More recent evidence that the AI boom is a bubble – and it's starting to burst; Ambulance handover ceremony in Kyiv

Today, I'm continuing my look at the AI boom and whether it's a bubble (see my e-mails from Monday and yesterday)...

1) This Wall Street Journal article details yet another indicator that the AI bubble is starting to burst: Failure of Australia's Largest IPO in Decades Sends Tremors Through AI Trade.

First, it summarizes how the current market environment is affecting AI IPOs:

Rising interest rates, a jumpy stock market and heightened political uncertainty are creating a dicey outlook for tech companies that had been aiming for a fourth-quarter public offering, causing some to reconsider their timing.

The planned IPOs of Anthropic and OpenAI have both been subject to that uncertainty and a driver of it, with other firms shifting their plans around what are expected to be among the largest public debuts in market history. Anthropic plans to host an investor day at its San Francisco headquarters on Wednesday after pushing its timeline back by several weeks.

Then it reports on the IPO failures from the title:

The latest signs of cooling interest came when two Australian data-center companies, Airtrunk and Firmus Grid, pumped their brakes on their public offerings within the span of a few days.

Firmus cited "recent market volatility and prevailing market conditions" in canceling what would have been the country's largest IPO in 30 years, with a planned raise of almost $5 billion. The company, which has a close relationship with Nvidia and is backed by Coatue and Blackstone, plans to raise funds privately and go public on a U.S. exchange next year, a person familiar with the matter said.

Airtrunk, which was preparing to list in Singapore, now hopes to go public in the first half of next year, The Wall Street Journal reported. The company, which is also backed by Blackstone, was concerned about market volatility and interest rates.

Some of us here in the U.S. might question the importance of two failed IPOs on the other side of the world... But when bubbles burst, the early warning signs are at the periphery – the overhyped, often-fraudulent latecomers to the party.

You might recall the two Bear Stearns subprime hedge funds that collapsed in the summer of 2007 that were famously nicknamed "Dumb and Dumber." They were an early catalyst for the global financial crisis.

But I'm reminded even more of the Internet bubble – when the little dominoes fell first, then the big ones. And investors realized they had massive exposure to all the little companies' ability to raise seemingly endless cheap money.

The WSJ article has parallels to this famous Barron's article from March 20, 2000, which highlighted how many Internet companies were running out of cash:

When will the Internet Bubble burst? For scores of 'Net upstarts, that unpleasant popping sound is likely to be heard before the end of this year. Starved for cash, many of these companies will try to raise fresh funds by issuing more stock or bonds. But a lot of them won't succeed. As a result, they will be forced to sell out to stronger rivals or go out of business altogether. Already, many cash-strapped Internet firms are scrambling to find financing.

An exclusive study conducted for Barron's by the Internet stock evaluation firm Pegasus Research International indicates that at least 51 'Net firms will burn through their cash within the next 12 months. This amounts to a quarter of the 207 companies included in our study...

The ramifications are far-reaching. To begin with, America's 371 publicly traded Internet companies have grown to the point that they are collectively valued at $1.3 trillion, which amounts to about 8% of the entire U.S. stock market. Any financial problems at these Internet firms would affect the myriad companies that supply them with equipment, including such giants as Cisco Systems and Intel.

And the bubble today is so much larger, both in terms of the money invested in AI infrastructure and the combined size of the relevant companies in the stock market.

All 371 publicly traded Internet companies at the peak of the bubble in 2000, including huge long-term winners like Amazon (AMZN) and Booking (BKNG), were worth $1.3 trillion. That's less than the valuation being discussed today just for Anthropic!

2) This second WSJ article also reminds me of the peak of the Internet bubble: Frenzied Stock-Options Trading Shows Individual Investors Are Still Hungry for Risk:

Individual investors aren't letting anything get them down these days.

Their fervor is showing up in frenzied trading of options, or contracts that allow holders to buy or sell a stock at a set price, by a certain date. Volumes in options tied to individual stocks have recently reached levels seen during the meme-stock mania of 2021, about $700 billion of notional volume a day...

3) A third WSJ article by investor Andy Kessler highlights How the AI Boom Ends – with pricing plunging by factors of 10:

Be careful about extrapolating anything from today's pricing. With AI, like all new technologies, the refinement stage is kicking in – just as mainframes gave way to cheaper minicomputers, microprocessors, web servers and now AI architectures. AI token prices have been dropping but are set to plunge faster than most anticipate. Tech growth is never a straight line.

Silicon Valley's mantra for decades has been smaller, cheaper, faster, better. I asked venture capitalist Steve Jurvetson to translate that into numbers. "We often see a technology shift when the new approach is 10 times better. It can be 10 times more performant or 10 times cheaper, sometimes both." He calls it the "factor of 10."

