AI's Energy Crisis Is the Real Infrastructure Story
The AI arms race isn't just about GPUs — it's about whether we can keep the lights on.
The biggest bottleneck in AI right now isn't models. It's megawatts.
The DOE just committed $4.42 billion to restart three shuttered nuclear plants — including Three Mile Island — specifically to power AI data centers.
Let that sink in. We're reopening nuclear facilities to keep up with compute demand.
This is the infrastructure story most people are sleeping on. Everyone's watching the GPU arms race (Meta's 'Prometheus' supercluster targeting 1 million GPUs at a single site is wild). But none of that matters if you can't reliably power the thing.
Energy is becoming a strategic moat. The companies and hyperscalers that lock in carbon-free capacity now will have a serious competitive advantage as regulatory pressure around AI's environmental impact intensifies.
The next AI infrastructure play isn't software. It's kilowatts.
Founders and investors — if you're not thinking about energy constraints in your AI buildout, you're missing a critical variable.
— Leo Giel, Founder & CEO @ Lemonhead.ai
