets.
There are not a lot of detailed open sources about project Maven’s development. However, it seems that the Ukraine war is a testing ground for some beta applications related to this project. For example, the Maven AI uses data from the theatre of operations to train itself. Especially, this training aims at identifying Russian equipment from the general landscape and destroying it (Jack Poulson, “Where drone warfare intersects with Ukraine and U.S information operations”, Tech Inquiry, 2023/13/03).
And then, there were drones
It is in this technological context that, by October 2023, Turkey would have provided Ukraine with 50 of the Bayraktar TB2 air drones (Agnes Helou, “With Turkish drones in the headlines, what happened to Bayraktar TB2?, Breaking Defense, October 6, 2023). These are unmanned autonomous drones, with observation and bombing capabilities (Maragarita Konaev, ibid).
In the same time, the Ukrainian military started to use small drones, including the U.S. Switchblade Tactical, developed in 2015 by the U.S. Army, in order to support the troops in Afghanistan. Those drones are for observation, telemetry, targeting and bombing (Michael Peck, “Kamikaze drones add a new layer of lethality to remote forces”, C4ISR, 19 September, 2015).
The Ukrainian military also uses the Phoenix Ghost drone, a loitering munition (.e. an explosive drone) (Konaev, ibid). Then, on 1 February 2024, swarm drones sank a Russian small warship in the Black Sea (Ellie Cook, “Ukraine deployed “swarming tactics” to sink Russian warship”, Newsweek, 8 February, 2024).
AI is a wolf for AI
However, if the Americans mobilise AI power to support the war in Ukraine, the same is true on the Russian side and its allies. The war triggers a race for counter-measures against the Western militarization of AI power.
For example, Russia rapidly developed a symmetric effort in the use of drones. Since 2023, the Russian military completes its own drones arsenal with the Iranian Shaheed 132 and Mojaheed drones. Those machines are equipped with active electronic warfare weapons systems. Then, those weapons systems disabled numerous Ukrainian drones, especially the Bayraktars (Konaev, ibid, Agnes Helou, “With Turkish drones in the headlines, what happened to Bayraktar TB2?, Breaking Defense, October 6, 2023 and Jeff Mason and Steve Holland, “Russia hundreds of Iranian drones to attack Ukraine, US say”, Reuters, June 10, 2023).
It also appears that, from the Ukrainian side, one of the answers to the massive Russian use of electronic warfare has been the integration of AI to drones. This way, the development of the autonomy of new drones would compensate for the loss of ground control (Konaev, ibid and Paul Mozur and Aaron Krolik, “The invisible war in Ukraine being fought on over radio waves”, International New-York Times, in Deccan Herald, 19 November 2023).
Drones attrition
As it happens, electronic warfare weapons systems are only part of the anti-aerial and conventional weapons systems arsenals, alongside artillery, missiles and anti aerial machine guns. According to a RUSI report, 90% of all drones used by both sides between February and July 2022 were destroyed (Mykhaylo Zabrodskyi, Jack Watling, Oleksandr V. Danylyuk, and Nick Reynolds, Preliminary Lessons in Conventional Warfighting from Russia’s Invasion of Ukraine: February-July 2022, RUSI, Nov. 30, 2022, p. 37).
This entails large production capabilities for Russia, Ukraine, the U.S. and the foreign suppliers.
As American AI companies, Russian AI companies test their innovations on the Ukrainian battlefield. For example, the Russian military used the Lancet-3 loitering munition. The Russian Zala Aerogroup produces the Lancet-3 (Konaev, ibid).
The Lancet-3 is able to autonomously target and strike in designated areas. For example, in June 2022, it was used during the fights in the region of Zaporozhnye region. In July 2023, those drones were also used against the Ukrainian forces during their failed summer counter-offensive (Konaev, ibid).
From robot combat to AI Performative warfare
It must be noted that Russia also launched ground drones, which exchange shots with Ukrainian air drones and with the first Ukrainian ground drones. In other words, Ukraine happens to be the first drone on drone combat battlefield (Zachary Kallenborn, “Robots are fighting robots in Russia’s war in Ukraine”, Wired, 30 January, 2024).
AI is also largely present in the area of what we call the “meta-theatre of operations”. We define this area as the universe emerging from the intersection of information warfare, cyber warfare, propaganda, and political warfare operations. The proliferation of messages, fake information, deep-fake videos, etc… cross the meta theatre of operations. As it happens, AI generates lots of those contents (Konaev).
