Microsoft is using Claude Code internally while selling you Copilot

Acccording to The Verge (paywalled), last week, Microsoft told employees across its Windows, Microsoft 365, Teams, Bing, Edge, and Surface divisions to install Claude Code, a competing AI coding tool built by Anthropic. Not as a curiosity and neither a one-off experiment. Instead, it was poised a serious internal tool to be tested alongside the company’s own GitHub Copilot.

The same GitHub Copilot that Microsoft has spent years positioning as the future of software development. The same tool that generated over 40% revenue growth in Q4 2024. The same product they’re actively selling to your company right now.

Are you kidding me?

If you’re wondering what Microsoft knows that you don’t, you’re asking the right question.

But looking deeper in both companies’ business interests, if kind of makes sense long-term if you check their mutual commitments right now:

The setup

The news broke via The Verge on January 22, 2026, in a scoop by Tom Warren that caught the industry off guard. According to the report, Microsoft’s Experiences + Devices division — one of the company’s largest — had been asked to install Claude Code just days earlier.

But the story goes deeper than a single division running an experiment.

Microsoft’s CoreAI team, led by Jay Parikh and reporting directly to CEO Satya Nadella, has been testing Claude Code for months (to no one’s surprise, it is that good). Software engineers across the company are now expected to use both Claude Code and GitHub Copilot, then provide feedback comparing the two (and I have a good hunch on those feedback forms will say).

And here’s the detail that should make Copilot’s product team nervous: even employees without any coding experience — designers, product managers, non-technical staff — are being encouraged to experiment with Claude Code.

“Even employees without any coding experience are being encouraged to experiment with Claude Code.”

Heh.

That last point matters more than it might seem. If non-technical employees can prototype effectively with Claude Code, it suggests a fundamentally different value proposition than Copilot’s autocomplete-style assistance.

It hints at something more autonomous, more capable, more threatening to the status quo.

And from the Microsoft camp:

Microsoft’s communications chief, Frank Shaw, offered the predictable corporate response: “Companies regularly test and trial competing products to gain a better understanding of the market landscape. OpenAI continues to be our primary partner.”

It’s technically true. It’s also not the whole story.

What is Microsoft trying to do here?

Let’s be clear about what Microsoft is doing: they’re hedging.

The company has invested an eyewatering $13 billion+ in OpenAI. They’ve built their AI strategy around that partnership. GitHub Copilot is the flagship product of that strategy. And yet, here they are, spending half a billion dollars annually with Anthropic and telling their own employees to use Claude Code.

This isn’t about “understanding the market landscape.” Rather, it’s about recognising that the AI coding tools market is shifting beneath their feet.

The agentic pivot: The AI coding assistant space is moving from autocomplete to autonomy. Copilot, for all its success, was built for the autocomplete era — suggesting the next line, completing functions, helping developers write code faster. Claude Code and tools like Cursor (now valued at $29.3 billion with $1 billion in annual recurring revenue) represent something different:

AI that can reason about entire codebases, make architectural decisions, and execute multi-step tasks with minimal human intervention.

Microsoft appears to be asking itself a difficult question:

Is Copilot built for the next era, or the last one?

The non-technical signal: When Microsoft encourages designers and product managers to prototype with Claude Code, they’re implicitly acknowledging that this tool might be accessible to people who would never touch Copilot. That’s either a massive expansion of the market or an existential threat to Copilot’s value proposition — or both.

The multi-vendor reality: Enterprise AI procurement is evolving. The days of betting everything on a single provider are ending. Microsoft, of all companies, understands this. They’re not just testing Claude Code — they’re building the infrastructure (Azure quotas, sales incentives, compute commitments) to offer it as a first-class option to enterprise customers.

The bigger picture

For developers watching this unfold, the implications are significant.

If you’re evaluating AI coding tools: Microsoft just did your competitive analysis for you. When a company with $13 billion invested in one AI partner starts seriously testing another, that’s signal. It doesn’t mean Claude Code is definitively better — but it means the gap between these tools is close enough that even the company selling Copilot isn’t sure which will win.

If you’re building on these platforms: The multi-model future is arriving faster than expected. Microsoft’s Azure strategy now explicitly includes Anthropic alongside OpenAI. Expect other cloud providers to follow. The winning strategy isn’t picking a side — it’s building abstractions that let you switch.

If you’re worried about your job: The junior developer employment market is already down roughly 20% from its 2022 peak, according to data from Stanford and Stack Overflow. Tools that let non-technical employees prototype without writing code will accelerate this trend. The question isn’t whether AI will change software development — it’s how quickly, and what roles will remain.

The question isn’t whether AI will change software development — it’s how quickly, and what roles will remain.

There’s a certain irony in Microsoft’s position. They’re the world’s largest seller of AI coding tools, and they’re telling their own people to try the competition. They’re OpenAI’s biggest backer, and they’re building a parallel relationship with OpenAI’s main rival.

They’re selling you Copilot while testing whether Claude Code might be better.

But maybe that’s not irony. Maybe that’s just good business.

Microsoft didn’t become a $3 trillion company by being loyal to its own products when better options emerged. They killed Internet Explorer when Chrome won. They embraced Linux after decades of calling it a cancer. They’re not sentimental about technology, they’re actually pragmatic about it.

The message to developers is clear: if Microsoft isn’t all-in on Copilot, maybe you shouldn’t be either.

The AI coding tools market is still being written. The company that knows the most about this space, because they’re building it, selling it, and now testing the alternatives, is hedging its bets.

Perhaps you should too.

References

• The Verge — Claude Code is suddenly everywhere inside Microsoft — Primary source for internal Microsoft testing details (Tom Warren, Jan 22, 2026)
• Microsoft Official Blog — Microsoft, NVIDIA, and Anthropic announce strategic partnerships — $5B investment and $30B Azure compute commitment (Nov 18, 2025)
• The Information — Microsoft’s Spending on Anthropic AI on Pace to Hit $500 Million — Annual Anthropic spending details (Jan 14, 2026)
• TechCrunch — Microsoft counting Anthropic sales toward Azure quotas — Sales incentive restructuring (Jan 17, 2025)
• Stack Overflow Blog — AI vs Gen Z — Junior developer employment data (Dec 26, 2025)