No other metal carries as much energy per kilogram as uranium. If every atomic nucleus in one kilogram is split, it releases roughly as much energy as three thousand tonnes of coal. And it has only one customer. All uranium mined today ends up in reactors. Demand is therefore determined wherever decisions about nuclear power are made. And those decisions are now moving in the same direction almost everywhere: seventy reactors are under construction, states are extending the lifespans of plants that were slated to close, and the world's largest technology companies are showing keen interest in nuclear power's potential as baseload power.

The world has been here before. Nuclear power has already lived through an entire cycle of optimism, expansion and retreat. To understand where demand is heading, and why supply has become a question of power politics, the story is best told from the beginning. And it begins with a speech.

Atoms for peace

The year is 1953. The nuclear arms race is in well under way, and during the warm summer months the Soviet Union detonates its first hydrogen bomb in Kazakhstan. A few months later Dwight D. Eisenhower, the US president, feels compelled, as he himself put it, to stand before the UN General Assembly and speak the language of atomic warfare. 

As he himself admits, the United States already possesses a nuclear capability and stockpile many times greater than the combined explosive force of all the bombs and shells fired from every aircraft and every gun, on every theatre of war, throughout all the years of the Second World War. At the same time, he points out that this dreaded secret (the machinery of atomic power) no longer belongs to the United States alone. 

Although the speech signals American power, Eisenhower also puts forward the proposal that a few years later led to the International Atomic Energy Agency, the IAEA. The idea was to give the agency access to fissile material that could serve humanity's peaceful endeavours. Experts would be mobilised to apply atomic energy to the needs of agriculture, medicine and other peaceful pursuits in the new world order. 

Eisenhower's speech is therefore often seen as the starting point for the commercialisation of nuclear power. The following year, 1954, the United States pushed through a new Atomic Energy Act, giving companies access to previously classified technology along with the right to own and operate reactors. The act thus laid the groundwork for the so-called Agreements for Cooperation, bilateral cooperation agreements that a large number of countries signed on to. The United States supplied research reactors, trained thousands of foreign nuclear technicians, and lent out enriched uranium as fuel. Ownership, as a rule, remained with the American state rather than being an outright sale, so that the material could be recalled if it were ever needed in wartime. 

On the surface, Atoms for Peace was a peace project, but behind the idealistic words was a strategic calculation. By sharing knowledge and technology, the United States could gain influence and bind recipient countries to American technology, American fuel and the American regulatory order. The Soviet Union did the same within its own bloc, and gradually a contest emerged over strengthening the energy supply of one's own allies.  

The logic behind that contest was simple: energy is power in its most fundamental form. The availability of energy sets the limit on a country's industry, its growth and, by extension, its military capability. Moreover, every barrel of oil that doesn't have to be burned in domestic power plants can instead be exported, stored or put to other uses. Whoever gained the atom gained capacity. 

The golden age: optimism, expansion and retreat

The promise of a peaceful atomic age took two decades to fulfil. The reactors planned during the 1960s were not completed until the early 1970s. Many were built on American licensed technology spread through Atoms for Peace, but Britain, the Soviet Union and France also had their own tracks. The timing could not have been better. When OPEC's oil embargo in 1973 quadrupled the price of oil within a few months, nuclear power stood there as a proven alternative, at the very moment the West realised how dependent it had become on oil from the Middle East. Energy security, the freedom not to have to rely on others' energy, became the driving force that led one country after another to build nuclear power. 

The construction statistics show how strong that driving force was. Global nuclear power generation stood at just over 80 terawatt-hours in 1970; by 1975 it had more than quintupled, and by 1980 around 700 terawatt-hours were being produced per year. Construction starts soared in the years after the oil crisis, and at the peak around 1980 more than 230 reactors were being built simultaneously. No period before or since comes close.

But while the curve pointed steeply upwards, a countervailing force grew in its shadow. As early as the late 1960s, researchers had begun asking questions, about safety and, of course, about waste. In the early 1970s the doubts coalesced into a movement. The 1975 protests against the reactor construction at Wyhl in southern Germany halted the process and inspired opposition across Europe. In Sweden, the Centre Party took a stand against nuclear power as early as 1973 and won the 1976 election partly on that issue. The optimism to which Eisenhower had given voice in his speech was giving way to scepticism.

When the meltdown at Three Mile Island occurred in 1979, it confirmed these fears, and nuclear power fell into a crisis of popular and political legitimacy. The accident triggered the Swedish referendum of 1980 and turned public opinion in the United States. Seven years later, when the reactor at Chernobyl exploded, opposition hardened across Europe. Italy voted out its nuclear power, and in country after country new construction ground to a halt.

With the Fukushima disaster in 2011 came the real retreat. Japan shut down its entire reactor fleet (some fifty reactors) and Germany began its phase-out over the following decade. For the first time, world production fell noticeably, from around 2,730 terawatt-hours in 2010 to about 2,430 in 2012, a drop of just over ten per cent.

