Last Friday, the stock market suffered a major crash, leading some to suspect that it may be on the verge of a much more severe crash. The two main factors were second thoughts about astronomical valuations of tech companies and rising inflation.
The tech-heavy Nasdaq fell about 4.2%, dragged down by losses of more than 10% at major chipmakers including Micron, Marvell, Intel, AMD, Qualcomm and Arm Holdings. Nvidia fell more than 6% and Broadcom fell almost 8%. The decline wiped out about $1.2 trillion in market value in one day. The Nasdaq recouped less than a quarter of Friday’s losses on Monday.
Other works by Robert Kuttner
The soaring stock market is one of the contradictions of the Trump economy. The general explanation is that although consumers feel squeezed by rising inflation, market valuations are strong thanks to the AI revolution and the profits of other technology platform monopolies such as Amazon and Google, as well as semiconductor manufacturers.
Friday’s selloff came against a staggering jump in prices for chipmakers. Intel stock has increased 453% over the past year. AMD stock is up 303%. Both rely heavily on the pending AI infrastructure build-out, which is expected to cost trillions of dollars.
On Friday, markets were also roiled by a stark collision of good and bad news. The Labor Department reported a faster-than-expected increase in employment in May to 172,000 people. Worried about further inflation, the bond market immediately raised interest rates on Treasury bills. Both factors made it more likely that the Fed would raise rates, which is always bad news for the stock market.
A close comparison is the dot-com stock bubble and crash of 2000.
Underlying this familiar interest rate movement was growing concern about the possibility of an AI bubble. It makes up an unusually large portion of the stock market, driven by AI valuations and the stock prices of other technology platform companies that rely on the AI boom.
The two largest AI companies, OpenAI and Anthropic, are privately held, but both plan to go public in the coming months. Keizai Shimbun reports that each IPO is likely to be valued at around $1 trillion. Neither company has formally filed with the SEC, but OpenAI’s filing could come at any time. According to , OpenAI was valued at $852 billion in a recent private funding round. wall street journal.
Another company in the trillion dollar club has already launched an IPO. Elon Musk’s SpaceX said in a filing that it plans to sell 555,555,555 shares at a price of $135 each. This puts the company’s valuation at about $1.77 trillion, roughly double the company’s valuation from six months ago. In 2020, SpaceX was valued at just $36 billion. All of its profitability is based on exclusive mistress contracts with the government. SpaceX is set to begin trading on the NASDAQ on Friday in the largest IPO in history. Assuming OpenAI and Anthropic move forward with their respective IPO plans, the three companies alone will amount to about $4 trillion, a huge amount of capital.
Artificial intelligence may be making a comeback Although it has been hyped as a revolutionary technology, it is still economically overvalued at this time. A close comparison is the dot-com stock bubble and crash of 2000.
In the late 1990s, the potential of new technologies related to the Internet led venture capitalists and other investors to pour money into technology startups. Burn rates for many of these companies far exceeded profits. Investors were betting on future profits to drive up stock prices. That money was earned from playing stocks, not from actual profits. The value of the tech-heavy Nasdaq nearly tripled between 1997 and early 2000.
When several startups failed and the Fed began raising interest rates in 1999, herd instincts reversed and investors began selling. In the ensuing crash, the Nasdaq lost 78 percent of its value between March 2000 and October 2002. This crash did not immediately collapse the economic scale. That then happened with the bursting of the housing bubble in 2008. Some technology companies with viable business models, such as Amazon, Google, and eBay, recovered and quickly became bigger than ever.
The AI analogy is imperfect but powerful. Similar to tech bubble stocks in 1999 and 2000, AI’s capital utilization and projected value growth far exceed its returns. A huge number of users are using OpenAI’s ChatGPT and Anthropic’s Claude for free. The devil’s bargain that consumers make is that in exchange for being able to use these applications for free, companies can use their data for AI training and God knows what else.
AI companies are treating these free applications as gateway drugs, trying to convince consumers, especially business consumers, to migrate to paid applications. This is definitely coming, but not necessarily at a fast enough pace to justify the astronomical valuations of AI and AI-related companies.
