What if the A.I. Economy Crashes?
I keep coming back to the idea that bubbles are only obvious in hindsight, and the AI boom feels like the fastest one we've ever had the chance to watch in real time. Azhar and Kedrosky are asking the right question, but the answer probably depends on which layer of the stack you're standing on. The chip makers and data center landlords are booking real revenue. The startups burning cash to build "AI for X" are a different story entirely.
What I'd want to know is how much of the current valuation depends on a handful of customers. If a few hyperscalers are buying each other's compute, we're really just watching them trade with themselves. The 2001 dot-com crash didn't kill the internet, it just killed the people who confused a network with a business model. Same could happen here, and the surviving infrastructure might actually be more useful afterward.
I don't have a strong prediction, but I'd bet the crash, if it comes, won't look like a single event. It'll be a slow realization that some problems don't need a trillion-parameter model, and that a lot of the "intelligence" being sold is just pattern matching with better marketing. The tech will survive. The hype tax, though, is going to be painful.