X-energy, the small modular reactor startup Amazon put money into eighteen months ago, just raised $1.02 billion in its IPO. Shares priced at $23, above the $16 to $19 range the bankers were shopping. The stock opened Friday on the Nasdaq under ticker XE and jumped 31%, closing the company at an $11.9 billion valuation.

This is the first meaningful nuclear startup IPO of the AI infrastructure era. It will not be the last.

The Math Wall Street Bought

X-energy sold 44.3 million shares at $23, raising roughly $1.02 billion, with an upsized offering that exceeded the $800 million target the company filed for ten days ago. Demand was strong enough that lead bankers lifted both the share count and the price.

What investors are actually buying: a company with no operating reactor and a multi-year runway to first commercial kilowatt. What they are pricing in is the queue.

X-energy has a signed agreement with Dow to provide heat and power to a Texas chemical plant. It has a separate agreement with Amazon to sell as much as 5 gigawatts of nuclear power by 2039. Amazon's Climate Pledge Fund led the Series C-1 round last year and stuck around through the IPO. That is the anchor tenant nuclear needed to become investable again.

The Reactor Itself

The Xe-100 is a high-temperature gas-cooled reactor rated at 80 megawatts of electric output per unit. Helium coolant flows over billiard-ball-sized graphite pebbles packed with TRISO fuel, uranium kernels wrapped in layers of carbon and silicon.

TRISO is not new. It was developed decades ago specifically to withstand higher temperatures than conventional fuel, reducing the probability of a meltdown. What is new is the commercial willingness to deploy it at scale. That willingness comes directly from the demand curve AI is bending.

Why Nuclear, Why Now

The AI build-out has broken the assumptions under every hyperscaler's power strategy. Grid interconnects that used to take eighteen months now quote five years. Gas turbines are backordered into 2028. Wind and solar are cheap but intermittent, and the data centers training frontier models need 24/7 baseload.

Nuclear solves the baseload problem. Small modular reactors solve the siting problem. Amazon, Google, Microsoft, and Meta have all signed nuclear deals in the last eighteen months. Google is restarting the 615-megawatt Duane Arnold facility with NextEra. Microsoft is reviving Three Mile Island Unit 1 with Constellation. Amazon backed X-energy and separately bought the Cumulus nuclear-adjacent data center campus.

X-energy's IPO is the public-market pricing of that entire thesis.

The Risk Ledger

Three things a bull case has to clear. First, regulatory timelines. The Nuclear Regulatory Commission has not licensed a new small modular reactor design under the Part 53 framework at commercial scale, and the clock on X-energy's first plant is measured in years, not quarters. Second, construction cost overruns. The last American nuclear new-build, Vogtle Units 3 and 4, ran seven years late and $17 billion over budget. Third, AI demand durability. If training compute demand plateaus or shifts to inference-dominated workloads with different power profiles, the 2039 off-take curve starts to wobble.

None of those risks stopped the book. The offering was oversubscribed. The stock popped 31%. Wall Street is pricing in the AI power crunch, and it is paying a premium for a startup that might, in a decade, actually ship the baseload.

Sources: Reuters, TechCrunch, Renaissance Capital, X-energy press release (April 24, 2026).