The AI boom is more overhyped than the 1990s dot-com bubble, says top economist

[Daniel Sims]

The big picture: As tech giants pour more money into AI, some warn that a bubble may be forming. Drawing comparisons to the dot-com crash that wiped out trillions at the turn of the millennium, analysts caution that today's market has become too reliant on still-unproven AI investments.

Torsten Slok, chief economist at Apollo Global Management, recently argued that the stock market currently overvalues a handful of tech giants – including Nvidia and Microsoft – even more than it overvalued early internet companies on the eve of the 2000 dot-com crash. The warning suggests history could soon repeat itself, with the buzzword "dot-com" replaced by "AI."

In the late 1990s, numerous companies attracted venture capital in hopes of profiting from the internet's growing popularity, and the stock market vastly overvalued the sector before solid revenue could materialize. When returns failed to meet expectations, the bubble burst, wiping out countless startups.

Slok says the stock market's expectations are even more unrealistic today, with 12-month forward price-to-earnings ratios now exceeding the peak of the dot-com bubble.

Click to enlarge

A company's P/E ratio measures the relationship between its stock price and the profit it generates, with a high ratio reflecting optimism about future returns. Comparing S&P 500 ratios at five-year intervals from 1990 to 2025 clearly shows the dot-com spike in 2000. Similar spikes in 2020 and 2025 suggest the AI bubble may be even more pronounced.

What's more concerning is that in each spike, the top 10 companies' ratios far exceed the rest of the index. Such disparity suggests investments in those firms – mostly tech giants heavily betting on AI – have detached from reality before their newest technology can generate real profits.

Companies like Nvidia, Microsoft, Apple, Amazon, Meta, Alphabet (Google), and Tesla account for most of the S&P 500's recent growth.

Slok's warning echoes concerns from other industry leaders about the risks facing AI companies. Robin Li, CEO of Chinese internet giant Baidu, predicted that only about one percent of AI firms will survive if and when the bubble bursts. He said this will eventually lead to a more stable market with more realistic AI applications.

Tech giants continue to make massive investments as AI's popularity grows, underscoring the high stakes in this rapidly evolving field. OpenAI is developing an AI-powered web browser to challenge Google Chrome's dominance. Meta is spending over $60 billion to build new AI data centers.

Microsoft recently cut 9,000 jobs to offset costs from its new AI infrastructure, estimated at $80 billion. Amazon has unveiled plans for agentic AI, signaling that the race for AI leadership shows no signs of slowing.

Permalink to story:

The AI boom is more overhyped than the 1990s dot-com bubble, says top economist

 

[PXAbstraction]

Could not agree more. I've been saying this since the bubble began and the same people laughing me off now are the same people who laughed me off when I said crypto was a bubble. Just wait. Like crypto, AI isn't going away but anyone who thinks it's advancing or is even now in the place the bros claim it is, just wait.

 

[opckieran]

Could not agree more. I've been saying this since the bubble began and the same people laughing me off now are the same people who laughed me off when I said crypto was a bubble. Just wait. Like crypto, AI isn't going away but anyone who thinks it's advancing or is even now in the place the bros claim it is, just wait.

Right… the problem with declaring something to be “a bubble” is that even if it’s true that a bubble is due to burst, none of the naysayers (yourself included) ever seem to know when exactly the bubble bursts. It could be 3 days from now, 3 weeks, 3 months, 3 years, 30 years; financial markets have historically been reliably unpredictable (since literally any financial instrument can be viewed in any timeframe), and any snippet of price data can be cherry picked to fit a narrative.

Sure, the dotcom bubble burst, but most people who held onto those stocks instead of panic selling is surely better off today than they were before buying. Same with crypto. Saying the bubble is going to burst without saying when or at what price point seems kind of pointless.

 

[Kirby1]

Could not agree more. I've been saying this since the bubble began and the same people laughing me off now are the same people who laughed me off when I said crypto was a bubble. Just wait. Like crypto, AI isn't going away but anyone who thinks it's advancing or is even now in the place the bros claim it is, just wait.

Crypto was a bubble? Even tho bitcoin is worth over 100k each now?

Might need to rethink that.

AI is definitely not a bubble, this economist is wrong. There are definitely instances where it's overhyped or used as a marketing term and nothing more, but overall it's definitely not overhyped.

With the dot com bubble, everyone wanted to be the next Google but we don't need 100 Googles, so all the other Googles went out of business. But with AI there won't be a shortage of usage for it.

But that doesn't mean there aren't instances of investors making poor investment decisions based on hype, but that is not at all unique to AI.

