When do you think the AI bubble will burst?

[Dani]

I was reading that OpenAI might file for an IPO sometime in 2026. In the meantime, I have a big stake in Nvidia and some other AI-based companies. When do you think the AI bubble will burst in the stock market?

[rproffitt]

Not yet. If you are asking but not asking about the last few days movement, that was Softbank.

"Nothing to see here. Move along."

[Dani]

I vaguely recall reading the other day that Softbank sold their entire stake in Nvidia so that they can instead purchase a stake in OpenAI (which is not yet public). That doesn't say anything about the AI bubble bursting, since even they are keeping their money in AI.

My question is more of when you think AI will become so ubiquitous that AI companies will stop with the 100% year over year growth and start to flat line or self-correct.

[rproffitt]

Here's the current prediction. 99% of AI companies will be dead by EOY 2026.
Mostly because most are using "wrappers" and don't own any of the underlying tech.

While the magnificent few will do well, you have to disregard the dumpster fires to achieve your 100% YOY growth.

Even with that, that bubble didn't burst. Which is your question.

[Dani]

Even with that, that bubble didn't burst. Which is your question.

I know the bubble didn't burst. That's obvious. That wasn't my question. My question was when in the future you think eventually the bubble will burst?

[rproffitt]

There are big and little bubbles. I'm with many others that some thousand little bubbles will burst and the big bubble will be fine for many years. That is, NVDA, MSFT, META, Goog and whatever OpeneAI is going to roll out as.

The .com bubble lasted about 5 years. So for now I'll go with 1 to 5 years. Why 1 year? Mike Johnson said some things but hey, you may think it political and yet all this is about US and China politics and golden ballrooms.

[jkon]

I don't think that there is a single AI bubble. But I strongly believe that there is a bubble in the valuation of companies that have to do with LLMs, and that will burst not with a big splash but like a domino of falling towers, some will stand but will be lower.

OpenAI and Anthropic keep lying about what their models can do and their impact now, in 6 months, or in a year. They have done that for over two years now, as a promotion, and from my standpoint, they are comical. Instead of facing consequences for that, their valuation as companies keeps going up. That, of course, is not sustainable. What we know about how ChatGPT-6 is being trained reveals that they are really desperate to present something different soon, but I think they already have run out of time, and probably sometime in 2026, we will see the first falls.

Although NVIDIA does what it can to inflate this hype (e.g., ''coding as a career might already be dead due to AI'' - Jensen Huang 2024), it produces something real that there will always be a need for. So even if its valuation might be affected, I don't think it will fall. On the other hand, Microsoft and Google have more diverse revenue streams, so although in some cycles they inflate the LLM hype, in others they retreat to reality.

I can't predict how much the fall of OpenAI will affect the real economy, if the USA will officially enter a recession due to that domino effect, or how this will translate for the rest of the world. In the long run, AI will transform almost every aspect of the economy, but not in the way that is presented now, making the workforce obsolete, but instead enhancing human productivity. There might be some fringe sectors where humans will be replaced at large, but for the most part, AI will be just an assistant.

[Dani]

"'coding as a career might already be dead due to AI'' - Jensen Huang 2024

That is the very real sentiment around silicon valley right now. I think a lot of individual contributors at major tech companies are all waiting with bated breath to see how things play out over the next few years. Especially with housing prices out of control here and mortgage rates pretty high as well.

[SCBWV]

But hey, the admin is pushing 50-year mortgages. 😡

[Dani]

But hey, the admin is pushing 50-year mortgages. 😡

With the average first-time homebuyer currently 40 years old, they're going to die before their mortgage is paid off.

[rproffitt]

When an entire industry is built upon massive theft and IP rights infringement, is involved in an ongoing series of legal battles because of this and has the double-edged Damocles sword of legal fines and crippling new regulations hanging over its head, can't even deliver something that works as advertised and hyped, and has a hard time convincing everyone to adopt its flawed product, invest as you would at a gambling table.

Then we think about how NVDA isn't the only AI chip maker.

The bubbles will pop. Not a matter of if but when and which bubble you are asking about.

[Salem]

https://www.nbcnews.com/business/economy/openai-nvidia-amd-deals-risks-rcna234806
See "AI funding’s circular web"

A game of musical chairs - who's holding the timebomb when the music stops?

[Dani]

When an entire industry is built upon massive theft and IP rights infringement, is involved in an ongoing series of legal battles because of this and has the double-edged Damocles sword of legal fines and crippling new regulations hanging over its head, can't even deliver something that works as advertised and hyped, and has a hard time convincing everyone to adopt its flawed product, invest as you would at a gambling table.

