Microsoft is releasing new AI models to catch up with Anthropic, shifting focus to enterprise users rather than consumers. AI chief Mustafa Suleyman emphasized the company's "Anthropic-style" approach targeting business applications and coding. The company unveiled seven models including one for reasoning comparable to Anthropic's Opus 4.6, conceding Anthropic still leads by several months.

Microsoft previously relied on OpenAI but now pursues "true self-sufficiency" after restructuring. The tech giant retains a 27% stake in OpenAI with access until 2032 and invested $5B in Anthropic as part of a $30B cloud deal. Suleyman, who joined from Inflection in a $650M deal, oversees Microsoft's push to build state-of-the-art models.

Anthropic's Cowork tool triggered market sell-off and fears about enterprise software, with Microsoft shares down 10% year-to-date. The company unveiled an "ultra efficient" coding model for GitHub and aims to build "thinking and coding" agents for business customers. Microsoft hopes to reduce costs by sacrificing "significant margin" to Anthropic when serving products to customers.