Developers love Claude Code, but Microsoft’s reach gives Copilot the enterprise edge

AI coding tools are now part of everyday software development for many engineers. But new data highlights a gap between developer preferences and the tools that gain traction inside large companies.

A survey published by The Pragmatic Engineer finds that Claude Code is the most-used AI coding tool among respondents, overtaking GitHub Copilot and other competitors in just 9 months since launch. Yet the same data points to a different trend inside large organisations, where Copilot appears to gain ground as company size increases.

The pattern reflects a familiar dynamic in enterprise software: developer enthusiasm and enterprise deployment rarely move in lockstep.

Developers embrace AI coding tools

The survey shows strong adoption of AI coding tools, with 95% of respondents confirming that they use AI tools “at least” weekly, while three quarters report that AI is involved in at least half of their development work.

Agent-style tools are also becoming common. Some 55% of respondents say they regularly use AI agents, often to help debug code, explore unfamiliar codebases, or automate other parts of the development process.

Within that group, Claude Code appears frequently. The tool launched into general availability in May 2025, and has snowballed among developers experimenting with newer approaches to AI-assisted coding.

The report also asked respondents which tools they “love” most. Claude Code ranked highest by a wide margin, with nearly half of respondents naming it as a favourite.

That momentum reflects a broader shift in how developers interact with AI systems. Earlier AI-enabled tools focused largely on autocomplete-style suggestions inside an editor. Newer tools increasingly aim to carry out larger tasks across multiple files or entire repositories.

The enterprise effect on AI tool use

However, the survey paints a different picture once company size enters the frame.

Claude Code appears most frequently among developers at smaller organisations. But usage trends shift steadily as companies grow larger. Among respondents working at organisations with more than 10,000 employees, Copilot becomes the most commonly reported tool.

One obvious explanation may have more to do with distribution than any technical rationale.

Microsoft owns large parts of the developer toolchain, including GitHub and widely used development environments. When AI features are shoehorned directly inside those tools, companies may find it easier to adopt them across large engineering teams.

This is a familiar pattern in enterprise software, where Microsoft’s reach often shapes how tools spread inside organisations. The company has frequently bundled newer products alongside its widely used productivity suite, helping them gain rapid traction in business environments.

Slack, for example, complained to regulators that Microsoft bundling Teams into Office gave the product a distribution advantage over competing workplace tools. Zoom has also expressed similar concerns, over how Microsoft was using its dominance in workplace software to steer customers toward its own collaboration tools.

Copilot may benefit from a similar effect. Companies already using GitHub or Microsoft development environments may encounter Copilot as a built-in feature rather than a new product requiring separate procurement.

For large organisations managing thousands of developers, that convenience can matter a lot.

Claude Code and GitHub Copilot serve different roles

Another point worth noting is that Claude Code and Copilot are not exactly the same category of tool.

Copilot largely functions as an assistant embedded inside development environments (though it is becoming more agentic), suggesting code as engineers type. Claude Code, by contrast, is closer to an agent-style tool that can examine repositories and perform multi-step coding tasks.

That difference may influence how quickly enterprises adopt each system.

Large companies often introduce new developer tools gradually, especially when those tools can run commands or modify multiple parts of a codebase. Systems that remain tightly integrated with existing development environments may face fewer barriers to adoption.

If so, the survey may reflect different stages of the same transition. Developers experiment first with newer tools, while enterprise adoption follows through existing platforms.

The limits of Survey methodology

The survey itself carries some caveats. Notably, it’s based on 906 respondents, drawn from The Pragmatic Engineer’s readership.

That audience tends to include experienced engineers and engineering leaders working in technology-focused companies. The report notes that the median respondent has more than a decade of professional experience.

Because participation is voluntary and drawn from a specific audience, the results should be interpreted as a snapshot of developer sentiment rather than a comprehensive picture of global software development.

The enterprise segment in particular becomes relatively small once respondents are divided by company size. In the survey, companies with more than 10,000 employees account for roughly 16% of the 906 respondents — around 140 developers. Once that group is further broken down by tool usage, the sample becomes even smaller. That makes the survey better suited to identifying broad trends than measuring precise adoption rates.

Still, the results capture an important moment in how software is built, and it largely backs up what much of the industry intuitively thinks already.

Microsoft’s distribution advantage remains powerful, but it doesn’t automatically translate into developer enthusiasm, and as recent reports have noted, the company faces an uphill climb turning its enterprise dominance into widespread adoption of its Copilot AI tools