The Peter Lynch Method Says the Death of AI May Be In

Legendary investor Peter Lynch used to say that if you paid attention to your daily life, you would see things that you could use to invest successfully, way before Wall Street figured it out.

The canonical example back in the 1980s was that if a local plant added a third shift, you would buy the stock.

I use Replit for all my AI agent building and received this email from them yesterday explaining how they have cut two “shifts”:

Let me do the price change math for you:

• Autoscale compute: -81%;

• Autoscale requests: -66%;

• Outbound data transfer: -50%; and,

• App storage: -50%.

I’m no AI expert like Leopold Aschenbrenner, but Gott in Himmel!, from an industry perspective, we got a problem here.

I use AI a lot, every day actually. And I’m using it the way you’re supposed to - not as a replacement for Google search, getting customized answers instead of 9,000,000 hits returned in 0.3 seconds, but building machines to help me do my work.

I’ve built over 20 apps and use half of them every day.

This is not me flexing, just me saying that I’m a relatively heavy user of AI, for a small business.

Check this out:

Image 2, above, shows that Bantam has spent $189.47 on AI in the past month.

Now that’s going to be cut significantly.

A tremendous amount of debt and equity has been issued on the back of not a lot of revenue… revenue which now appears to be significantly declining because of market pricing pressure.

Buffett once said that in a commodity business, you’re only as smart as your dumbest competitor. Well, compute is a commodity. You want proof? Replit just voluntarily repriced compute down 81 percent. That’s commodity pricing. (Rolex never cut the price of a Submariner from $10,000 to $2,000 overnight.)

China probably plays the role of “dumbest competitor” in AI, namely because they will illegally dump compute on the market at five percent the cost of U.S. compute. But the commodity pricing is coming even if the U.S. government bans the use of Chinese compute.

In no universe does this end well for AI stocks. Most will go BK and the winners will be down 70+ percent.