Most of the trillions of dollars that AI companies are raising and spending are expected to go toward building data centers. OpenAI is spending about $1.4 trillion on data center infrastructure over the next eight years and doesn’t expect to become free cash flow positive until 2030. NVIDIA CEO Jensen Huang recently said that total spending on AI infrastructure could total $3 trillion to $4 trillion by 2030.
Some of the smartest people in the world are trying to find ways to monetize all that capital expenditure. They have only been partially successful.
In 2024, one of the most influential AI bears, David Kahn of Sequoia Capital, released a much-cited report raising what he called AI’s $600 billion problem. $600 billion was the revenue needed to justify the massive capital investment in AI. He was skeptical that AI would generate enough revenue.
Since then, Khan acknowledged that AI companies’ revenues have increased. OpenAI reports $20 billion in revenue in 2025. But some of this growth is due to re-pricing of AI for business consumers, leading to astronomical costs for companies that were promised savings by laying off employees and replacing them with technology. One company reportedly spent $500 million on Claude in one month after Anthropic moved to usage-based billing. If companies don’t incur these costs, AI companies could lose out on their only reliable revenue growth potential.
This equation is further complicated by a widespread and growing popular uprising against AI data centers that are raising electricity prices for other users and straining the power grid.
Illinois Governor JB Pritzker announced that his administration will stop approving new state tax breaks for data centers in Illinois starting July 1. He ordered the state Department of Commerce to stop processing new applications and asked lawmakers, utilities, labor and environmental groups to devise an entirely new approach.
Pritzker wants tech companies to pay in their own way and have separate fee classes for data centers. Rather than have ratepayers fund the AI boom, he wants them to fund clean energy themselves. He wants the power to be shut off first when the grid becomes strained.
In Virginia, as my colleague Gabriel Gurley reports in this article, outlook County after county is revolting against data centers. In the Republican state of Utah, a plan to build a massive data center on 40,000 acres (more than twice the size of Manhattan) in Box Elder County is drawing growing opposition from residents. The project would require about 9 gigawatts of electricity, more than the entire state of Utah currently consumes. It would also use a lot of water in a county that has experienced severe drought in recent years. In response, the project’s sponsor, celebrity investor Kevin O’Leary, offered to reduce the size to 10,000 acres.
The backlash against data centers is still in its infancy, and questions will also arise about the soaring valuations of AI companies and their burn rates as further restrictions are imposed. The national debate about how to regulate AI is still in its infancy.
Bernie Sanders, on the other hand, came up with a rare bad idea. Sanders has proposed nationalizing some parts of AI. Not only President Trump but also some AI executives have jumped on board this trend. But the devil is in the details. Bringing the government in as a partner at the top of the bubble will only inflate the bubble and leave taxpayers with the bag of bailouts in the downturn. And, as our friend Matt Stoller asks, if we’re going to give the government a stake in explicit AI companies like OpenAI and Anthropic, why not put some of the more diversified tech monopolies, like Google, which are heavily invested in AI and have their own AI products, into public ownership?
Finally, let’s take a closer look The positive employment report triggered panic selling in financial markets on Friday. Despite President Trump’s economic boasts (“It’s raining jobs!”), job creation hasn’t led to wage growth.
Average hourly wages for private sector workers rose by just 0.2% in May, for an annualized rate of 2.4%. But prices are rising almost twice that amount, at about 3.8% a year. In other words, real wages are falling.
So why is this relatively slack economy creating jobs at this rate? The answer is that the federal deficit currently stands at about $1.9 trillion, or about 5.9% of GDP. This is nearly double the 50-year average. In other words, job creation is the result of a pure Keynesian stimulus.
Unfortunately, that’s the wrong kind of Keynesian stimulus. It is driven almost entirely by tax cuts for the wealthy rather than by reduced public investment. And while it’s not countercyclical, it continues year after year, which means it’s unsustainable. If the Fed raises interest rates, the cost of carrying all that debt will itself be a drag on the economy.
There was a time when Republicans, as fiscal conservatives, might have opposed it. However, the Republican Congress continues to impose President Trump’s budget deficit without a single protest.
Tax cuts that put more money into the pockets of wealthy investors also interact nicely with stock market bubbles. All of this suggests that when a crash occurs, it will be a disaster.
