Microsoft Pushes Claude Code Across Teams, Even as It Sells Copilot

Microsoft is rolling out Claude Code across major divisions and encouraging both developers and non-developers to use it. This move builds on earlier adoption of Anthropic’s models and reflects perceived usability advantages. Engineers are asked to compare Claude Code with GitHub Copilot, despite Microsoft continuing to sell Copilot.

Key Points

• Microsoft is encouraging thousands of employees across major teams to adopt Anthropic’s Claude Code.
• The rollout follows earlier internal adoption of Claude Sonnet 4 and prioritization for paid GitHub Copilot users.
• CoreAI has tested Claude Code, and the Experiences + Devices division was asked to install it.
• Non-developers are being urged to use Claude Code for prototyping; it’s approved across Business and Industry Copilot repos.
• Engineers are expected to use both Claude Code and GitHub Copilot and provide comparative feedback, signaling an internal preference for Claude Code’s usability.

Sentiment

The community overwhelmingly agrees with the article's thesis. The dominant sentiment is that Claude Code is genuinely better and that Microsoft's adoption is unsurprising given the widely perceived inferiority of Copilot. Commenters use the story as a springboard to critique Microsoft's broader organizational dysfunction, from branding chaos to incentive structures that punish quality work. There is a strong undercurrent of schadenfreude toward Microsoft squandering its massive AI investment and first-mover advantage. The minority defending Copilot generally concede the harness is inferior while arguing underlying models or specific configurations can be competitive.

In Agreement

• Claude Code has genuinely superior UX, tool orchestration, and polish compared to all competitors, and Microsoft's adoption reflects real product quality differences
• Anthropic succeeded by letting engineers solve real problems organically, while Microsoft failed by mandating "Copilot everywhere" from the top without understanding developer needs
• Microsoft had every advantage — the OpenAI partnership, distribution, enterprise relationships, massive investment — and still built a product their own engineers don't want to use
• The fact that Microsoft isn't dogfooding its own AI products undermines the case for paying for Copilot externally
• Claude Code's CLI-first approach is fundamentally better than IDE-coupled approaches that suffer from integration quality issues
• Microsoft's "Copilot" branding across dozens of unrelated products has diluted the brand to meaninglessness
• This repeats Microsoft's historical pattern of failing to build a product and then adopting someone else's better solution

Opposed

• GitHub Copilot with Anthropic models (especially Opus) is actually quite effective, and the quality gap is exaggerated by people who don't configure Copilot properly
• The Copilot CLI has been improving with features like plan mode and Agents.md support, making it competitive for daily use
• Copilot is best understood as a platform or marketplace rather than a single product — it can offer access to all the best models
• The wave of positive Claude Code press coverage may be coordinated Anthropic marketing rather than organic enthusiasm
• Microsoft is pragmatically evaluating tools and asking engineers to compare with feedback — this is healthy engineering culture, not failure
• Copilot's smart autocomplete remains the single most useful AI coding feature for many developers