Then, those are projected in dozens of millions of brains and minds through social media platforms. Facebook, Twitter, Snapchat, Youtube, Telegram, the Russian VK, Chinese WeChat, become both medium and “weapons of massive distribution” of militarized contents. In other terms, the hybridation of AI generation of content with social networks is tantamount to an industrial production of performative warfare operations (Matthew Ford and Andrew Hoskins, Radical War, Data, Attention and Control in the 21stCentury, Hurst Publishing, 2022).
Politics of the AI Battlefield
So, it now appears that Ukraine is a gigantic testing ground both for the militarization and weaponization of AI power and for the political use of AI power in a time of war.
Suspending state failing
The latter reveals itself in the way Amazon, Starlink and Palantir have literally become the electronic and data platforms and infrastructures of the Ukrainians. Thus, at least, those giant U.S. tech companies were able to literally “suspend” the risk of administrative discontinuity of the Ukrainian state ( On warfare and collapsing states, see Lawrence Freedman, The Future of War: a History, Penguin Books, 2017 and David Kilcullen, The Dragons and the Snakes, How the Rest Learned to fight the West, Hurst, 2020).
That was a crucial political and strategic move. Indeed, the Ukraine state, i.e. the complex of institutions guaranteeing civil as well as military service, is ensuring the security and safety of its territory and population. At the very least, its economic, fiscal and finance systems is what enables the Ukraine military functioning.
As Clausewitz establishes, war is the continuation of politics with other means. So, the projection of AI (military) and (fire) power on the Ukrainian theatre of operations and on the battlefield is one of the families of the means of this continuation. Historically, it is also the first iteration of AI power in war operations and combat.
In the case, for example, of Palantir, its capability to treat massive amounts of data turns it into an entity that establishes a continuum through the “fog of war” to the daily complexities of a relatively (dis)functional Ukraine public service and economy in a time of war (Slawomir Matuszak, “A struggle to survive. Ukraine’s economy in wartime”, Centre for Eastern Studies, 2022-10-18).
In other terms, Palantir becomes both the digital mean of the continuity of state as well as its military “continuation through other means”. The machine learning of Palantir is both politics and its OODA loop continuity.
What about electricity?
This hybridization of the political and military institutions as well as of combat zones with AI power turns the electric grid into a strategic and existential asset. In this perspective, the constant Russian hammering of the Ukrainian electric grid and power utilities takes on an anti-AI power dimension.
So, depriving Ukraine of electricity means not only weakening the Ukrainian economic and life conditions, but also the power conditions of the U.S. AI hybrid entity that the Ukrainian state and military have become (Talya Vatman, “Russian attacks on Ukraine energy sector have escalated again as winter sets in”, International Energy Agency / IEA, 17 January 2024).
The strategic significance of this new political situation reveals itself since the start of the war. For example, Palantir has played a key role in communications during several battles, including the fierce siege of Mariupol, while helps restoring civil and military communications. (Konaev).
However, in October 2023, several Ukrainian units had to face communication’s outages because of a failing Starlink internet connection. So, they were forced to stop and commanders had to urgently come on the frontlines (Ronan Farrow, ibid).
In other words, the hybridation of the Ukraine military and of Starlink created for Ukraine a situation of profound technological and political dependency upon the will of a private foreign person. In return, this person has acquired an immense power through its satellite constellation and communication devices.
Reciprocally, the political status of Amazon, Starlink, and Palantir becomes something quite complex and ambivalent. Those are private U.S. companies that are able to project their AI power in order to hybrid themselves with the civilian and military structures of a foreign state. Thus, they make this state dependent upon their good will.
AIs and the flight of the Fates
In other words, Palantir, Starlink, Amazon, Alphabet and some others are becoming the technological and political equivalent of what the Greek gods were for the troops, warriors and political leaders during the Trojan war. Following this line of thought, AIs become the equivalent of transcendant beings able to intervene in human affairs, such as war (Peter Sloterdijk, After God, Polity Press, 2020).
However, if this is quite true on the Ukrainian side of the war, it is not so on the Russian side. President Putin, its government and the Russian state are not dependent on foreign AI platforms. The conventional war they wage against Ukraine also integrates and tests AI power. But in this case, the AI power development dynamic is at the service of the Russian national interest.
