Five Billion Dollars Says AI Isn't a Fad

AMD committed up to five billion dollars in equity to Anthropic in August. Nvidia reportedly backed Safe Superintelligence with another five billion. A consortium proposed a hundred-billion-dollar AI computing campus in Kentucky.

When people ask me if AI is a bubble, I tell them to ignore the apps and follow the concrete.

Apps come and go. Startups pivot. Features get absorbed into platforms. But nobody pours a hundred billion dollars into physical infrastructure on a hunch. You build a 2-gigawatt data center campus because your demand projections for the next decade say you'll need it. That's not speculation. That's industrial planning at a scale we haven't seen since the interstate highway system.

I work in cybersecurity consulting. I've watched technology cycles for thirty years. The internet bubble burst, but the fiber optic cables stayed in the ground. Cloud computing went through a hype cycle, but the data centers got built anyway. The pattern is always the same: the narrative overshoots, the stock prices correct, and the infrastructure that got built during the boom becomes the foundation for everything that follows.

That's where we are with AI. The apps might shuffle. The platforms will consolidate. Some of today's darlings will disappear. But the silicon, the power plants, the cooling systems, the chip foundries — those don't go away. And the capabilities they enable only get cheaper and more accessible with time.

For builders, the signal is clear: what you learn now about working with AI will compound. The tools will change, the models will improve, but the fundamental skill of knowing how to create with AI assistance — that's as durable as learning to use a computer in 1995. The specific applications shift. The underlying capability only grows.

Five billion dollars isn't a bet on hype. It's a bet on infrastructure. And infrastructure bets have a very good track record.