Table of Contents
GitHub Copilot sign-ups are officially suspended, leaving developers locked out as Microsoft battles a massive AI capacity crunch. The code-hosting giant has paused new subscriptions for its Copilot Pro, Pro+, and Student plans to prevent severe service degradation for existing users. This sudden freeze highlights the growing infrastructure strain caused by advanced AI coding assistants.
According to Joe Binder, GitHub's VP of Product, the rapid rise of "agentic workflows" has fundamentally altered the platform's compute demands. Long-running, parallelized AI sessions are now consuming significantly more resources than the original subscription models were built to handle. Without immediate intervention, the sheer volume of automated tasks would degrade service reliability across the board.
This bottleneck is not isolated to Microsoft. A recent surge in demand for AI tools like OpenClaw has caught infrastructure providers off guard across the industry. Google recently enacted a similar policy for its Gemini CLI and Code Assist tools, while OpenAI implemented its own usage balancing earlier this month. Even major cloud providers like AWS and Microsoft Azure are struggling to keep pace with the proliferation of autonomous AI workloads.
New Usage Limits and Model Changes
For existing users, GitHub is tightening the reins to cut costs and realign its service. The platform is enforcing stricter session and weekly limits to control long-trajectory requests that result in prohibitively high compute costs. Exceeding a session limit will now force users to wait until their usage window resets before Copilot can be used again.
Furthermore, GitHub is shifting away from its flat-rate, per request billing model toward token-based consumption. Complex queries were previously costing Microsoft more to process than they charged. As part of this transition, Anthropic's Opus 4.5 and 4.6 models are being removed from Pro+ subscriptions.
They are being replaced by the newly launched Opus 4.7, which is available to Pro+, Teams, and Enterprise customers with a 7.5x premium request multiplier until April 30th. While Opus 4.7 performs better, it consumes 20 to 40 percent more tokens. Existing Copilot Pro and Pro+ subscribers who are dissatisfied with these changes have until May 20th to request a refund, a move that has already sparked heavy debate in the GitHub Community forum. The free tier remains available for now.
The End of the Flat-Rate AI Era
The suspension of GitHub Copilot sign-ups signals a harsh reality check for the generative AI industry: the "all-you-can-eat" subscription model is no longer sustainable for autonomous coding agents. When AI was just auto-completing single lines of code, flat-rate pricing made sense. Now that developers are deploying agents to refactor entire codebases in parallel, compute costs have skyrocketed beyond what a standard monthly fee can cover.
By forcing a shift toward token-based billing and introducing heavy multipliers for premium models like Opus 4.7, Microsoft is actively trying to train developers to be more efficient with their prompts. This capacity crunch will likely force competitors to follow suit, meaning the days of unlimited, unmetered AI assistance are rapidly coming to a close.