From AI (fire) power to AI World
Thus, from a geopolitical point of view, one may say that the AI dimension of the Ukraine war opposes the U.S. public-private AI power integration into the Ukrainian state structures to the Russian AI national power. It can also be said that, for now, the sovereignty of the Ukrainian state on the unoccupied parts of its territory is technologically dependent from the same US AI power.
However, as the Starlink-Musk incident emphasizes, this sovereignty is very fragile if, or when, the foreign AI power switches off. Those political continuities induced by AI power are thus creating powerful political and strategic ambiguities and vulnerabilities.
This political ambiguity is even more pervasive. Indeed, on the Ukrainian side, private capabilities appear as being the bulk of AI power. Meanwhile, the US Department of Defense satellite and space power capabilities have to combine with them. Thus, the Ukraine war becomes an additional driver of the US political and military authorities dependency on private technologies and interests.
Then, the massive injection of AI power in the Ukrainian war turnspower generation and distribution as the very existential condition for the continuity of the Ukrainian state and the efficiency of its military arm. This dynamic also means that the war has turbocharged the integration of Ukraine to “AI World”, that is to say the world emerging from the development and implementation of AI (Hélène Lavoix, “When Artificial Intelligence will Power Geopolitics – Presenting AI”, The Red (Team) Analysis Society, November 27, 2017).
This current reality begs the question of the use of AI power in other theatres of operations, especially in asymmetric warfare.
Indeed, how efficient is AI on theatres of operations and battlefields that drastically limit the efficiency of targeting while dramatically increasing the very costs of ultra-modern war for the AI power equipped foe?
That is what we shall see with the Gaza and Red Sea war case study.
“In today’s highly volatile and rapidly changing environment, it is imperative for our contemporary societies and for all players, whether private or public, to reduce the uncertainty associated with the future. The primary objective is to survive and progress in the best possible way. To achieve this, we need to identify and assess the changes that are already underway and that we will have to face in the future.”
Julien Devaureix, Introduction to the Podcast “Anticipate and Get Ready for the Future – Hélène Lavoix”, Sismique, 2 April 2024
Interview in French (check below on Youtube for an automatic translation in English) Access the page of the interview on Sismique (transcript and notes).
What are we talking about? 0:00 Introduction 9:04 The importance of Strategic Foresight 17:36 Understanding World Dynamics 22:16 Weak signals and anticipation 40:54 Mechanisms to restrict power. 49:29 Towards a Mad Max life. 52:41 Tensions between democracies and authoritarianism. 55:22 Kant on democracy and universal law. 1:06:14 Democracy in question 1:12:53 The impact of violence on society 1:19:58 Anticipate to better protect yourself 1:23:13 The importance of individual foresight“
The sequences of the interview of Dr Lavoix by Julien Devaureix, Sismique
The Red Team Analysis Society uses cookies to ensure that we give you the best experience on our website. This includes cookies from third party social media websites if you visit a page which contains embedded content from social media. Such third party cookies may track your use of the Red Team Analysis Society website.
If you click on "Accept", you accept our policy, we'll assume that you are happy to receive all cookies on the Red Team Analysis Society website and this will close this notice. Accept AllGDPR and Cookie Policy
Privacy & Cookies Policy
Privacy Overview
This website uses cookies to improve your experience while you navigate through the website. Out of these, the cookies that are categorized as necessary are stored on your browser as they are essential for the working of basic functionalities of the website. We also use third-party cookies that help us analyze and understand how you use this website. These cookies will be stored in your browser only with your consent. You also have the option to opt-out of these cookies. But opting out of some of these cookies may affect your browsing experience.
Necessary cookies are absolutely essential for the website to function properly. This category only includes cookies that ensures basic functionalities and security features of the website. These cookies do not store any personal information.
Functional cookies help to perform certain functionalities like sharing the content of the website on social media platforms, collect feedbacks, and other third-party features.
Performance cookies are used to understand and analyze the key performance indexes of the website which helps in delivering a better user experience for the visitors.
Analytical cookies are used to understand how visitors interact with the website. These cookies help provide information on metrics the number of visitors, bounce rate, traffic source, etc.
Advertisement cookies are used to provide visitors with relevant ads and marketing campaigns. These cookies track visitors across websites and collect information to provide customized ads.
