The race to power artificial intelligence is shifting from the data center to the mine shaft. While tech giants like Google are investing in small modular reactors (SMRs) to fuel their AI infrastructure, a critical bottleneck is emerging: the uranium needed to keep those reactors running. A newly public mining company and a coalition of nuclear startups are now competing to establish a domestic uranium supply chain that could determine whether the United States can sustain its AI ambitions without relying on foreign energy sources .

Why Is Uranium Supply Suddenly Critical for AI?

The connection between uranium and artificial intelligence might seem indirect, but it reflects a fundamental shift in how the tech industry thinks about power. Data centers running large language models and other AI systems consume enormous amounts of electricity, and unlike traditional computing workloads, AI compute cannot tolerate power interruptions. Nuclear energy offers something renewables cannot: reliable, carbon-free baseload power that runs 24/7 without weather dependency .

The U.S. Department of Energy is actively pushing this vision. Secretary of Energy Chris Wright has made nuclear energy central to his strategy for powering AI infrastructure, emphasizing both energy independence and national security . However, the supply chain for nuclear fuel has atrophied over decades. The United States currently imports most of its uranium, creating a vulnerability that policymakers and industry leaders now view as a strategic risk.

What's Happening on the Ground Right Now?

Three major developments are unfolding simultaneously, each addressing a different piece of the puzzle:

• Reactor Construction: Kairos Power, a California-based nuclear startup, broke ground on Hermes 2 in Oak Ridge, Tennessee, which will produce 50 megawatts of carbon-free electricity to power Google data centers in Tennessee and Alabama. This is the first fourth-generation nuclear reactor to demonstrate power production in the United States .
• Uranium Mining: Eagle Nuclear Energy, which went public on NASDAQ in February 2026 under the ticker NUCL, is developing the Aurora Uranium Project at the McDermitt Caldera in Oregon. The company has identified an estimated 37 million pounds of uranium, described as the largest extractable uranium deposit in the country .
• Bridge Power Solutions: A partnership between Denham Capital and First American Nuclear is deploying temporary on-site gas power generation to data centers while waiting for nuclear reactors to come online, potentially reducing interconnection delays by five or more years .

These three pieces represent a coordinated bet that the nuclear industry can move faster than traditional utility timelines. Kairos Power's approach is particularly instructive. The company is using modular construction methods, including precast concrete and seismically isolated foundations, to reduce both costs and timelines. The lessons learned from building Hermes 1, a demonstration reactor, are being directly applied to Hermes 2 .

"For nuclear projects to be successful, we need more than just the right technology. We need to understand every aspect of project delivery. Hermes 2 is where that all comes together," said Mike Laufer, co-founder and CEO at Kairos Power.

Mike Laufer, Co-founder and CEO at Kairos Power

Eagle Nuclear Energy's CEO Mark Mukhija frames uranium mining as both an economic and national security imperative. The company is not just mining uranium; it is also planning to refine it at a site in Nevada, attempting to rebuild domestic capability for fuel production that has largely disappeared from the United States .

"I see it as a national security issue. I see it as an energy independence issue," said Mark Mukhija, CEO of Eagle Nuclear Energy.

Mark Mukhija, CEO of Eagle Nuclear Energy

How Does This Solve the AI Power Bottleneck?

The U.S. data center market is expected to exceed 100 gigawatts of demand by 2028, according to industry projections . Traditional utility interconnection can take five to ten years, creating a gap between when companies need power and when the grid can provide it. The nuclear supply chain strategy addresses this in multiple ways:

• Immediate Power: Gray Oak Power, a platform launched by Denham Capital, provides on-site gas generation that can be deployed within months, not years, allowing data centers to begin operations while waiting for nuclear infrastructure .
• Long-Term Reliability: Once SMRs like Hermes 2 come online, they provide stable, carbon-free power that does not depend on grid interconnection or weather conditions, making them ideal for the continuous operation AI workloads require .
• Fuel Security: By developing domestic uranium mining and refining capacity, the U.S. reduces dependence on foreign suppliers and creates a closed-loop supply chain that can sustain multiple reactor deployments .

The economics are also shifting. Kairos Power is using standardized, repeatable designs that can be manufactured at scale, reducing per-unit costs. Google's Global Head of Data Center Energy, Amanda Peterson Corio, emphasized this point when discussing the Hermes 2 project .

"By pioneering a standardized, repeatable design, Kairos Power is addressing the historical challenges of nuclear construction costs. This shift toward a more efficient, factory-based manufacturing approach is a proven path toward lower-cost, cleaner power for our operations and the communities we serve," said Amanda Peterson Corio, Global Head of Data Center Energy at Google.

Amanda Peterson Corio, Global Head of Data Center Energy at Google

Steps to Building a Nuclear-Powered AI Infrastructure

• Secure Fuel Supply: Develop domestic uranium mining and refining capacity to ensure long-term availability and reduce geopolitical risk, as Eagle Nuclear Energy is doing at the McDermitt Caldera .
• Deploy Modular Reactors: Use standardized, factory-built reactor designs like Kairos Power's SMRs to reduce construction timelines and costs compared to traditional large-scale nuclear plants .
• Bridge the Timeline Gap: Install temporary on-site power generation while nuclear reactors are under construction, allowing data centers to begin operations years earlier than traditional utility interconnection would permit .
• Align Policy and Incentives: Create regulatory frameworks that allow private companies to develop, own, and operate nuclear infrastructure for data centers, as the Department of Energy is actively promoting .

What Are the Timeline Challenges?

Despite the momentum, experts caution that nuclear power will not immediately solve the AI power crisis. In India, where similar nuclear-AI infrastructure plans are underway, analysts estimate that meaningful nuclear contributions to data center energy demand will not materialize until after 2032, with large-scale impact closer to 2040 . The United States is likely facing a similar timeline.

Kairos Power's Hermes 2 is expected to begin operations in the coming years, but scaling from one 50-megawatt reactor to the dozens or hundreds needed to power a national AI infrastructure will take time. Eagle Nuclear Energy's uranium mining project still requires state and federal permits before extraction can begin . These are not overnight solutions.

However, the combination of bridge power solutions, modular reactor designs, and domestic fuel supply development creates a pathway that did not exist even two years ago. The nuclear industry is moving faster than it has in decades, driven by the urgency of AI infrastructure demands and the recognition that energy independence is now a competitive advantage.

For data center operators and AI companies, the message is clear: nuclear power is no longer a theoretical future solution. It is becoming a practical, near-term option for securing reliable, carbon-free energy. The companies building the reactors, mining the fuel, and bridging the timeline gaps are betting that the AI boom will sustain long enough to justify the investment. So far, the market is agreeing with them.