EVERYBODY KNOWS AI is a HUGE bubble.  Nobody wants to be the first one to leave the PAHHTY though.  Has Ray Dalio dumped his Tech stocks?  "Close" to bursting.  How Close is Close?  Who's gonna jump ship first?  What will be the spark to get the rats running?


RE

Nice summary of the many reasons the AI Bubble will pop.

Insufficient ROI for end users.

Currently in the land rush phase, where everyone is trying out the new tool. This distorts true demand.

Pushback over: job losses, bots, hacking, scams, NIMBY, teen health, deepfakes & privacy.

Most of the investment is by 5 huge Hyperscalers whose free cash flow has turned negative.

AI adoption is harder than the promoters think.

Token prices are dropping, squeezing frontier large language model (LLM) margins.

Much of this pressure is coming from Chinese open-source models.

The Hyperscalers are tired of giving all their profits to Nvidia and other suppliers and are increasingly using their own chips.

Over half of the Hyperscalers' backlog is from 2 unprofitable LLMs.

Massive circular financing.

Increasingly turning to exotic off-balance sheet financing.

Massive depreciation is coming that will reduce Hyperscaler earnings.

Oracle is in danger of bankruptcy if there is an AI slowdown.

Investors have repeatedly sold Hyperscaler stocks after announcements about increasing AI Capex.

Future AI will need less power as efficiencies are found.

Until then, there are huge power constraints.

Commoditization of LLMs.

Very high prices for memory are making AI even more costly.

Lack of use cases to justify all the investment.

Parallels to the internet bubble and other bubbles, such as railroads.

Tokenmaxxing, which has driven usage, is now declining.

Dead Internet Theory—bots, by some measures, are now over half of internet traffic.

Blown AI budgets—many companies are scaling back after quickly running out.

Increased regulation is coming—already happening in various states and in Europe.

RE