I'm not so sure I can agree with all of that. As a digital publisher, I do feel as if AI is built upon massive theft and IP rights infringement. I can get behind that. However, I don't think that LLM "has a hard time convincing everyone to adopt its flawed product." Specifically, I think that LLM is revolutionary and groundbreaking in its current state, and is being adopted left and right by just about every industry in tech. If that wasn't the case, we wouldn't be in as big of a bubble as we currently are. However, i's still in its infancy, and it's constantly improving, mostly because so much money is being thrown its way, and I think it's too early to tell where the dust will settle after the shake-up.

I, for one, own stock in all the big AI companies. However, I've owned them all for decades. I bought Nvidia at $0.70 back in 2007. I've owned Microsoft also since 2007, Oracle since 2009, and Google since 2013. I have enough faith in these companies that they were here before LLM and they'll be here after the AI bubble bursts. Sure, AI is fueling their current growth spurt, but I believe their business models are strong enough to make it through the AI wave.

[Dani]

Also, I used AI and LLM interchangeably in my above post. Please ignore that.

[Pebble94464]

Here's my thinking...

We usually see a Santa Rally towards the end of the year, when fund managers like to show something on their books. Given how overpriced tech stocks are at the moment, and the fear of AI bubble bursting, I suspect we're going to see some profit taking and rebalancing that will help drive up commodities and less volatile non-AI stocks.

There's plenty of chatter about the AI bubble, with many commentators saying a crash is now inevitable, but none of them will tell you exactly when it will happen. No one, except me - It'll happen the moment go back into the market. 100% guarantee! :-)

(Disclaimer: I have no idea what I'm talking about)

[rproffitt]

I wish you would have defined which bubble. Maybe it's just NVDA and the AI-based companies.

1. NVDA. Probably fine for years.
2. GOOG. Looking at doubling AI compute power every 6 months. The problem here is a lack of electric power generation. For now that will be its limiter.
3. Smaller AI companies. Let's put those in the less than billion dollar a year category. Those look to be out of business every year. Few are actually going to survive year over year. So while a sub-percent point will make it past a year that is a burst or bust in my view.
4. Consolidation vs. bust. The better AI companies will likely be acquired.

[wwwalker]

2 years so November 2027. Compare tech wreck of 2001.

[Dani]

Nvidia doesn’t seem like it’s doing that great lately. Just saying.

And yes, I know this is on the heels of Meta announcing they will use Google processors instead of Nvidia.

[David_326]

The AI boom feels unstoppable, but like any hype, it might see a slowdown or “bubble” burst someday. Meanwhile, we can enjoy the moment and share excitement with those around us.

[rproffitt]

Remember that the "bubble" is bubbles. Maybe you are only concerned with NVDA but the AI BUBBLE collection is going to keep going for years.

Additional inflators and constraints are:

1. Project Genesis. I'll slide by detailing the scale and which country this is being saddled with.
2. GOOG doubling AI compute centers every 6 months. As to your NVDA mention, guess whose hardware they are using.
3. While Dani's concern seems to be centered on NVDA, AI in China appears to be building with design and chips they make there.
4. NVIDIA's trailing-twelve-months (TTM) P/E ratio is approximately 44.39 as of November 2025. The S&P is about 31.
5. The rising AI tide is not lifting all boats. This statement is worth it's own discussion.

The WHEN of a NVDA stock bubble burst will be when you see about a 10% stock price drop which will spook the herd and cause a secondary dip or crash if you will.

-> The top question needs a lot of refinement.

[rproffitt]

Update needed about GOOG's TPUs. Work in progress, GOOG is a generation behind NVDA so let's see if they catch up.

[Dani]

Maybe you are only concerned with NVDA

I’m mostly concerned with the Magnificent 7.

As to your NVDA mention, guess whose hardware they are using.

Their own!

The WHEN of a NVDA stock bubble burst will be when you see about a 10% stock price drop

We’re pretty much there. Nvidia is down about 8% this week.

The top question needs a lot of refinement

How so?

[Dani]

Update needed about GOOG's TPUs. Work in progress, GOOG is a generation behind NVDA so let's see if they catch up.

Even if they don’t, they are cheaper than Nvidia and might be “good enough” at a lower price point. Perhaps why Meta is ditching Nvidia for them.

[rproffitt]

New data/ai centers by GOOG use and this is not backed by fact is the new GOOG TPUs. This alone isn't a bubble pop.

Project Genesis alone needs a discussion. Who wins there?

PS. Paul Krugman says the fed can't rescue AI. But they will try with Project Genesis so the pop is delayed until about November 2026 or later.

[Dani]

Project Genesis alone needs a discussion. Who wins there?

What does terraforming have to do with the price of beans?