admin GitHub Copilot gave Microsoft a commanding start in the AI coding race. It was launched before many competitors had mature developer products, and it benefited from GitHub’s position at the centre of global software collaboration. More than 150 million developers use GitHub, while Copilot has become one of Microsoft’s most visible AI products for engineers, businesses and public-sector technology teams.
That advantage now looks less secure. Developers and enterprise customers have grown more vocal about GitHub’s reliability problems after repeated service disruptions affected repositories, pull requests, Actions, Copilot and related developer workflows. For engineering teams working on continuous deployment, security patches and AI-assisted code generation, even short interruptions can slow product releases and expose the risks of depending too heavily on a single development platform.
GitHub’s own service updates have shown the scale of the challenge. During April, Copilot-related services were affected by model request failures, cloud agent disruption and dashboard problems. The company has acknowledged strains across parts of its infrastructure as demand for AI-powered development tools rises sharply. The pressure is particularly acute because AI coding agents require far more compute and orchestration than traditional code hosting or autocomplete features.
Microsoft’s ownership of GitHub was once seen as a strategic masterstroke. The $7.5 billion acquisition in 2018 gave Microsoft direct access to the world’s largest community of developers and helped reposition the company as a partner to open-source software communities after years of distrust. Copilot then turned that asset into a commercial AI wedge, embedding Microsoft deeper into daily coding workflows.
The problem for Microsoft is that developer loyalty can shift quickly when tools fail at crucial moments. Programmers may tolerate occasional outages in a social network or consumer app; they are less forgiving when source-code access, CI/CD pipelines or AI coding assistants break during production deadlines. Frustration has been amplified by questions over Copilot pricing, performance consistency and the growing complexity of enterprise billing for AI usage.
Leadership turnover has added to the uncertainty. Thomas Dohmke announced last year that he would step down as GitHub chief executive, and Microsoft moved GitHub more tightly into its CoreAI organisation rather than preserving the same degree of independence it had after the acquisition. That shift reflects Microsoft’s ambition to make GitHub central to its broader AI platform strategy, but it has also raised concerns among some employees and users about whether GitHub’s developer-first culture is being diluted.
Security has become another source of concern. A compromise linked to a malicious Visual Studio Code extension exposed thousands of internal GitHub repositories, underscoring the growing threat from software supply-chain attacks. GitHub has said customer repositories were not affected, but the episode highlighted a broader vulnerability: the tools developers use to build software have themselves become prime targets for attackers.
Rivals are moving aggressively into this opening. Cursor has gained traction among developers who want an AI-native coding environment built around multi-file editing, project context and fast iteration. Its valuation has surged as investors bet that the next major software company could emerge from the developer productivity market rather than from traditional enterprise applications.
Anthropic’s Claude Code has become another formidable competitor, especially among start-ups and engineering teams working on complex codebases. Its appeal lies in agentic workflows that can reason through architecture, edit files, run tasks and assist with broader software development processes. Microsoft itself has added Anthropic models to GitHub Copilot, a sign that the market is no longer defined by a single model provider or one assistant.
OpenAI is also pushing deeper into software development with Codex-style agents, while Google, Amazon and other cloud providers are tying coding tools to their own infrastructure and enterprise platforms. The result is a fragmented race in which no company can rely only on distribution. Reliability, model quality, workflow integration, security and cost are becoming equally decisive.
Microsoft still retains powerful advantages. GitHub remains deeply embedded in enterprise development, Visual Studio Code is one of the world’s most widely used code editors, Azure provides the cloud backbone for large-scale AI deployment, and Microsoft’s enterprise sales machine can bundle developer tools into broader contracts. Many large companies prefer Copilot because procurement, identity management, compliance and support fit more easily into existing Microsoft arrangements.
Yet the contest has shifted from adoption to trust. Developers want tools that stay online, respect established workflows and produce useful code without adding new operational risk. Enterprise technology chiefs are asking whether AI coding assistants can be governed safely at scale, whether generated code can be audited, and whether agentic tools will create new security and compliance burdens.