Kessler argues it's already happening:

Last week, according to OpenRouter, OpenAI and Anthropic were charging a premium for some of their models – $1 to $2 per million input tokens. Comparable Chinese open-weight models like Z.ai's GLM and Alibaba's Qwen charged 6 to 12 cents, around 90% less – that factor of 10 inflection point. American AI labs have gone from 70% market share last year to 30% today. All are still growing. But prices matter...

I often hear, that if tokens get cheaper, demand will go up. That's Jevons Paradox. Sure, elasticity works. But if prices drop 90%, demand must increase 900% just to stay flat. Remember when internet demand grew while debt-laden Global Crossing spectacularly failed because prices dropped too fast? Historically, semiconductors got 30% cheaper every year. When prices dropped by more than 50%, disaster struck with fabrication plant closings and company failures.

He concludes:

Will 2027 see peak data center? Again, timing is dicey. Current demand is gangbusters, but spending is a game of chicken until multiple factor-of-10 improvements kick in and prices collapse. This could set off a cascade of data-center halts and debt defaults, even as token demand increases.

Yes, model training, drug discovery and winning prizes like the recently solved Navier-Stokes equations require hot chips in data centers. But chatbots and agents and other tasks may increasingly be handled on PCs and phones and smart glasses and whatever new gizmos Silicon Valley invents over the next decade. I've heard this called the "Great Migration." Mr. Jurvetson's factor of 10 is visible now, as it often is, but ignored. Change doesn't happen overnight. It happens gradually, then suddenly.

4) Lastly, I want to highlight an essay by Ed Zitron on his Substack (subscription required): Annualized Enron Rate.

In it, he argues that the "entire AI bubble is inflated based on stupid, made up numbers that only exist to pump the bags of investors and make the AI startups look more successful than they are."

Regarding the huge annualized run rates claimed by OpenAI, Anthropic, and others, he writes:

There is absolutely no value in using "annualized run rates" for anyone other than the company and its investors, and the fact that modern tech and business journalism continues to accept it for what are meant to be the largest and most-important companies in the world is a sign that something is very wrong with how we cover the tech industry.

We don't even know the AI revenues, much less profits, of these giant companies:

We do not know Google, Amazon or Oracle's AI revenues, and we have a single year – fiscal year 2026 – of Microsoft's AI revenues based on an estimate from Bloomberg calculated from its disclosure of its OpenAI revenues in an SEC filing. We do not know how much each individual service makes, how much their true revenue shares are with the AI labs or, as I discussed a few weeks ago, how much actual, operational AI capacity exists as a result of the trillion-plus dollars in capital expenditures...

What we do have is "annualized run rates" pegged to non-specific time periods and non-specific numbers...

It is astonishing that we, four years into the AI bubble, do not know the revenues of more than a handful of AI companies outside of OpenAI and Anthropic. All we have is a mishmash of annualized run rates...

Zitron also shares this chart that highlights the worrying degree of customer concentration OpenAI and Anthropic have:

He concludes:

Anthropic and OpenAI Are Only Growing Because Of AI Startups' Perpetual Unsustainable Spending On Tokens – Once [Venture Capital] Dollars End, So Does Their Growth...

• The Revenues: 80% of Anthropic and OpenAI's revenues come from 1% of their customers, the vast majority of which are AI startups that spend anywhere from 300% to 1000% of their revenue on AI tokens.
• AI Startups Run Services at a Loss, Allowing Users to Burn Thousands of Dollars of Tokens for Hundreds of Dollars a Month, Requiring Constant Venture Funding
• Every New Model Release Artificially Boosts Revenues Through AI Startups
• All Of This Is Only Made Possible By Perpetual Venture Funding, And Once It Expires, It Eats Anthropic and OpenAI's Revenues Alive

Every AI startup is effectively an extraction mechanism for OpenAI and Anthropic, reselling their tokens through their services at a massive discount because users are unwilling to pay the actual cost of using [large language models].

If the AI bubble is starting to burst, what should investors do to protect themselves? I'll address this critical question in upcoming e-mails, so stay tuned...

Best regards,

Whitney

P.S. I welcome your feedback – send me an e-mail by clicking here.

P.P.S. On my second day in Kyiv, we handed off the last seven of the 57 ambulances to the Ukrainian military in a ceremony at the Motherland Statue, after driving them from Munich.

But we first had to go into this shelter, as a Russian drone was shot down nearby just as we arrived:

When the skies cleared, we proceeded with the ceremony:

I'm training for the New York City Marathon two weeks from Sunday (on my 60th birthday!), so I went for long runs around the city each day.

I ran by Maidan Square, the site of the February 2014 "Revolution of Dignity" that ended the regime of the Russian puppet president and tilted Ukraine toward democracy. But it also led to Russia's seizure of Crimea and, ultimately, the full-scale invasion in February 2022.

Maidan Square is also the site of a large memorial for fallen soldiers. I recorded a video that you can watch here. And here's a photo: