Remember November 1965? A single mis-set relay near Niagara Falls knocked out power to 30 million people across the Northeast. One tiny mistake cascaded into total darkness. The grid wasn’t broken—it just couldn’t handle the load.
Fast forward 61 years. We’re about to have the same problem, except the load isn’t air conditioners and elevators. It’s artificial intelligence, and nobody’s laughing about it anymore.
Here’s the thing: AI demand is growing so fast that serious engineers are now debating whether we need nuclear reactors or data centers floating in orbit. That’s not a joke. That’s an actual investment debate happening right now, and the outcome will determine which companies become trillion-dollar winners and which ones become cautionary tales.
**The Energy Crisis Nobody Wants to Talk About**
The hyperscalers—Amazon, Microsoft, Google, Meta—are already dumping tens of billions into data centers. But here’s the plot twist: the bottleneck isn’t chips anymore. It’s power. Raw, reliable, massive amounts of electricity.
Two competing visions are emerging. On one side, next-generation nuclear is getting fast-tracked. Companies like Oklo (OKLO), Constellation Energy (CEG), and the Global X Uranium ETF (URA) are positioning themselves as the baseload energy suppliers for AI clusters. On the other side? Orbital compute is moving from sci-fi to actual engineering.
Amazon’s acquisition of Globalstar (GSAT) is the real tell here. Sure, it looks like a connectivity play to compete with Starlink. But the deeper play is orbital compute. You can’t run data centers in space without a dense satellite constellation, licensed spectrum, and an anchor customer. Amazon just grabbed all three—right before SpaceX goes public and Elon starts launching his own orbital data centers.
**The Supply Chain That’s About to Explode**
If you want to play this infrastructure buildout, the picks-and-shovels approach is your friend. We’re talking compute owners (Amazon, Microsoft, Alphabet, Meta), rocket companies (Rocket Lab), orbital network operators (AST SpaceMobile, Globalstar, Iridium), space compute suppliers (GlobalFoundries, Nvidia), space power companies (Boeing, Lockheed Martin), and optical interconnect players (Coherent, Lumentum).
The commercial space era has officially begun. Both the nuclear path and the orbital path lead to massive buildouts. The suppliers win either way.
**The Real Risk Nobody’s Pricing In**
Here’s where it gets contrarian: the biggest threat to the AI trade isn’t valuations or competition. It’s populist backlash. Negative AI headlines are piling up. Bubble fears. ROI skepticism. And yeah, some of it’s valid. But here’s the thing—this backlash is actually a sign the buildout has reached critical mass. Every major tech wave, from railroads to the internet, faced populist headwinds once the investment got big enough.
The question is whether backlash turns into legislation that actually constrains AI investment. My read? The fundamentals are too strong, the capital commitments too deep, and the competitive dynamics too urgent for politics to slow this down materially.
**The Bottom Line**
TSMC just reported 30%-plus revenue growth and maxed out capex at $56 billion. When the company that physically makes every AI chip tells you demand is “extremely robust,” you listen. The ecosystem is on fire.
The next 12 months belong to patience and selectivity. But the long-term thesis? It’s never been stronger.
