Amazon and X-energy have signed another deal to deploy nuclear power to the grid. This one funds 12 small nuclear reactors at a former coal plant in Indiana, with total output of nearly 1 gigawatt. It's a partnership announcement, not a crypto story — but it's the kind of infrastructure bet that eventually shapes what compute costs, and who can afford to tokenize it.
The coal-to-nuclear detail most coverage will skip
The site matters. A retired Indiana coal plant comes with something a greenfield project doesn't: a live grid interconnection. Reusing it can shave years off deployment and avoid the interconnection queues that have become the real bottleneck for new generation. X-energy hasn't published a construction timeline, and Amazon hasn't said how much it's putting in. What's clear is that the company isn't shopping for reactor designs — it's shopping for grid access.
📊 Market Data Snapshot
That's the template. Hyperscalers are racing to lock in long-duration, carbon-free baseload power for AI data centers, and the firms that get there first will set the marginal cost of compute for everyone behind them.
Twelve reactors is the actual tell
Nearly 1 gigawatt across 12 units works out to roughly 80 megawatts each. These aren't the 300-megawatt-plus designs the SMR crowd usually pitches. Smaller units mean factory production rather than bespoke construction, which is faster — if the Nuclear Regulatory Commission agrees to license a fleet instead of 12 individual builds. That regulatory question is the one to watch. Get a favorable precedent and delivery timelines compress. Don't, and the math gets ugly.
X-energy is the quiet winner here. A second Amazon anchor order de-risks its fuel supply chain and makes it the most plausible SMR listing candidate going into next year. The SMR race is won on order books and fuel fabrication, not reactor blueprints.
What this has to do with crypto
Nothing directly. Bitcoin is drifting around $81,400 with high dominance, and this headline won't move it. But the longer arc — cheap baseload power feeding AI compute — is the substrate under every tokenized compute and DePIN narrative in the market.
The catch: most decentralized compute protocols rent capacity from hyperscalers. They don't own the power. Until a protocol signs a direct PPA or co-locates behind the meter at one of these sites, the 'energy is the new moat' thesis stays theoretical for crypto holders. Watch for DePIN-energy partnerships, not sympathy plays on nuclear headlines.
Next on the calendar
X-energy's filings with the NRC on fleet licensing, and any signal from Amazon on how much of the 1 gigawatt it plans to take itself. Neither has a public date yet. Until they do, this is a backdrop story — one more hyperscaler nuclear commitment that validates the long-term compute thesis without giving traders anything to act on today.
