A prominent technology critic is drawing a stark parallel between OpenAI and Lehman Brothers, warning that the ChatGPT maker’s financial fragility and outsized influence over the broader artificial intelligence sector could set the stage for a market correction with systemic consequences. Ed Zitron, whose commentary on the technology industry has attracted significant attention in recent years, argued in a widely circulated essay that what markets are calling an AI bubble is, more precisely, an OpenAI bubble — and that its eventual deflation could be far more damaging than investors currently appreciate.
The warning arrives as scrutiny of OpenAI’s underlying economics intensifies. According to leaked OpenAI financials, the company recorded approximately $21 billion in losses against $13 billion in revenue, a gap that raises serious questions about the path to profitability for a business valued at more than $300 billion. Zitron’s central argument, as reported by Business Insider, is that OpenAI functions as the gravitational centre of the entire AI investment ecosystem — and that its collapse would not be an isolated event but a trigger for cascading failures across the sector.
A Valuation Built on Expectation, Not Earnings
The Lehman Brothers comparison is deliberately provocative, but Zitron’s underlying logic follows a recognisable financial pattern. Lehman’s 2008 collapse was not merely the failure of one institution; it was the exposure of a system that had priced risk incorrectly and distributed that mispricing across interconnected balance sheets. Zitron contends that OpenAI occupies a structurally similar position — not because it is a bank, but because its valuation has become a benchmark against which the entire AI sector is measured. When the benchmark is wrong, everything anchored to it is wrong.
The numbers underlying that valuation remain difficult to justify on conventional financial terms. A $21 billion annual loss, sustained while the company simultaneously pursues an initial public offering and a restructuring from nonprofit to for-profit status, would be alarming in almost any other sector. The AI industry has, to date, been largely insulated from the scrutiny applied elsewhere, in part because investors have accepted the premise that the current spending phase is a necessary precursor to an eventual payoff of extraordinary scale. That premise is now being tested more seriously. Concerns about AI sector sentiment have already begun to surface in adjacent markets, including memory chips, where forward guidance has grown more cautious.
Infrastructure Spending and the Demand Gap
Central to the bearish case is the mismatch between what the technology industry is spending on AI infrastructure and what end users are actually paying for AI-powered products and services. Hyperscale cloud providers have collectively committed hundreds of billions of dollars to data centre construction, chip procurement, and model development over the next several years. The assumption embedded in those commitments is that enterprise and consumer demand will scale to meet the supply. Evidence for that assumption remains thin.
A separate analysis published by TheStreet examining Apple’s AI gamble identified a related structural weakness: that consumer-facing AI features have so far failed to drive the kind of hardware upgrade cycles or subscription revenue that would justify the capital expenditure underpinning them. That finding reinforces Zitron’s broader argument — that the AI sector has built a vast and expensive supply chain for a product whose demand curve remains speculative.
Contagion Risk and What Comes After
The implications of an OpenAI stumble extend well beyond the company itself. Anthropic, which is separately navigating its own path toward a public market debut — a process that Wall Street banks have been actively facilitating through investor roadshows — has raised capital at valuations that are partially derivative of OpenAI’s own pricing. If OpenAI’s valuation compresses sharply, whether through a troubled IPO process, a deteriorating loss trajectory, or a loss of enterprise confidence, the repricing pressure on Anthropic and other privately held AI firms would be immediate and significant.
Zitron stops short of predicting a precise timeline, but the structural argument he advances is difficult to dismiss on its merits. The AI investment cycle has been sustained by a combination of genuine technological progress, competitive anxiety among large technology firms, and a willingness among institutional and retail investors to tolerate losses in exchange for future optionality. All three of those supports are now showing signs of strain simultaneously. Whether OpenAI proves to be the catalyst for a broader reckoning or manages to stabilise through an IPO that resets expectations on more conservative terms will likely define the trajectory of AI investment for the remainder of the decade.