Back
AzureMay 29, 20262 sources

Microsoft builds Copilot 'super app' and ships in-house coding model at Build, distancing from OpenAI

AI Analysis

Two coordinated reports this week make clear Microsoft is restructuring its AI strategy around independence from OpenAI. Fortune exclusively reported on May 29 that Microsoft is building a Copilot 'super app' bundling coding, chat, and other AI capabilities into one product — an attempt to reclaim AI leadership lost after the $13B OpenAI alliance failed to keep Copilot ahead of Anthropic's Claude Code and ChatGPT-as-consumer. The Information separately reported Microsoft will launch a new in-house AI coding model at Build, kicking off June 2; MSFT shares rose nearly 3% on the news.

The shift is visible at the product level too. Microsoft unveiled a 'buttoned-up' Copilot redesign that pulls AI out of some Windows 11 apps after user backlash, and Lifehacker published a how-to guide on fully removing Copilot from Windows 11 — reflecting that user pushback was loud enough that Microsoft retreated from forced AI placement. CEO Satya Nadella posted: 'We've redesigned Copilot to be simpler, faster, and more intuitive.' Microsoft AI CEO Mustafa Suleyman teased Build with a 'Loud and clear' post.

The broader narrative: Microsoft's exclusive-partner era with OpenAI is ending. Renegotiated partnership terms have been disclosed, and the in-house coding model — Microsoft's first serious frontier-class attempt — explicitly reduces dependence on OpenAI, Anthropic, and Google models that have powered GitHub Copilot. The pressure: Anthropic's Claude Code has decisively overtaken Copilot as the developer-preferred AI coding assistant, per multiple developer surveys this week.

Skeptical takes from the developer community: super-app strategies historically muddle brand and confuse users, and Microsoft's Copilot brand is already fragmented across GitHub Copilot, Microsoft 365 Copilot, Copilot Studio, and Windows Copilot. Watch next: the Build keynote on June 2, the in-house model's benchmark numbers, and whether enterprise customers care enough about OpenAI-independence to switch off Claude.

Sources
AI Briefing
·Curated by AI agents · Updated daily · 2026
Built by Koby Almog