Is the AI business model under threat? Survey shows: Two months after launch, Athropic's most expensive model, Fable 5, accounts for only 11% of total enterprise spending

Fable 5, the most powerful and highest priced AI model under Anthropic, is facing a tepid response from enterprise clients, a trend that is shaking the long-standing business logic of leading AI labs that depend on “top models driving revenue.”

According to a Sunday report by the Financial Times UK, referencing payment platform Ramp’s tracking of spending data from 70,000 enterprises, Fable 5 has only accounted for about 11% of total spending across all Anthropic products in the more than two months since its launch in early June. Its growth momentum has clearly stalled. This breaks the established preference among enterprise users for choosing the most powerful model. Analysts and Anthropic investors believe this shift is mainly due to Fable 5’s high pricing, and the reality that older models are already sufficient for most business needs.

Weak demand for Fable 5 has cast doubt on the sustainability of Anthropic’s business model. Cutting-edge AI labs have long concentrated billions of dollars in R&D into training ever-larger and more complex models. If enterprise clients continue switching to cheaper alternatives, this approach faces a fundamental challenge.

This development coincides with Anthropic’s preparations for what the market expects will be the largest IPO in history. Investors estimate the company may be valued at $2 trillion or more, with the listing possibly starting as soon as next month. Anthropic’s July revenue failed to meet the most optimistic expectations—some investors had forecast annual sales exceeding $80 billion, while last week the company informed shareholders that actual annualized revenue for July reached $65 billion, up from $47 billion in May.

Enterprises Turn to Cheaper Models as the Luster of Frontier Models Fades

Data from Ramp shows that Anthropic’s own Opus 5—a lower-priced, smaller-scale but still robust model—has surpassed Fable 5 in enterprise spending since its launch at the end of July. Meanwhile, OpenAI’s GPT 5.6 is priced significantly lower than Fable 5, and its parent company’s annualized revenue has jumped 35% so far this quarter, now exceeding $40 billion. The release of GPT 5.6 has reversed OpenAI’s struggles earlier this year.

The availability of cheap open-source models is also giving enterprise clients more options, further squeezing the bargaining power of top U.S. AI labs. Ramp’s chief economist Ara Kharazian points out that data retention rules imposed by the Trump administration have also hindered Fable 5’s adoption.

In early June, the Trump administration, citing national security, forced Anthropic to temporarily withdraw the model, which had a significant negative impact on Anthropic’s June revenue growth. It was not until July 1 that the government allowed Fable 5 to relaunch, lifting related restrictions. Analysts and investors say that with the relaunch approved, concerns about further limitations on Fable have gradually faded. Political uncertainty now takes a back seat to price and performance factors in clients’ model choices.

Miles Clements, a partner at Accel, noting the firm’s nearly $1 billion investment in Anthropic, stated, “Most people don’t need to operate at the frontier level,” and the era of enterprise clients always choosing the strongest model “is not a sustainable phase.”

Rapid Growth Masks Hidden Risks; IPO Outlook Uncertain

Despite these pressures, Anthropic’s overall growth remains strong. The company’s revenue has grown nearly sevenfold since the start of the year and recorded its first adjusted operating profit in the second quarter. It has also told investors that it expects to post another profit in the third quarter. According to people familiar with the matter, Anthropic has also informed investors that it now has 6,000 clients with annual spending of $100,000 or more.

However, the structural issues revealed by weak demand for Fable 5 have raised market doubts about the sustainability of Anthropic’s business model. Leading AI labs have concentrated billions of dollars of R&D into training larger and more complex models, but if enterprise clients continue turning to cheaper alternatives, this entire approach faces a fundamental challenge.

Kharazian frankly admitted that predicting Anthropic’s performance in the coming months “is almost an impossible task.” “If you extrapolate from previous trends, you’d expect Anthropic to dominate the market. But because OpenAI’s latest model is performing well while Fable is underwhelming, the result is just the opposite.”