Key insights
- The article argues that while AI technology is real and transformative, the current investment cycle exhibits bubble-like characteristics. Valuations are being priced as if future success is guaranteed, with many companies leveraging the 'AI' label to justify high prices. This mirrors past bubbles where overvaluation, not the underlying technology's lack of value, caused issues. The author questions what metrics (revenue, profit, productivity) would truly validate AI's economic impact and signal it's not in a bubble, suggesting a potential correction if these aren't realized.
I don’t understand why every AI debate turns into two extreme camps.
One side acts like AI is fake and useless.
The other side acts like questioning AI valuations means you’re some anti-technology caveman who doesn’t understand the future.
The more realistic answer is probably this:
AI is real, useful, and will change a lot of industries.
But the current AI investment cycle still looks like a bubble.
Not because the technology is fake, but because the economics are being priced like everything will go perfectly. Every startup is suddenly “AI-native.” Every SaaS company has an AI assistant. Every CEO is talking about agents. Every data center is “strategic infrastructure.” Every layoff is “AI transformation.” Every chip stock is treated like it deserves a permanent premium.
That is usually how bubbles work.
The internet was real in 1999. Railroads were real. Housing was real. Crypto had real use cases. The problem was never that the underlying technology or asset had zero value. The problem was that people overbuilt, overpaid, overhyped, and assumed the future would arrive in a straight line.
That feels like where we are with AI now.
The winners may become massive. But a lot of companies using the AI label are probably just borrowing credibility from the future to justify today’s valuation.
My question is this:
What would actually prove that AI is not in a bubble?
Is it revenue? Profit? Productivity gains? Lower costs? Real enterprise adoption? Or are we all just assuming the economics will eventually catch up because the technology feels important?