The U.S. government has set a countdown clock that could fundamentally reshape how artificial intelligence data centers get their power. On May 13, 2024, the Prohibiting Russian Uranium Imports Act created a hard deadline: all waivers allowing Russian low-enriched uranium (LEU) imports must completely terminate by January 1, 2028. This is not a gradual phase-out or an indefinite grace period. It is a 12 to 18-month window before utilities with unhedged reactor requirements face a genuine supply crisis.

To understand why this matters for AI infrastructure, you need to know the scale of the dependency. Russia's state-owned Rosatom controlled approximately 44% of global enrichment capacity and supplied nearly 25% of the low-enriched uranium used by U.S. utilities to power 20% of the national electric grid. That uranium was funding Russia's nuclear weapons complex and military-industrial apparatus. The ban cuts off that supply entirely, forcing domestic alternatives to fill the gap within a fixed timeline.

Why Does This Matter for AI Data Centers?

Global data center electricity use is projected to double to 945 terawatt-hours (TWh) by 2030, with artificial intelligence representing 50% of that growth. Unlike solar or wind power, which depend on weather, nuclear energy provides continuous, dispatchable gigawatt-scale power that can run 24/7. For hyperscalers like Microsoft, Google, and Meta building massive AI training facilities, nuclear is increasingly the only carbon-free baseload option that can reliably power their operations.

Microsoft's decision to restart Three Mile Island in Pennsylvania and Google's first-ever corporate small modular reactor (SMR) agreement signal that major tech companies are betting on nuclear as their long-term power solution. But those deals depend on a stable, affordable uranium supply. The Russian import ban creates both urgency and opportunity: urgency because utilities need domestic enrichment capacity online before 2028, and opportunity because companies that can secure that supply will have a competitive advantage in the AI infrastructure race.

How Does the Government Plan to Build Domestic Uranium Capacity?

• Direct Equity Investment: The U.S. government is deploying preferred stock, convertible instruments, and long-term power purchase agreements, not just grants. The Department of Defense acquired a 15% ownership stake in MP Materials via $400 million in preferred stock and warrants, becoming the largest shareholder.
• Strategic Capital Office Funding: The Pentagon's requested budget for the Office of Strategic Capital (OSC) is $20.2 billion for fiscal year 2027, up from $1.5 billion in 2026, a 1,247% increase in a single budget cycle. The OSC offers direct loans pegged to Treasury rates with deferred payment grace periods matched to the useful life of physical assets.
• Supply Chain Security Designation: The government signals future capitalization through regulatory triggers: DPA Title III invocation for critical materials, OSC Covered Technology designation for sub-market Treasury-pegged loans, NRC docketing and license progression to compress time-to-market, and federal land leases to reduce environmental permitting costs.

The structural difference between this intervention and previous energy subsidies is ownership. When the government holds equity in a strategic asset, the political incentive to protect that investment persists across administrations regardless of party. This is not a subsidy that expires with a change in administration. It is an ownership position with symmetric upside participation.

Is This a Real Structural Shift or Just Hype?

Investors have watched clean energy narratives collapse before. In 2021, renewable energy stocks crashed despite compelling environmental arguments. In 2019, cannabis stocks imploded after years of hype. The honest question is whether the nuclear thesis is genuinely different this time, or whether current stock valuations already reflect perfection.

The structural case rests on four distinct layers. First, whether AI power demand genuinely requires nuclear baseload power, not just renewable energy. Second, whether federal intervention represents durable non-dilutive capital or temporary political spending. Third, whether the Russian uranium ban creates genuine structural tightness or a manageable transition. Fourth, whether current equity valuations reflect the structural thesis or have already priced in all future growth.

The January 2028 deadline is the key variable. It is not a soft target or a negotiable timeline. It is a hard legislative deadline with progressively decreasing annual caps on Russian imports leading up to the cutoff. In the 12 to 18 months before that date, utilities with unhedged reactor requirements are expected to compete aggressively for domestic uranium supply. That competition will either validate the structural case or expose it as overblown.

For AI infrastructure investors and data center operators, the implication is clear: companies that secure long-term nuclear power agreements before 2028 will have a competitive advantage over those still dependent on grid power or renewable energy. The uranium deadline is not just a supply chain issue. It is a structural catalyst that could reshape the entire economics of AI infrastructure for the next decade.