For years, nuclear sat in the corner at every clean energy conversation, awkward, expensive, carrying decades of political baggage. Investors avoided it. Sustainability teams sidestepped it. And the industry slowly faded into the background while solar and wind collected all the headlines.
Then AI showed up and changed everything.
In October 2023, X-Energy attempted a SPAC merger valued at around $1 billion. It collapsed. Public markets were "persistently volatile," the company said, and Ares Acquisition Corporation simply liquidated. In less than two years, Amazon-backed X-Energy walked back into public markets and left with over a billion dollars, and investors still wanted more.
The Problem Nobody Wanted to Say Out Loud
The core issue is simple but uncomfortable: the AI boom has a power problem.
The International Energy Agency projects that global data center electricity consumption could double by 2030, and that number alone does not capture the real challenge. The issue is not just volume; it is consistency. AI training runs and inference workloads require constant, at-scale power without interruption. Solar panels go dark at night. Wind turbines depend on the weather. And neither can carry a GPU cluster through a long weekend without a battery backup system that would cost a fortune to build.
More than half of all U.S. AI data centers planned for 2026 have been delayed or canceled entirely due to interconnection backlogs, with grid operators warning of multi-year wait times in markets such as Northern Virginia, Phoenix, and Dallas-Fort Worth. PJM Interconnection alone is sitting on a queue of over 286 gigawatts of pending generation requests.
Hyperscalers have responded the only way they know how, by throwing money at the problem. Microsoft brought Three Mile Island's Unit 1 reactor back online through a deal with Constellation Energy, securing 835 MW of carbon-free power through 2054. Google contracted Kairos Power for up to seven SMRs totaling 500 MW. Meta put out a request for proposals for up to 4 gigawatts of new nuclear capacity. The pattern is clear. Nuclear is no longer a climate statement. It is infrastructure procurement.
What Amazon-backed X-Energy Is Actually Building
Amazon-backed X-Energy's core product, the Xe-100, is not the image that comes to mind when most people think of nuclear power. Each module delivers 80 megawatts of electricity or 200 megawatts of thermal energy, and the reactor uses helium cooling with TRISO fuel particles, a design that can withstand extreme temperatures without any meltdown risk. You do not need to understand engineering to appreciate what that means operationally: it removes the catastrophic failure scenario that has haunted nuclear's public reputation for fifty years.
The reactor's compact footprint is another advantage. SMRs can be deployed closer to demand centers than conventional gigawatt-scale plants, which matters enormously for data center operators who want to reduce transmission loss and avoid depending on an already strained grid.
Modules can be deployed in groups of four, producing 320 megawatts, and scaled up to 12 units, reaching 960 megawatts. Beyond electricity, the high operating temperatures also make the Xe-100 suitable for industrial steam, hydrogen production, and desalination. This versatility is part of why the customer list stretches well beyond tech companies.
Amazon's Bet Goes Beyond Climate Commitments
Amazon led X-Energy's $500 million Series C-1 funding round and signed a binding agreement to purchase up to five gigawatts of nuclear power by 2039. The first project, the Cascade Advanced Energy Facility in Washington state, begins as a four-unit, 320-megawatt installation developed alongside public utility Energy Northwest, with room to expand to 12 units and 960 megawatts.
Amazon's nuclear strategy does not stop there. The company acquired Talen Energy's data center campus adjacent to the Susquehanna nuclear plant in Pennsylvania for $650 million, secured a 1,920-megawatt power purchase agreement through 2042, and is exploring a 300-megawatt SMR project with Dominion Energy in Virginia.
This is not a sustainability pledge dressed in press release language. It is vertical integration, the same logic Amazon applies to logistics and chip manufacturing, now pointed squarely at electricity supply. And Amazon-backed X-Energy is the clearest expression of that strategy made public.
The Bigger Picture
The IEA reported this week that AI data center electricity consumption is expected to triple by 2030, and that the pipeline of conditional offtake agreements between data center operators and SMR projects has nearly doubled from 25 gigawatts at the end of 2024 to 45 gigawatts today. The capital is not waiting for commercial reactors to start operating. It is locking in positions before the supply crunch becomes impossible to ignore.
Amazon-backed X-Energy's total customer pipeline now exceeds 11 gigawatts, anchored by Amazon, Dow Chemical, Centrica, and a letter of intent with Talen Energy for gigawatt-scale deployment in the PJM market. Dow's Seadrift project will replace ageing fossil fuel infrastructure with four Xe-100 units delivering both electricity and industrial steam, while Centrica signed a six-gigawatt joint development agreement for the UK's first advanced reactor fleet.
The companies that will lead AI at scale over the next decade will not just be the ones with the sharpest models. They will be the ones who figured out the power problem early. Amazon-backed X-Energy's IPO is the market's first public acknowledgment that this race is already underway and that nuclear has earned a seat at the table it was never supposed to have again.
JMC continues to track how capital, technology, and infrastructure are reshaping global business strategy.
