OpenAI collapse could trigger a 50% AI sector rout, critic warns

Ed Zitron, a veteran tech critic who lived through the dot-com bust, is warning that OpenAI's collapse could become the Lehman Brothers moment for a $1.8 trillion AI industry.

The artificial-intelligence sector is repeating the same mistakes that wiped out hundreds of dot-com companies in 2000, according to Zitron, who founded the digital advertising firm 24/7 Media in 1995 and saw its valuation collapse from $2 billion to nine cents a share before WPP acquired it for $649 million.

"OpenAI is going down, and taking investors and an entire industry with it," Zitron said in a note published July 16. The critic drew a direct parallel to the 2008 Lehman Brothers bankruptcy, arguing that a high-profile AI company failure would trigger a contagion effect across the technology sector and broader equity markets.

Independent AI startups — those without existing revenue streams from search or social media — collectively owe billions of dollars while generating minimal income, Zitron said. The economics are brutal: training a single frontier model can cost more than $4 billion in compute power alone, with inference costs running $0.003 to $0.008 per 1,000 tokens depending on the model. Microsoft Corp., which has invested more than $13 billion in OpenAI, would face the most direct exposure if the startup collapsed, potentially dragging down the Nasdaq and other tech-heavy indices.

The Profitability Problem

The core issue mirrors the dot-com era: companies are prioritizing growth over profit, Zitron said. During the late 1990s, investors poured money into internet startups on the assumption that scale alone would eventually produce returns. Pets.com raised $82.5 million in its 2000 IPO and was bankrupt within nine months. Today, Chewy Inc. — a profitable online pet retailer — dominates the same market.

AI companies face a similar reckoning. OpenAI, Anthropic and other independent labs are spending billions on Nvidia Corp.'s H100 and B200 graphics processing units — each costing $25,000 to $40,000 — while their revenue remains a fraction of their operating costs. Data center construction alone is projected to consume $1 trillion in global capital expenditure by 2030, according to industry estimates, with hyperscalers like Amazon.com Inc., Microsoft and Alphabet Inc.'s Google accounting for the majority.

"The conversation should change about profit," Zitron said. "They absolutely need to find a path to profitability and fast."

The Public Trust Gap

A second risk factor is growing public resistance to AI. A recent commencement speaker at the University of Central Florida was booed for calling AI the next industrial revolution. Many workers view the technology as a direct threat to their jobs, a perception reinforced by recent layoffs at companies that have cited automation as a factor.

Zitron urged AI leaders to stop talking about artificial general intelligence — the hypothetical ability of machines to outperform humans at most economically valuable work — and instead focus on practical applications. "This doesn't help them seem relatable to the average Joe and creates a scenario where the public questions whether or not they're right in the head," he said.

What Comes Next

If Zitron's thesis proves correct, the fallout would extend well beyond AI startups. Microsoft trades at 35 times forward earnings, with much of that premium tied to its OpenAI partnership and Azure AI revenue growth. A collapse in AI valuations could force a broad re-rating of the so-called Magnificent Seven stocks, which together represent about 30 percent of the S&P 500's market capitalization.

Not everyone agrees with the bear case. Nvidia, which supplies the vast majority of AI training chips, reported data center revenue of $22.6 billion in its most recent quarter, up 427 percent from a year earlier. The company's guidance suggests demand remains insatiable. But Zitron argues that the current spending cycle is unsustainable — and that when it ends, the companies with the weakest business models will fail first.

"The dot-com bubble existed because the companies involved were focused on the wrong things," he said. "Their idea was solid; people did want to buy their dog food online. But Pets.com wasn't implementing the product in a compelling way. Today, Chewy.com does."

This article is for informational purposes only and does not constitute investment advice.