Microsoft is significantly expanding its internal use of Anthropic’s Claude Code, tasking thousands of employees within prolific teams to experiment with the AI tool, according to reports. This move comes despite Microsoft heavily marketing and selling GitHub Copilot, its primary coding assistant powered by OpenAI models, to external customers.
The adoption extends beyond the developer division, with Microsoft’s Experiences + Devices unit—which oversees Windows, Microsoft 365, and Surface—now being asked to install and utilize Claude Code. This internal testing phase aims to evaluate Claude Code's efficacy across Microsoft’s core product engineering groups.
Sources indicate that Microsoft engineers are expected to use both Claude Code and GitHub Copilot concurrently, providing direct feedback comparing performance across various coding tasks. Furthermore, the company is encouraging nontechnical staff, such as designers and project managers, to use Claude Code for rapid prototyping.
Microsoft's CoreAI team, the organization's central AI engineering group, has been testing the model for several months, signaling high-level organizational interest in Anthropic’s capabilities. This broad internal validation contrasts with the external narrative where Copilot remains the flagship offering.
This increased reliance on Anthropic is strategic, as Microsoft is already one of the AI firm's top customers, according to The Information. The company is reportedly counting sales of Anthropic models toward Azure quotas, an unusual incentive structure usually reserved for internal or OpenAI products.
While Microsoft communications chief Frank Shaw affirmed OpenAI remains the primary partner for frontier models, the simultaneous, deep integration of Claude Code demonstrates a pragmatic approach to heterogeneous AI toolsets. Companies often trial competing products to fully map the market capabilities, Shaw stated.
If these extensive internal pilots prove successful, observers suggest Microsoft could potentially offer Claude Code directly to its Azure cloud customers in the future. This would mark a significant diversification of the AI coding tools Microsoft offers commercially, moving beyond its established OpenAI relationship.
The widespread testing of Claude Code, especially by nontechnical users, underscores a trend toward making code generation more accessible, potentially reshaping junior developer roles within the organization. This broad adoption places further pressure on the established utility of existing tools like Copilot within the Microsoft ecosystem.
