New York University professor emeritus Gary Marcus has warned that OpenAI could become “the AI version of WeWork.” The comparison comes as OpenAI rules out a 2026 IPO and carries a reported private valuation of $852 billion, renewing questions about how investors can test that valuation—and the funding behind AI infrastructure—without a public listing.
OpenAI delays IPO as private valuations climb
OpenAI reportedly submitted a confidential IPO filing in June, but CEO Sam Altman later said the company would not go public in 2026. Its most recent private financing took place in March, raising $122 billion at a reported valuation of $852 billion.
Market talk has put OpenAI’s valuation in a potential new funding round at between $1.2 trillion and $1.4 trillion. Anthropic’s pre-IPO valuation has also drawn attention in private markets, with on-chain trading markets indicating a figure above $1 trillion. Neither company’s private-market pricing has yet been tested through a public offering and exchange trading.
Marcus’s concern is not limited to the companies’ fundamentals. A shift in investor sentiment toward AI valuations could also affect expectations for chipmakers such as Nvidia and for institutions financing AI infrastructure.
Marcus points to WeWork’s failed 2019 IPO
Marcus cited WeWork as a cautionary comparison. The company’s 2019 IPO filing showed a loss of nearly $700 million in the first half of that year. Its private valuation had reached $47 billion, but fell to about $10 billion within weeks.
WeWork did not complete an IPO that year. It went public in 2021 through a merger with a special purpose acquisition company (SPAC) and filed for bankruptcy in 2023. Marcus’s example highlights how a high valuation in private funding rounds may not carry over to public markets. The two companies have different businesses and operating conditions, however; the comparison is about the potential gap between valuation expectations and market sentiment.
Anthropic’s listing could offer a valuation benchmark
Anthropic is preparing for a November listing. Marcus said the company still hopes to secure a valuation of about $1.5 trillion, while also citing market estimates close to $1.1 trillion. A target of roughly $2 trillion had circulated a month earlier, underscoring that the offering price has not been settled.
If Anthropic lists first, its pricing and subsequent share performance could give investors a public-market reference point for AI company valuations. OpenAI reported revenue of $13 billion through 2025; Marcus estimates that its operating losses are close to $2 billion a month. The combination of revenue at that scale and ongoing losses puts capital needs and the path to profitability at the center of valuation questions.
AI computing finance and chip spending in focus
Marcus singled out newer cloud providers that rent AI computing capacity to businesses as potentially more exposed to higher financing costs. If investors lower their return expectations for AI infrastructure, funding conditions for these companies could tighten further.
OpenAI’s hardware spending through 2030 is estimated at about $600 billion. Anthropic’s public-market valuation could shape how investors assess that spending, as well as the chip suppliers and lenders behind it. A strong debut for Anthropic would challenge the view that AI valuations are set for a broad decline. Until an offering price and public trading record are in place, private valuations remain untested in the open market.