I’m guessing it’s hard to go short OpenAI without also going short a bunch of other companies riding the AI wave that aren’t led by Sam Altman?

Would love to hear how that could work.

Shorting a security means risking exponential losses if the stock you're shorting continues to increase in value. As the saying goes: the market can stay irrational longer than you can stay solvent.

Hmm. Maybe he might do... a bet!

And then maybe he might ... change the bet! when he was about to lose?

Maybe!

Who's to say, really? Certainly not me! I'm just a neural network!

Just short Nvidia. If the thing goes bang then that’ll be one of the big losers.

"Just short Nvidia"

Is this financial advice? :-)

If so, keep in mind that it's contingent advice. The question was how to profit from predicting an AI bubble popping [or something along those lines]. The answer is shorting Nvideo (assuming your prediction also includes a timeline).

It's always a way to lose a massive amount of money if you're wrong, so the advice is also contingent on confidence level.

I'm not trying to be facetious here, but I think it's very naive to assume you get to "profit from predicting an AI bubble." In theory, maybe, but in reality you will lose money. Shorting is never the solution... it's a very niche tool for very niche group of investors.

When retail guys talk shorting, it's very hard to take them seriously.

Retail guys generally think they can time the market based on vibes, rather than specific insider-y info. But if you (retail investor or not) have that specific insider-y info--something resulting in justified, high probability, time-bounded knowledge about a future change, shorting can be the rational decision.

It was more “if you’re going to short then short nvidia seeing as it’s not possible to short companies such as OpenAI which are private”.

What's your complaint about this article? I wish there were a way to incentivize comments to put effort into specific criticism, before seeking attention for their ideas.

I'm not aiming this at GP specifically, but there seems to be a culture around gen AI that the burden of proof is on sceptics, not the people claiming we're about to invent God

Gary is an insufferable blowhard but he's had skin the AI game for awhile. I believe he sold an AI startup to Uber back in the 2010s.

Many of his criticisms of OpenAI and LLMs have been apt.

It's possible to notice a trend while still having the wit to realize you can't precisely time that trend well enough to profit from it. Another example might be, "Trump is increasingly old, feeble, and incapable of doing his job... but I'm not sure how that will translate into how long he's able to keep the job. It's possible that he could be a vegetable at some point and still POTUS."

Demanding that people gamble with their often limited finances to prove a point orthogonal to the one they're actually making feels disingenuous and dismissive to me.

It's not orthogonal. And you will find people will change their mind when forced to put a little money on the line.

"Team X is definitely winning, I'm certain". "So you'll offer me 1000-1 on the opponent?". "No". "6-1?", "No". They often realize they are about 65% certain at some point. And they often aren't being hyperbolic, they are just not thinking clearly.

There was a point after which the outcome of WWII was obvious and inevitable, but could you have timed the ending to within a week or two based on that knowledge? Should the inability to precisely time something imply that the arc of its future can’t be obvious?

Clearly not. Does the fact that few of us know the date of our death imply that we might live forever?

You've misunderstood the point. It's not about being exactly right. It's about thinking clearly about the probabilities of different events. In your case, you explicitly call out that it's difficult to know the exact timing - so if someone had offered you even money for some particular week, you'd say no. If they offered you 100-1 for said week, it's a great bet, even if you lose.