 

[Thatsdisgusting]

Even more beautiful is how many people they're gonna fire now cuz "AI-AI-AI", and how many people will lose their jobs when this thing bursts.

Oh the world's gonna be in ruins if all this takes long enough

 

[yRaz]

Right… the problem with declaring something to be “a bubble” is that even if it’s true that a bubble is due to burst, none of the naysayers (yourself included) ever seem to know when exactly the bubble bursts. It could be 3 days from now, 3 weeks, 3 months, 3 years, 30 years; financial markets have historically been reliably unpredictable (since literally any financial instrument can be viewed in any timeframe), and any snippet of price data can be cherry picked to fit a narrative.

Sure, the dotcom bubble burst, but most people who held onto those stocks instead of panic selling is surely better off today than they were before buying. Same with crypto. Saying the bubble is going to burst without saying when or at what price point seems kind of pointless.

So the way I interpret bubbles is that "it's just a bad time to invest." The way I invest is that when I already made a bunch of money, I sell and I put the money into other emerging markets. I sold nearly all of my nVidia stock earlier this year, I still have some but I'm plenty happy with the gains I've made and If I'm wrong and they go to 8 trillion, I wont care. When someone says something is a bubble I wouldn't exactly say you should sell, just that you probably shouldn't buy it right now. Like you said, people who held onto tech stocks through the dotcom crash are doing just fine right now.

 

[VaRmeNsI]

The best investment is made after there is "blood on the streets". Invest in those that remain after it pops, just like after DotCom.

Like the internet, AI is only in its infancy.

 

[flipspot]

Dot com went Dot boom.

Still, the survivors crawled out of the wreckage, and created far more wealth than the Dot com investors ever imagined.

"What was the price of Amazon stock in 2000?
The closing price for Amazon (AMZN) in 2000 was $0.78, on December 29, 2000. It was down 80.9% for the year."

Anyone holding Amazon since Dot com went Dot boom? Congrats.

 

[Dr Roboto]

I learned this lesson the hard way during COVID.

**"The market can remain irrational longer than you can remain solvent"** is a famous quote attributed to John Maynard Keynes. It highlights the idea that even if an investor believes a market is fundamentally flawed and overvalued, it can continue to rise (or fall) for an extended period, potentially bankrupting those who bet against it prematurely. In essence, it emphasizes the unpredictable nature of markets and the risks of trying to time them based solely on perceived irrationality. (Internet Quote - Not my own)

 

[AnilD]

One key difference from the dot-com bubble is that cloud infrastructure and data are much more mature now. AI companies today can scale faster and more efficiently, but the question is whether they can convert that into sustainable profit before the hype runs dry.

 

[BuckarooBonzaii]

We should all know that AI will be for future generations after Gen Z, perhaps for Gen A, B, etc. as they're the ones who will grow up with it and it will be as common as the microwave oven. We're the testing platform. And I believe CEO Robin Li is right because once this bubble pops, the Tech Giants will be doing something else

 

[brucek]

Sure, the dotcom bubble burst, but most people who held onto those stocks instead of panic selling is surely better off today than they were before buying.

LOL no. You're thinking of the relative handful of companies that won in their segments. Yes, for these few winners, they spiked like everyone else, dropped like everyone else, but eventually emerged as solid businesses and over the years made a lot of money for their investors. Anyone who invested ONLY in these winners did fine to great to spectacular.

But the dotcom bust drove many companies to zero, or pennies. Investments in them were wiped out and did not recover. And at the time a lot of sophisticated investors could not accurately predict which were going to be the winners and which were going to be everyone else.

To me this high-risk some-winners many-losers is the practical meaning of this bubble warning. AI is going to be real, and is going to make a lot of people a lot of money. But there's also going to be a ton of money lost on the investments that don't pan out.

 

[Theinsanegamer]

Crypto was a bubble? Even tho bitcoin is worth over 100k each now?

Might need to rethink that.

AI is definitely not a bubble, this economist is wrong. There are definitely instances where it's overhyped or used as a marketing term and nothing more, but overall it's definitely not overhyped.

With the dot com bubble, everyone wanted to be the next Google but we don't need 100 Googles, so all the other Googles went out of business. But with AI there won't be a shortage of usage for it.

But that doesn't mean there aren't instances of investors making poor investment decisions based on hype, but that is not at all unique to AI.

Yes, it was a bubble, that has burst and reformed several times, screwing over hundreds of thousands if not millions of "investors" in the process.

It's also still a greater fool scam, with the only prospect of value being another sucker paying more for said asset, thus providing liquidity for early buyers to get out. There is nothing else backing the value of the currency.