But history sometimes seems to move in waves, and nuclear power is no exception. Twice before, the pursuit of energy security and energy sovereignty had propelled the technology forward (in 1953 and 1973), and that force was now returning for a third time.

The new crisis, and perhaps the new golden age 

With Russia's invasion of Ukraine in 2022, the winds turned. Energy security once again became the words on everyone's lips, and nuclear power once more appeared as the answer. In Europe, reactor lifespans were extended and phase-out plans were torn up. The United States turned too, with broad cross-party support, government subsidies and restarts of reactors that had previously been shut down. Sweden, which after the 1980 referendum had decided to phase out nuclear power by 2010, abolished the ten-reactor ceiling in 2023, opened the door to reactors at new sites, and is planning new capacity at Ringhals. And since January 2026 the ban on uranium mining, introduced as recently as 2018, has been lifted, and uranium is once again a concession mineral.

In the media this is called nuclear power's renaissance. But the renaissance had already been under way for nearly two decades, just in another part of the world. While the West hesitated after the Fukushima accident, China kept building, from a handful of reactors in the 1990s to the world's largest programme today. What looked in 2022 like a sudden reversal was rather the West joining a movement already in full swing in the East.

In the West, the very language around nuclear power has changed as the question of energy security has grown. Having been portrayed as a danger after the Fukushima accident, in the second half of the 2010s it became a climate tool, a fossil-free power source that could carry the energy transition. After 2022 the rationale shifted again, from climate to energy security. What was once to be phased out is now presented, in country after country, as a guarantor of independence.

Data centres as the new uranium consumer

Even as the world's governments look to nuclear power as an answer to energy security, a new buyer has stepped forward. The big technology companies' data centres for artificial intelligence require enormous amounts of electricity, and they need it around the clock, all year round. It is a baseload profile that weather-dependent solar and wind struggle to deliver on their own, but which nuclear power fits well. The IEA singles out data centres as a new, dedicated market for nuclear power. That gives demand for stable baseload power a commercial driving force.

In the autumn of 2024, Microsoft signed a twenty-year agreement with the utility Constellation to restart a shut-down reactor in Pennsylvania. The plant is called Three Mile Island. The very same facility that in 1979 triggered nuclear power's first crisis of confidence is now to supply AI models with electricity, and Microsoft has committed to buying the entire output for twenty years. The place where the previous nuclear wave stalled has become the starting point for the new one.  

Microsoft is not alone. Other major AI companies have struck similar deals totalling more than ten gigawatts (to be set against the United States' entire nuclear fleet of 97 GW) spread across restarts, lifespan extensions and next-generation small reactors. Their data centres need electricity every hour of the year, and they pay for guaranteed supply rather than haggling over price. One gigawatt consumes around 150 tonnes of uranium a year, so every deal translates into uranium demand already this decade. 

The outlook for uranium demand

Over its short history, nuclear power has served both as a promise of peace and as a lever of power. Today a new era of nuclear growth appears to be on the way: around seventy reactors are under construction worldwide (the highest level since 1990) and construction is already under way in China, India and Turkey. A reactor that comes online consumes uranium for at least sixty years to come. Demand for the coming decade is therefore, in practice, decided when the plans are put into action. 

The World Nuclear Association expects reactors' fuel requirements to grow sharply through 2040, from today's roughly 68,900 tonnes of uranium a year. In the lower scenario the increase stops at just over 107,000 tonnes; in the reference scenario the need more than doubles, to just over 150,000 tonnes; and in the upper scenario it approaches 204,000 tonnes, almost a tripling. Behind the figure lie both new builds and old plants given new life: more than sixty reactors (roughly one in seven in the world fleet) have had their operating licences extended, and at the climate summits 33 countries have pledged to at least triple capacity by 2050. Investment in nuclear power has grown by 50 per cent in five years and, according to the IEA, is on track towards around 75 billion dollars a year as early as 2025, a rise the agency expects to continue through the rest of the decade. 

Against that demand stand buyers who have long postponed their purchases. At the end of 2024, US operators had, according to the Energy Information Administration (EIA), contracted around 106,000 tonnes of uranium of their ten-year requirement, while just over 83,000 tonnes (four in ten) still lacked a supplier. And the United States does not appear to be alone in the lag. According to industry data, for more than thirteen years the world's operators have signed new contracts at a slower pace than the reactors burn the fuel; between 2021 and 2025 alone, just over 310,000 tonnes were consumed while around 225,000 were contracted. The difference is waiting to be bought up. 

The delay carries a price that is already visible in the contracts. The uranium delivered to American reactors in 2024 cost, according to the EIA, an average of just under 53 dollars per pound, while the contracts newly signed that same year stood at just over 86. 

The demand side is therefore fairly easy to read. Construction of new reactors is in full swing, the buyers are states and technology companies with security of supply as their top priority, and postponed purchases are waiting to be executed. The interesting question lies on the other side: can the ground deliver what the market demands?