Why This Analyst Says The AI ​​Bubble Is 17 Times Bigger Than The Collapse Of Internet Companies

A version of this story will appear in CNN Business’ Nightcap newsletter. To receive it in your inbox, sign up for free here.

new York
—

At this point, even the concept of an “AI bubble” seems to be a bubble. (In fact, Deutsche Bank analysts said last month that the Bubble “AI bubble” has already broken out.)

Maybe some corners of the internet are bored with the bubble talk. That doesn’t make the market any less bubbly.

Just this week, the Financial Times wrote that 10 AI startups — not a dollar in profit among them — gained nearly $1 trillion in market value over the past 12 months. (That is, to use a technical term, bananas.)

Although Wall Street analysts and the tech media increasingly question this hype, difficult comparisons with the late 1990sthe AI ​​industry’s response has been to shrug its shoulders and watch its valuations rise higher and higher. AI enthusiasts believe the technology will disrupt (hopefully in a good way!) virtually every aspect of modern life, from phone operating systems to pharmaceuticals to finance. And even if there is a bubble, proponents say, the dotcom bubble gave us companies like Amazon, and the internet became…the internet.

Many skeptics object to the AI ​​hype, although few professional market analysts have done so as forcefully as Julien Garran, a researcher and partner at the British firm MacroStrategy Partnership.

Earlier this month, Garran released a report saying we are in “the largest and most dangerous bubble the world has ever seen.” Garran concludes that there is a “misallocation of capital in the United States” that makes the current frenzy 17 times larger than the dotcom bubble and four times larger than the 2008 housing bubble.

Needless to say, this is a bold claim regarding a phenomenon that is notoriously difficult to predict.

I sat down (virtually) with Garran earlier this week to talk about bubbles and why he thinks AI fervor is, to quote his report, not just “a bit bad” but rather “the antithesis of socio-economic progress.”

The following interview has been edited for length and clarity.

Nightcap: Your latest report on AI made a lot of waves among financial and tech media junkies like me. Can you explain the main lines to me?

At the heart of the note is a golden rule that I developed, which is that if you use large language model AI to create an application or service, it can never be commercial.

One reason is the way they were built. The original large AI language model was built using vectors to try to understand the statistical probability of words following each other in the sentence. And although they are very intelligent and this requires a very good level of engineering, they are also very limited.

The second thing is how LLMs have been applied to coding. What they’ve learned – the coding that exists, both in and outside the public domain – means that they’re effectively showing you pieces of code learned by rote. Again, this will be limited if you want to start developing new applications.

And the third set of problems, in terms of how it’s built, is around the idea of ​​scaling. There is a real problem at some point in terms of how much money needs to be spent to improve them. I would say that it is certain that (the developers) hit a wall to scale. Otherwise, they would release obviously better and better models every time they hit the market with a new product. And since the release of ChatGPT 4 in March 2023, they I haven’t really increased the bar significantly.

Nightcap: What about the argument that ChatGPT, while imperfect, is capable of low-level work that could increase productivity?

Garran: There are some bullshit jobs – in certain areas of management, consulting, jobs where people don’t check if you’re doing things right or don’t know if you’re doing them right. So you can argue that you can replace bullsh*t with bullsh*t, and, yes, OK, I’m willing to accept that you probably can, but that doesn’t really make it any more useful.

Nightcap: So what should ordinary people think about all the huge amounts of money flowing through the industry?

Garran: The AI ​​ecosystem can’t really sustain itself. Nvidia is making a ton of money… Everyone else – the data centers, the LLM developers, the software developers that use the LLMs – are all very loss-making.

Therefore, to maintain the process requires continued funding, which is why it looks like a permanent fundraising tour. But despite all this, there is no obvious way to turn this into profit. It’s hope that takes precedence over realistic expectations… When you lack investors, then everything will turn upside down.

Last drink: are investors really pulling out?

Garran: The willingness (of venture capitalists) to fund some of these startups, especially software developers, is starting to diminish due to their high value. That basically leaves you with SoftBank, which had to accumulate a lot of debt against its actions to finance the first tranche of the OpenAI commitment it made, and it still has a second, larger tranche to finance.

There are foreign states like, for example, Saudi Arabia. But few countries have unlimited purchasing power. And that then leaves Nvidia as the last man standing.

Nightcap: Are we deflating or is the bubble still growing?

Garran: With the AI, I can’t say it’s starting to deflate. We’re barely a week after the all-time highs, so it would be a little arrogant to say that this was definitely the peak. But it certainly comes close.

Nightcap: I have to ask, because I ask myself this all the time: what if you’re wrong? What if the hype was real?

Garran: Well, there are two reasons why I would be wrong.

The first is that it just takes longer to break than I thought. Which, to be honest, it already is. And what happens if I’m wrong is they just continue to build things that are not fundamentally useful to economic society.

If it goes on for another year or two because they managed to persuade someone to provide funding, more people are doing things that won’t make any money. The future won’t be so bright. The future (gross domestic product) will be less than it would be if they didn’t do this stuff, and just did mundane things that people actually enjoyed.

And if I’m completely wrong… (and) someone could invent a “superintelligence”, well, that would completely change the world, and we would depend on whoever controls the systems. We could be in some sort of utopia, or we could be in the “Brave New World.” Or we could be in a utopia like Kurt Vonnegut’s book Player Piano, where everyone is employed except for some people living in their ivory towers.

To be honest, I think it’s beyond our current capabilities as an industrial society to achieve this. If this starts to change, I will change my mind very quickly. I just didn’t see it.