Much like crypto, there are edge cases where AI has an actual use. Nothing justifies the insane valuations and investment though. Imagine how much more good could be done if we invested like this into nuclear fusion, or electrical infrastructure.

 

[scoffer]

I don't think there's any "MAY" about it............

 

[Neatfeatguy]

I'm still waiting to see AI.

Right now all I see are programs that gather information given to them and return a result asked of them to best of it's ability based on the data they've been fed. That's just a search engine.

 

[wiyosaya]

IMO, it does not take an economist to figure this out. All that is needed is common sense.

And just wait for the fall.

 

[Steveb8189]

I'm still waiting to see AI.

Right now all I see are programs that gather information given to them and return a result asked of them to best of it's ability based on the data they've been fed. That's just a search engine.

But it's not is it. A search engine just returns existing content rather than novel content.

 

[PXAbstraction]

Right… the problem with declaring something to be “a bubble” is that even if it’s true that a bubble is due to burst, none of the naysayers (yourself included) ever seem to know when exactly the bubble bursts. It could be 3 days from now, 3 weeks, 3 months, 3 years, 30 years; financial markets have historically been reliably unpredictable (since literally any financial instrument can be viewed in any timeframe), and any snippet of price data can be cherry picked to fit a narrative.

Sure, the dotcom bubble burst, but most people who held onto those stocks instead of panic selling is surely better off today than they were before buying. Same with crypto. Saying the bubble is going to burst without saying when or at what price point seems kind of pointless.

Every AI company is losing billions of dollars, far faster than they can bring it in. This is an industry in search of a business model. If I knew exactly when the bubble would pop, I'd be shorting a bunch of these companies, but I don't, nor does anyone else. But none of these companies have the remotest concept of a plan of when they'll be profitable and the VC money won't last forever.

 

[PXAbstraction]

Crypto was a bubble? Even tho bitcoin is worth over 100k each now?

Might need to rethink that.

AI is definitely not a bubble, this economist is wrong. There are definitely instances where it's overhyped or used as a marketing term and nothing more, but overall it's definitely not overhyped.

With the dot com bubble, everyone wanted to be the next Google but we don't need 100 Googles, so all the other Googles went out of business. But with AI there won't be a shortage of usage for it.

But that doesn't mean there aren't instances of investors making poor investment decisions based on hype, but that is not at all unique to AI.

Yeah, and how's basically every other coin doing? And how many times has Bitcoin crashed already? One coin riding its own bubble (driven largely by failing companies trying to be perpetual motion machines with Bitcoin holdings) does not a sustainable industry make.

 

[PXAbstraction]

I suspect most people who are into the hype are going to use "it's too long" as an excuse to dismiss it, but Ed Zitron (who has an excellent track record of predicting bubbles) goes into great detail about this here.

 

[Kirby1]

Yeah, and how's basically every other coin doing? And how many times has Bitcoin crashed already? One coin riding its own bubble (driven largely by failing companies trying to be perpetual motion machines with Bitcoin holdings) does not a sustainable industry make.

How many other publicly traded companies fail?

If crypto is an investment, bitcoin is the Nvidia of the market.

 

[PXAbstraction]

How many other publicly traded companies fail?

If crypto is an investment, bitcoin is the Nvidia of the market.

I don't think that point has the strength you think it has. Bitcoin has crashed and cost thousands (at least) their shirts multiple times now. That it rode back up again (and again) doesn't nullify that. And make no mistake, while NVIDIA will be doing just fine after this bubble pops, anyone buying in now is going to get a rude awakening.

 

[user556]

... but Ed Zitron (who has an excellent track record of predicting bubbles) goes into great detail about this https://www.wheresyoured.at/the-haters-gui/

Great read. Thanks.

To put it bluntly: Cursor's growth story was a lie. It reached $500 million in annualized revenue selling a product it can no longer afford to sell, suggesting material weakness in its own business and any and all coding startups.

This isn't any criticism of his article or points he's making but, to be fair to Cursor, Cursor isn't the one making the lies. They're just passing the lies on. OpenAI and co are the ones creating the false pricing structures and then pulling the rug, effectively, moments later.

 

[goatonastick]

I was hoping they would have some sort of evidence of the AI bubble. Pointing to the .com boom where a bunch of useless sites got too much funding isn't comparable to AI that is not only proven to be of use, but is in great demand.

 

[user556]

I was hoping they would have some sort of evidence of the AI bubble. Pointing to the .com boom where a bunch of useless sites got too much funding isn't comparable to AI that is not only proven to be of use, but is in great demand.

Wow! So a trillion dollars isn't an overspend?
AI totally hasn't proven itself to anything close to worth that.