OpenAI’s “WeWork moment” has arrived? Gary Marcus warns: The price war is destroying AI’s moat

```

The longstanding doubts about the "moat" of the AI industry are surfacing amid an increasingly intense price war. After reports that OpenAI is considering significantly cutting token pricing, Gary Marcus, who has long been bearish about the large language model industry, called this a "weak signal" and once again cited his judgment from years ago—OpenAI may be heading towards a "WeWork moment".

According to prior media reports, OpenAI is evaluating a plan to sharply reduce AI service token charges, directly targeting competitor Anthropic. Marcus promptly warned on social media that this is nothing but "burning money to keep clients", further evidence of OpenAI's troubles, and he predicts that if OpenAI falters, Nvidia, Oracle, and CoreWeave and other ecosystem partners may be affected as well.

This situation is particularly tricky for IPO prospects. OpenAI has already secretly submitted its application for listing this week, and whether its price reduction strategy can effectively seize market share without further widening losses has become a central question for investors assessing its IPO value. Meanwhile, SoftBank’s second attempt to use OpenAI shares as collateral for a loan was rejected by banks, raising further doubts about the appropriateness of its valuation. Marcus believes that Anthropic could be the biggest winner in this contest.

The shadow of the price war reflects the business model dilemma faced by leading AI companies: both sides are suffering tens of billions of dollars in losses due to huge computational costs, while the high substitutability of their products makes it very easy for clients to switch, calling into question the long-term durability of their moats.

Price war corroborates "no moat" theory

More than two years ago, Gary Marcus predicted publicly that the large language model industry would fall into a pattern of "no moat, thin profits, and price war", listing specific prediction items. Now, these predictions are being validated by successive events.

Reports indicate that OpenAI is evaluating a plan to significantly reduce token charges. OpenAI CEO Sam Altman recently admitted at an event that the cost of AI usage has become "a huge problem", and stated, "We will have many ways to help users get more value for less spending." Discussions are ongoing and subject to change.

Marcus characterized these developments on social media as a "sign of weakened strength", directly citing the collapse of WeWork's IPO precedent, and saying "won’t be surprised if OpenAI’s IPO plan ultimately falls through". He also noted that although critics often claim he "always predicts incorrectly", several of his judgments are gradually being confirmed by the market.

Anthropic’s competitive pressure spikes, enterprise spending enthusiasm cools

The immediate cause for OpenAI’s price reduction discussion is rising competitive pressure from Anthropic.

Anthropic’s programming tool Claude Code has become highly popular among software engineers, driving the company’s revenue sharply higher; its valuation recently surpassed OpenAI for the first time. In response, OpenAI has made its own programming tool Codex a key focus of development.

However, enterprise clients’ enthusiasm for spending on AI is cooling. Some companies with significant investment in Anthropic's products have begun seeking to control costs. Earlier this year, an Uber executive said the company’s autonomous AI usage budget for 2026 had already been exhausted; another company leader said last month that it was difficult to directly tie improved AI programming efficiency to the rollout of new customer features. These statements have triggered widespread discussion and reflection in Silicon Valley about “tokenmaxxing”—that is, consuming as many tokens as possible without regard for return in hopes of boosting productivity—and whether such behavior is reasonable.

SoftBank loan rejected, IPO story harder to tell

The price war will serve as an early stress test for both companies’ business models, with both currently at the cusp of high-profile IPOs.

OpenAI has secretly submitted its IPO application this week, and Sam Altman has said in a recent Slack message to employees that the company plans to complete its listing "within the next year". However, according to Marcus, SoftBank’s second attempt to use OpenAI shares as collateral for a loan was rejected by banks, which he called "embarrassing", indicating banks do not buy into OpenAI’s valuation narrative.

Investors have previously identified a potential risk: the products of OpenAI and Anthropic are highly substitutable, and the barriers for clients to switch from one to the other are relatively low. In its confidential IPO filing, OpenAI stated, "Some things may be more convenient to do as a private company", but gave no further details. If price cuts further compress already limited profit margins, it will become increasingly difficult for both companies to argue a long-term profitability story in their IPO roadshows.

Ripple effects: Nvidia, Oracle may be affected

Marcus’s concerns go beyond OpenAI itself. He clearly stated, once OpenAI loses ground, Nvidia, Oracle and CoreWeave, deeply connected in its ecosystem, are likely to be dragged down as well.

OpenAI and Anthropic currently occupy major shares of revenue from emerging AI products, sustaining their rapid growth. Both companies are losing tens of billions of dollars due to the computational costs needed for their AI systems to handle queries and tasks. If the price war further squeezes the revenue side, upstream computational power suppliers will also feel more pressure.

Marcus believes that in this competitive reshuffling, Anthropic is the most likely beneficiary, and its IPO prospects may be strengthened as a result. However, he also raises a larger uncertainty: whether Anthropic can fend off Google’s strong competition. On this, he admits "there’s no answer".

Risk DisclaimerThe market carries risks and investment should be cautious. This article does not constitute personal investment advice, nor does it take into account individual user’s unique investment goals, financial status or needs. Users should consider whether any opinions, views or conclusions in this article suit their particular circumstances. Investing based on this article is at your own risk. ```