Microsoft's AI chief stated that the Anthropic model is extremely expensive, and they plan to completely eliminate token procurement costs by developing their own model.

According to Beating, Microsoft's AI lead, Mustafa Suleyman, stated in an interview with Bloomberg that Anthropic's models are extremely expensive, and many users are urgently seeking alternatives. Suleyman revealed that Microsoft currently needs to pay Anthropic a huge sum of money, and the team's clear goal is to reduce and eventually eliminate related procurement costs. To reduce its reliance on external cutting-edge models, Microsoft released seven self-developed MAI series models at its Build 2026 developer conference in June. The newly released models include MAI-Thinking-1, which has inference capabilities. Microsoft claims that the new model can significantly reduce token costs while maintaining performance and compete with Anthropic's flagship model, Claude 4.6 Opus, in tasks such as programming. The urgency of cost control has prompted Microsoft to adjust its toolchain internally. Microsoft is gradually reducing its reliance on third-party tools such as the Anthropic programming assistant Claude Code, migrating internal R&D and daily development work to its self-developed GitHub Copilot CLI to control internal token consumption. Microsoft's accelerated development of its own models comes against the backdrop of a restructuring of its partnership with OpenAI. On April 27, 2026, the two companies signed a revised agreement that ended Microsoft's exclusive licensing of OpenAI technology and terminated revenue sharing payments from Microsoft to OpenAI. This restructuring paves the way for Microsoft to independently develop cutting-edge models and directly compete with OpenAI, Google DeepMind, and Anthropic.