Artificial intelligence has created an energy crisis that solar panels cannot solve. The hyperscalers building AI infrastructure, including Microsoft, Amazon, Google, and NVIDIA, have discovered a hard constraint: electricity demand. Training large language models and running inference clusters requires round-the-clock, reliable power that solar and wind cannot consistently deliver. This realization is reshaping how companies secure energy and which investments are capturing the returns.

Why Are Tech Giants Abandoning Solar for Nuclear?

The math is straightforward. A single ChatGPT query consumes roughly ten times the energy of a Google search. Training next-generation large language models requires power equivalent to small cities. Goldman Sachs Research projects global data center power demand will surge up to 165 percent by 2030 compared to 2023 levels. The grid was built for predictable growth of 1 to 2 percent annually, not the sudden demand spikes hyperscalers are creating.

When Microsoft, Amazon, and Google show up at utility offices requesting hundreds of megawatts on three-year timelines, the answer is often no. Berkeley Lab found that more than 70 percent of grid interconnection requests in the United States are ultimately withdrawn because the grid simply cannot accommodate them. Kevin O'Leary, the Shark Tank investor and longtime infrastructure backer, has estimated that 50 percent of the data centers currently planned across the United States will never get built, not for lack of money or demand, but because the grid cannot deliver the power.

Solar and wind are intermittent sources. They cannot guarantee the 24/7 baseload power that AI training clusters require. Nuclear power, by contrast, runs continuously and produces zero carbon emissions. This fundamental difference explains why the smart money is moving.

What Nuclear Deals Are Tech Companies Actually Signing?

The commitments are substantial and long-term. Microsoft signed a 20-year deal to restart the Three Mile Island nuclear plant, a facility that has been offline since 2019, specifically to feed its AI ambitions. Amazon paid $650 million for a data center campus directly co-located with the Susquehanna nuclear station in Pennsylvania. Google announced agreements with Kairos Power for small modular reactors, which are smaller, more flexible nuclear facilities than traditional plants. Meta has been pursuing similar nuclear partnerships and recently issued a request for proposals seeking up to 4 gigawatts of new nuclear capacity.

These are not speculative moves. They represent billions of dollars committed over decades, signaling that power is the binding constraint on AI expansion. If Microsoft is willing to restart a dormant nuclear plant to secure power, the scarcity and value of reliable electricity are real.

How Is This Reshaping Investment Markets?

The shift from solar to nuclear is showing up in fund performance. The VanEck Uranium and Nuclear ETF (NLR) is up 22.94 percent over the past year and 158.7 percent over five years, with shares closing at $125.64 as of mid-June 2026. By contrast, the Invesco Solar ETF (TAN), which holds panel manufacturers, inverter makers, and residential installers, has underperformed because hyperscalers are not signing solar agreements. TAN's holdings are not the counterparties on the nuclear deals reshaping the energy sector.

The mechanism matters. When Constellation Energy signs a power purchase agreement with a hyperscaler, the revenue flows to a company that NLR holds. When BWX Technologies wins reactor work tied to federal nuclear expansion or military microreactor programs, the same shareholders benefit. TAN holders capture none of that revenue. The performance gap reflects which fund owns the companies actually selling the megawatts.

For investors seeking concentrated uranium exposure, the Sprott Uranium Miners ETF (URNM) is up 38.17 percent over the past year. The Range Nuclear Renaissance Index ETF (NUKZ), which tilts toward reactor technology and small modular reactor developers, is up 32.39 percent over the past year. NLR remains the steadier, more diversified vehicle because it spans utilities, miners, and engineering firms rather than concentrating in a single segment.

What Are the Key Risks and Timeline Challenges?

Nuclear is sentiment-driven and volatile. NLR is down 2.83 percent over the past month, even after the multi-year run. One analyst warned in December 2025 that the rally had outrun fundamentals and that a correction was likely in 2026. The fund holds pre-revenue small modular reactor developers, which adds binary risk. Policy and permitting timelines could slip, and projects like Google's Kairos partnership are targeted for 2030, not next quarter.

NuScale Power, the only U.S. nuclear company with an NRC-approved small modular reactor design, exemplifies this timeline challenge. The company is expected to open its first commercial reactor in 2033, a seven-year wait. NuScale shares trade about 75 percent lower than their all-time highs, as investors have been more cautious on small modular reactors in 2026 than in the prior year. For patient investors, that pullback could create an opportunity if NuScale can successfully bring its first commercial small modular reactor online by 2033.

How to Position for the Nuclear-Powered AI Future

• Diversify across the nuclear supply chain: Rather than betting on a single small modular reactor developer, consider funds like NLR that hold utilities, miners, and engineering firms. This spreads risk across companies at different stages of the nuclear expansion.
• Understand the timeline mismatch: Most small modular reactor projects are targeted for 2030 or later, while AI data center demand is accelerating now. Companies with existing nuclear capacity or near-term projects may outperform longer-dated bets.
• Monitor grid interconnection bottlenecks: The real constraint is not reactor availability but grid capacity. Companies positioned to solve transmission and distribution challenges may capture outsized returns as utilities struggle to accommodate hyperscaler demand.
• Watch for policy shifts: Nuclear expansion depends on regulatory approval and government support. Changes in permitting timelines or federal incentives could accelerate or delay the thesis.

The energy crisis driving this shift is real and structural. AI workloads are not going to become less power-hungry. Hyperscalers have already signaled they will wait years and spend billions to lock in nuclear capacity rather than rely on intermittent renewables. That commitment is reshaping which companies will profit from the AI boom and which energy sources will power the next decade of computing.