A skeptical take that AI is a financial bubble built on hype, bad for investors betting on it but likely to leave useful leftovers behind when it pops.

• The world has spent $1.4 trillion on AI, but the whole industry only brings in about $50 billion a year, making it the biggest money-loser in history.
• Each new model costs more and earns less, the opposite of the early internet, where every new user made the web more profitable.
• The real selling point to bosses isn't doing the work — it's scaring workers into accepting worse pay and conditions by threatening to replace them.
• When the bubble bursts it should leave cheap data centers, chips, and trained workers behind, much like the dot-com crash left cheap fiber and servers.
• AI is powerful word-guessing, not real reasoning, which is why it still makes basic mistakes and floods editors with unpublishable writing.

Outlook: The bubble is expected to pop and wipe out a lot of the money poured in, even if the underlying tech sticks around as a useful tool.