Claude Code is suddenly everywhere inside Microsoft

Developers have been comparing the strengths and weaknesses of Anthropic’s Claude Code, Anysphere’s Cursor, and Microsoft’s GitHub Copilot for months now, looking for a winner. While no individual AI coding tool manages to be the best at every task that software developers do each day, Claude Code is increasingly coming out on top for its ease of use, both for developers and nontechnical users.

It seems like Microsoft agrees, as sources tell me the company is now encouraging thousands of its employees from some of its most prolific teams to pick up Claude Code and get coding, even if they’re not developers.

Microsoft first started adopting Anthropic’s Claude Sonnet 4 model inside its developer division in June last year, before favoring it for paid users of GitHub Copilot several months later. Now, Microsoft is going a step beyond using Anthropic’s AI models and widely adopting Claude Code across its biggest engineering teams.

Microsoft’s CoreAI team, the new AI engineering group led by former Meta engineering chief Jay Parikh, has been testing Claude Code in recent months, and last week Microsoft’s Experiences + Devices division were being asked to install Claude Code. This division is responsible for Windows, Microsoft 365, Outlook, Microsoft Teams, Bing, Edge, Surface, and more.

Even employees without any coding experience are being encouraged to experiment with Claude Code, to allow designers and project managers to prototype ideas. Microsoft has also approved the use of Claude Code across all of its code and repositories for its Business and Industry Copilot teams.

Software engineers at Microsoft are now expected to use both Claude Code and GitHub Copilot and give feedback comparing the two, I’m told. Microsoft sells GitHub Copilot as its AI coding tool of choice to its customers, but if these broad internal pilot programs are successful, then it’s possible the company could even eventually sell Claude Code directly to its cloud customers.

Microsoft is now one of Anthropic’s top customers, according to a recent report from The Information. The software maker is also counting selling Anthropic AI models toward Azure sales quotas, which is unusual given Microsoft typically only offers its salespeople incentives for homegrown products or models from OpenAI.

Microsoft’s decision to adopt Claude Code more broadly among its engineering teams certainly looks like a vote of confidence in Anthropic’s AI tools over its own, especially as it’s encouraging nontechnical employees to try out coding. But the reality is that Microsoft’s developers are likely to use a mix of AI tools, and adopting Claude Code is another part of that tool set.

“Companies regularly test and trial competing products to gain a better understanding of the market landscape,” says Frank Shaw, Microsoft’s communications chief, in a statement to Notepad. “OpenAI continues to be our primary partner and model provider on frontier models, and we remain committed to our long-term partnership.”

While Microsoft remains committed to OpenAI, it is increasingly working with Anthropic to bring its models and tools to Microsoft’s own teams and the software it sells to customers. Microsoft and Anthropic signed a deal in November that allows Microsoft Foundry customers to get access to Claude Sonnet 4.5, Claude Opus 4.1, and Claude Haiku 4.5. The deal also involves Anthropic committing to purchasing $30 billion of Azure compute capacity.

Microsoft has also started favoring Anthropic’s Claude models inside Microsoft 365 apps and Copilot recently, using them in specific apps or features where Anthropic’s models have proved more capable than OpenAI’s counterparts.

The big question here is, what does the increased use of Claude Code at Microsoft mean for its more than 100,000 code repositories? Microsoft told me last year that 91 percent of its engineering teams use GitHub Copilot and a variety of teams have been using the AI tool to speed up mundane tasks. Microsoft’s use of AI tools has been largely restricted to software engineers, but with Claude Code and Claude Cowork, Anthropic is increasingly focused on making coding and non-coding tasks more approachable, thanks to AI agent capabilities.

Microsoft is embracing the ease of use of Claude Code to allow more nontechnical employees to commit code using AI, and this broad pilot will certainly highlight the challenges and benefits of that shift. It also puts further pressure on junior developer roles, with fears in the industry that these roles are increasingly disappearing because of AI. Microsoft just took another big step toward a future where more autonomous AI agents are creating code, further wrestling control from its software engineers.

Microsoft is getting ready to show off two of its biggest Xbox games this year, Forza Horizon 6 and Fable, later today as part of its Xbox Developer Direct stream. There will also be a first in-depth look at Beast of Reincarnation and at least one other game shown, I’m hearing. Double Fine is ready to show off Kiln, a multiplayer, team-based brawler. I understand Double Fine has been holding playtests recently, where you play as a spirit that can inhabit pottery and carry water to douse an opponent’s kiln and put out a fire.

I wouldn’t be surprised to see Kiln appear as an early preview in the coming months, followed by Forza Horizon 6 in May and then Halo: Campaign Evolved. I keep hearing that both Fable and Gears of War: E-Day are currently targeting a release in the second half of this year. Microsoft is keen to release new Forza, Gears, Halo, and Fable games in 2026 to mark 25 years of Xbox.

I’m always keen to hear from readers, so please drop a comment here, or you can reach me at [email protected] if you want to discuss anything else. If you’ve heard about any of Microsoft’s secret projects, you can reach me via email at [email protected] or speak to me confidentially on the Signal messaging app, where I’m tomwarren.01. I’m also tomwarren on Telegram, if you’d prefer to chat there.

Thanks for subscribing to Notepad.