Yes, AI is a bubble.

Thursday 16, 2026

It started out as optimism. The rate of apparent improvement looked impressive. Then money started to flow. And that’s when the cheap hookers showed up with the cocaine and all went to shit. Nerdy men tend to lose the ability to think under the extreme stress of having too much money all of sudden.

Most kids, if they have never seen a soccer ball won’t be able to kick one the first time. But, give them a few weeks and nearly any child will advance at a surprising rate of skill at soccer. That doesn’t mean you, as a parent, should buy a knew house because your kid is going to play professionally.

Unless you, yourself played soccer and have coached soccer, you likely can not gage just how fast your child’s improvement really is. Just because someone else who gets paid to teach kids how to play soccer says your kid is the next Messi, that doesn’t mean they are either.

I am sorry to all the computer programmers out there. Computer programming is no more difficult than learning French. And four year old French kids are wizzes at French. The rate at which languages can be learned is not reflective of whether or not speaking French will make you a billionaire. The problem is as adults we don’t remember how easy it was to learn as a kid. This is because of developmental amnesia.

Shaun and I have been using AI since before OpenAI was founded. Back when it was called machine learning. I was playing with neural networks in the 1990s. Shaun got his graduate degree from Georgia Tech in the subject.

But, you don’t have to believe me.

Just ask this AI generations parent, Yann LeCun. He knows none of today’s AI companies are going pro. He is so confident this current generation of AI is not going anywhere he started a company to try and make the next generation.

Spoiler alert. Even if he succeeds unless patents a ternary architecture or similar that next generation will also be a bubble.

Let’s look at a pretty comprehensive list of “AI” i.e machine learning technologies. Everything from speech recognition to routing is freely available as open source.

Granted spam detection is incredibly helpful. No one likes robocalls. But, does your phone still ring from scam calls? Did this AI “tech” actually solve the problem? Or just slightly outpace it?

And do you explicitly pay any of this? Or is just included in whatever device you buy and you can choose to use it or not. Name the trillion dollar company that allows you to talk to your TV remote.

I will wait.

The fact is, everyone knows AI is a bubble. Everyone knows hyperscalers are wasting money. Everyone knows the SPVs will blow up. Most people in the know don’t care because they get paid not to.

What people don’t yet know is how to extricate themselves and their clients without looking foolish. This is the hard part.

Investment managers can’t say sell. If the market runs another 50% they look stupid. Burry manages his own money and has a herd. He can scream fire anytime he wants and push his trades into the money.

Goldman Sacs can also do this… but most of their money is actually their clients money.

Managers also can’t be seen selling while leaving their clients in tech stocks. This looks dishonest.

What people are slowing beginning to realize is that when this market ends there is no discernible bottom. Look at Micron. At this scale it doesn’t look that bad, right? The average per share cost is -60% below the current share price. A “cheap” entry is about 75% below. That sort of collapse renders most people numb.

Maybe this scale makes the unusual nature of this chart more apparent. That is a right angle almost.

And notice those purple dots. Those are liquidity alerts. I call them BOB because I don’t have a better name. Notice there has not been one during this run? Notice the red dots. That is funds dumping shares.

On this time line the per share cost is -85+% from here. Based on all the institutional buying before the move that is likely more accurate.

Psychologically there is a threshold where the loss seems total (even though it is not) and people just close the brokerage window and stop looking.

For years.

Or worse. Never invest again.

This is where we might end up. This is reality.

Managers and banks have exposed their clients to private and public markets in a way that can not be unwound. Let’s say you have a client that is up 300%. If they sell they will face capital gains.

Then what?

Where would you tell them to put their money? In private equity? In real estate? What part of today’s asset landscape looks cheap to you? What if your clients miss out on the last squeeze? Will they fire you and find another manager to put them back in the market?

Isaac Newton famously sold the South Sea Bubble and then watched the market run over the next few months. So he bought back. And proved why you shouldn’t poke an ice pick into your sinus to see what happens.

I said before I was close to wearing a cape. Well, I am pretty much there now.

This is the last earnings cycle I will ride the market. From here any speculative purchases will have to have a specific temporally focussed narrative for me make a trade.

And, I will continue to ride/flip QQQ 0.00%↑ PUTS until this all implodes. If takes a decade I take little to no risk and honestly do you really think the market is going up?

So the odds are still good for PUTS. Unless everyone is short and there is a magic squeeze event… which seems unlikely. Because of those hookers and blow.

— AJ