The head of Microsoft's AI division aims to cut back on spending related to Anthropic.

      Mustafa Suleyman has pinpointed Microsoft's main rival in the AI space, and it isn't OpenAI. “Anthropic is extremely costly, and many are urgently searching for alternatives,” the head of Microsoft's internal model initiative stated in a Bloomberg interview.

      This remark goes beyond mere competitive analysis; it signifies a commitment. “We invest a significant amount of money in Anthropic, so our objective is to decrease and ultimately eliminate that expense,” Suleyman noted.

      To reinforce this claim, Microsoft unveiled seven new internal AI models during its annual Build conference for developers, including MAI-Thinking-1, a reasoning model that it asserts competes with Anthropic’s Claude Opus 4.6 on a popular coding benchmark, while being offered at a lower cost.

      The pricing rationale

      Suleyman's perspective comes at a time when enterprise AI expenditures are turning into a significant issue. Uber depleted its entire 2026 AI development budget in just four months and introduced a monthly limit of $1,500 per employee per tool. Similarly, Walmart has restricted access to its internal AI assistant after usage surpassed expectations.

      For Microsoft, which consumes vast amounts of AI tokens across Copilot and its engineering teams, the cost issue is critical. “Many people in our organization are spending millions of dollars” on AI tokens, Suleyman commented. Developing cost-effective in-house options is not only a competitive strategy against external providers but also a means to safeguard Microsoft’s profit margins.

      After enhancing its models for the consulting firm McKinsey, Microsoft claims it achieved ten times better cost efficiency compared to OpenAI's GPT 5-5, as per Suleyman. The company has also been in discussions with Adobe regarding the use of its in-house models, according to Bloomberg.

      From contractual restrictions to independent development

      Until late 2025, Microsoft was contractually barred from independently pursuing advanced AI development due to its partnership with OpenAI. A renegotiated agreement allowed the company to create competing models while retaining licenses for everything OpenAI develops until 2032.

      The time is now pressing. Suleyman’s MAI Superintelligence team, established in November 2025, has launched its first public models within six months. For a firm with little history in advanced model development, this pace is aggressive, but the gap still exists.

      Suleyman candidly acknowledged this. “We’ve closed a huge gap in six months,” he told Benzinga, while admitting that Anthropic has released two more advanced models since Opus 4.6, giving it a lead of several months.

      The focus on Anthropic over OpenAI

      The strategic reasoning is telling. Microsoft maintains discounted access to OpenAI’s models through 2032 but does not have a similar arrangement with Anthropic. This makes Anthropic the pricier reliance, and it's the one Suleyman can directly address by developing in-house alternatives.

      There’s also a competitive rationale. Anthropic’s Claude models have become the standard choice for enterprise AI coding tools, and the company is gearing up for an IPO that could value it over $1 trillion. If Microsoft can deliver comparable performance at a lower cost, packaged with the Azure infrastructure its enterprise clients already use, it weakens the justification for purchasing Anthropic separately.

      Whether Microsoft's models can truly rival Anthropic’s latest versions, as opposed to older ones, remains an unanswered question about Suleyman's seven new models. Matching Opus 4.6 on a benchmark is a strong starting point, but Anthropic has already advanced beyond it, and in a landscape where the cutting edge evolves every few months, bridging the gap and maintaining that position are notably distinct challenges.