What happened
Major technology companies Microsoft and Meta are significantly scaling back internal employee usage of Anthropic's Claude artificial intelligence models. According to reports published on October 5, 2026, by The Information, Microsoft has reduced its projected internal spending on Anthropic models by more than one-third. Earlier in the year, executives at Microsoft had anticipated spending at least $1 billion annually on internal staff access to Anthropic's technologies. Additionally, within Microsoft's Cloud and AI division, individual monthly AI usage spending caps were lowered from $100,000 to approximately $10,000 in most instances.
Simultaneously, Meta has seen the number of internal users relying on Anthropic's Claude Code coding assistant decline by half, falling from approximately 60,000 to 30,000 employees. Both corporate giants are taking active steps to direct their workforce toward alternative tools. Meta is promoting its homegrown software solutions, including MetaCode and Muse Code, while Microsoft is steering developer workflows back to OpenAI models delivered through GitHub Copilot alongside its own proprietary infrastructure.
Why it matters
The spending cutbacks by two of tech's largest companies present notable enterprise headwinds for Anthropic as the artificial intelligence developer prepares for a high-stakes initial public offering. In its recent IPO filings, Anthropic revealed that just two corporate customers account for approximately 25 percent of its total annual revenue. While Anthropic did not explicitly name those entities in its SEC filings, industry analysts and market observers widely believe Microsoft and Meta are among those top revenue sources.
A contraction in internal tool utilization by primary enterprise partners highlights broader shifts across the technology sector. Corporate clients are increasingly scrutinizing the soaring operational costs associated with running top-tier large language models. Furthermore, growing considerations surrounding data privacy, vendor reliance, and direct strategic competition are prompting major software corporations to champion internal artificial intelligence solutions rather than routing developer traffic to third-party providers.
Deep dive
The reduction in Claude usage highlights growing strategic friction between Anthropic and the established technology firms that helped adopt its early developer ecosystem. At Meta, internal adoption of Anthropic's Claude Code coding assistant spiked dramatically earlier in the year before executive leadership began prioritizing in-house developer tools. Meta reportedly views Claude Code as a direct strategic competitor to its business. To replace Claude, Meta launched proprietary coding applications across its software engineering teams. MetaCode has now surpassed 30,000 active internal users, while a secondary internal tool, Muse Code, has gained more than 6,000 users.
At Microsoft, the spending adjustment reflects a deliberate recalibration across its Cloud and AI division. While Microsoft maintains a massive commercial partnership with OpenAI, internal teams had previously made heavy use of rival Anthropic models for specialized engineering workflows. By introducing strict $10,000 monthly usage caps per employee and encouraging staff to utilize OpenAI models via GitHub Copilot, Microsoft is seeking to consolidate its infrastructure expenses while solidifying its primary corporate alliance with OpenAI.
These internal pullbacks come at a critical financial moment for Anthropic. According to reports from Bloomberg, Anthropic experienced explosive top-line growth in 2025, with revenue surging to approximately $4.6 billion from $386 million in the prior year. However, the vast computing costs required to train and serve frontier models caused Anthropic's net loss to expand nearly fivefold, reaching almost $42 billion.
Despite these massive operational losses, prospective public investors continue to value Anthropic at historical premiums. The company's most recent private funding round established a valuation of $965 billion, while prospective public investors are reportedly valuing the firm between $1.8 trillion and $2 trillion. Anthropic confidentially filed for an initial public offering on June 1, 2026. Reporting from Bloomberg and SmartAsset.com indicates that Anthropic could begin formally marketing the listing to investors as early as the week of November 9, 2026, targeting public trading prior to the Thanksgiving holiday on November 26, subject to market conditions and regulatory approval by the Securities and Exchange Commission.
Report check
Reporting by The Information on October 5, 2026, established that Microsoft reduced its projected internal spending on Anthropic models by over one-third, down from an earlier expectation of at least $1 billion per year, and reduced individual monthly usage caps in its Cloud and AI group from $100,000 to roughly $10,000. The same outlet established that Meta's internal users of Claude Code fell from roughly 60,000 to 30,000 as MetaCode grew past 30,000 users and Muse Code passed 6,000 users. Outlets citing these reports include TradingView, Stocktwits, TipRanks.com, daily.dev, AI Weekly, and The Decoder.
It is widely alleged by market observers that Microsoft and Meta are the two unnamed corporate clients that Anthropic reported in its regulatory filings as accounting for 25 percent of its overall revenue. It is also alleged that Meta management explicitly views Claude Code as a commercial competitor to its business. Neither Microsoft nor Meta has publicly confirmed the reductions in internal Claude usage.
Reporting by Bloomberg on October 1, 2026, and SmartAsset.com on October 2, 2026, established that Anthropic's 2025 revenue reached roughly $4.6 billion against a net loss of almost $42 billion, with a private valuation of $965 billion and prospective IPO valuations between $1.8 trillion and $2 trillion. What remains unknown is the official public offering date, share price, final valuation, and whether the SEC review will permit trading before late November 2026.
Open questions
The central question for prospective public investors is whether Anthropic can diversify its commercial customer base quickly enough to offset declining internal usage from major tech firms ahead of its stock market debut. It remains unclear how heavily other enterprise clients will follow Microsoft and Meta in paring back reliance on top-tier external models in favor of proprietary software or open-weight solutions.
Another open question is whether Anthropic's upcoming listing will achieve its ambitious target valuation of up to $2 trillion while carrying substantial annual net losses. Market observers will be watching whether Microsoft or Meta make official statements regarding their internal usage caps, and whether Anthropic will clarify the identities of the key corporate customers representing a quarter of its revenue when formal prospectuses are released to the public.