Why Uranium Is Near $90—and What AI Power Demand Has to Do With It

Uranium’s return to roughly $90 per pound is best understood as a collision of future power demand, delayed utility buying and constrained supply. AI data centres have made reliable electricity strategically valuable, helping validate nuclear restarts, life extensions and new capacity. Yet the strongest direct signal comes from the uranium market itself: utilities need to secure fuel years ahead, while available long-term supply looks limited.

The key price signal is in long-term contracts

Uranium spot reached $90.20 per pound on September 10, 2026. Meanwhile, TradeTech’s long-term uranium price indicator reached $97/lb at the end of June, up $10 from the end of 2025. 1216

That gap matters. Spot pricing reflects material available for relatively prompt delivery; the long-term indicator reflects the cost of securing multi-year supply. A higher term price suggests buyers place a premium on delivery certainty, rather than simply competing for today’s available pounds.

This is why a headline about uranium above $90 should not be read as an AI-only trade. The near-term market is primarily responding to procurement risk: utilities’ need to replace consumed fuel and lock in supply well before it is needed.

AI is creating a new reason to commit to nuclear power

Large technology companies are seeking firm electricity for data-centre expansion, and nuclear power has become one option in that search. Microsoft, Google, Amazon and Meta had contracted close to 9.8 gigawatts of nuclear capacity through 13 deals as of August 2026, according to Pitt Street Research. It estimated that about 80% of that capacity was not yet generating electricity. 17

The distinction is important. A power-purchase agreement, reactor restart agreement or capacity commitment does not immediately increase uranium burn. Its effect on the fuel market arrives as reactors operate, extend their lives, restart or are built—and as their owners arrange fuel procurement.

In other words, AI is strengthening the long-duration demand outlook. It gives nuclear developers and utilities a clearer commercial case for preserving or adding firm generation, but it does not by itself explain every move in the spot market.

Under-contracting makes future buying more consequential

Sprott reports that U.S. and European utilities face substantial uncovered uranium requirements after 2030. It also reported only about 37 million pounds of global uranium contracting as of August 10, 2026, excluding two announced Indian purchase agreements. 224

Utilities can defer purchases when they hold inventories or have existing contracts. But deferral does not eliminate fuel needs. If more buyers return to the market at once to cover later years, competition for long-term production can increase sharply.

This is the core structural thesis behind the rally: contract coverage is falling as reactor requirements rise, while developing reliable new mine capacity takes time.

Mine output still trails reactor requirements

The supply imbalance is not necessarily a simple annual physical shortage, because secondary sources can fill part of the gap. Still, the scale is notable. A comparison based on the World Nuclear Association’s 2025 Nuclear Fuel Report puts annual reactor requirements near 70,000 tonnes of uranium, versus mine production near 60,000 tonnes. Secondary supply—including inventories, government stockpiles and reprocessed fuel—has helped bridge the difference. 22

Pitt Street Research similarly estimated 2025 reactor requirements at roughly 68,900 tonnes, with primary mine supply providing about 160 million pounds of U3O8 and secondary sources supplying the balance. 17

The implication is not that reactors will suddenly run out of fuel. It is that inventories and secondary supplies remain important to market balance. If those buffers decline while utilities remain under-contracted, new mine projects and restart capacity become more valuable—but also more necessary.

Sprott can tighten the investable uranium market—but flows can reverse

The Sprott Physical Uranium Trust is an important financial-market link because it can acquire and hold physical uranium. One report put Sprott Asset Management’s holdings through the trust at 81.4 million pounds of U3O8 in mid-2026. 4

When physical funds raise capital and buy uranium, they can reduce material readily available to other buyers and add support to a relatively small spot market. But this is not a one-way mechanism. Fund purchases depend on investor demand and the trust’s ability to raise capital; weaker flows, discounts to net asset value or redemptions can reduce that support.

That makes the uranium price partly a fundamentals story and partly a capital-flows story.

What the TSX30 says about the broader trade

Canada’s 2026 TSX30 illustrates how broadly investors have rewarded the resource, infrastructure and power theme. The exchange’s top 30 performers posted a record 785% average dividend-adjusted share-price gain over three years and added $225.7 billion in market value, according to the Toronto Stock Exchange. 37

Mining companies accounted for 18 of the 30 companies in the ranking. 32 The TSX30 is based on three-year dividend-adjusted performance, so it is not a measure of a single year’s uranium return or proof that all mining shares benefited from the same catalyst. Still, its composition shows strong investor interest in metals, resource development and the buildout of physical infrastructure.

AI contributes to that wider investment narrative because data centres require not only generation capacity but also transmission, grid equipment, cooling and construction. That broadens attention beyond uranium to materials associated with electrification and infrastructure, including copper. The result is better described as a power-and-infrastructure cycle than as a pure nuclear-fuel trade.

Is the uranium rally sustainable?

There is a credible fundamental case for elevated uranium prices: long-term contract prices are high, utility coverage declines later in the decade, mine supply remains below reactor requirements without secondary sources, and technology companies are adding long-dated support for nuclear generation. 2161722

But the case is not risk-free.

• AI-linked nuclear capacity is largely prospective. Much of the announced capacity is not yet generating, and projects can face permitting, financing, construction and restart delays. 17
• Spot and term markets can move differently. A strong long-term indicator does not guarantee a sustained spot-price rise. 1216
• Secondary supply and inventories still matter. They can cushion the market, though deferred buying may leave more demand to be covered later. 1722
• Uranium equities are not uranium itself. Mining shares depend on operations, costs, permitting, financing and valuation. Uranium mining equities fell 3.9% and juniors fell 7.4% in the first half of 2026 even as the long-term uranium price rose to $94/lb, according to one market analysis. 4

The durable conclusion is that AI has made dependable power a more valuable strategic asset, and that supports the long-term nuclear narrative. The current uranium rally, however, rests most directly on a practical fuel-market issue: utilities must secure future supply in a market where dependable new production is slow to arrive.