GitHub Availability Degraded Due to Surge in AI Coding

GitHub, the code-sharing platform under Microsoft, is struggling to cope with a traffic surge driven by AI-assisted coding and agentic development workflows. According to a report by The Register, the platform has faced declining availability over the past several months. While efforts to expand capacity and migrate to Azure infrastructure are ongoing, reliability improvements have not kept pace.

The Reality of the Traffic Surge

According to GitHub’s Availability Report for May 2026, there were nine incidents that degraded service quality. While this is a decrease of one from April’s ten incidents, it remains at a high level.

The primary driver of the traffic increase is the widespread adoption of the AI assistant “GitHub Copilot” and the rapid expansion of agentic development workflows that accompany it. In 2025, the platform processed 1 billion commits per year, but by 2026, that figure jumped to 1.4 billion commits per month. The more developers use Copilot, the more the frequency of pull requests and commits grows exponentially.

Jakub Oleksy, SVP at GitHub, stated, “We are making structural changes to permanently eliminate failure modes,” adding, “We have work to do. We want to make GitHub reliable when you need it.”

This situation presents an ironic picture. The more Microsoft actively promotes Copilot as an AI feature and encourages developers to use AI-assisted coding, the more strain is placed on its own platform. In a sense, the platform has fallen into a contradictory state where “the spread of AI is straining its own infrastructure.”

Revised Capacity Expansion Plans

GitHub’s capacity planning has failed to keep up with the surge in demand. As of October 2025, the plan was to expand capacity tenfold. However, by February 2026, it became clear that a 30-fold expansion was necessary. This shows that demand forecasts completely failed to keep up with actual traffic growth rates.

Behind this lies a change in the quality and quantity of commits and pull requests generated by AI agents. Compared to code written by human developers, AI agents tend to generate large volumes of changes continuously. As a result, the write load on Git repositories has significantly exceeded previous assumptions.

While Microsoft is also focusing on developing LLM-integrated tools such as the Microsoft MarkItDown Markdown Converter for LLMs, strengthening the foundation of its own platform has not kept pace.

Progress and Challenges of Azure Migration

To address the availability issues, GitHub is accelerating the migration of workloads to Azure. According to Oleksy SVP, the percentage of monolith (single architecture) traffic handled by Azure increased from 8% in February to 40%. Git traffic is now 30% on Azure, and repository replication is 99% on Azure.

He stated, “We have more than doubled effective capacity in four months.” The company is also working to isolate primary database clusters and move user authentication and authorization to separate domains to prevent cascading failures.

However, these measures have not yet proven sufficiently effective. Part of the problem is that the target Azure infrastructure itself is facing capacity issues. The risk of cloud dependency has become apparent, as Azure’s resource constraints delay GitHub’s recovery.

As demonstrated with the announcement of Microsoft’s Solara with an AI Agent-Dedicated OS, Microsoft is pushing forward with new platform strategies for AI agents. However, ensuring the reliability of the existing GitHub infrastructure remains a pressing issue.

Current Status Shown by Unofficial Data

Separate from official incident reports, the unofficial project “The Missing GitHub Status Page” reports more severe numbers. This project tracks GitHub service issues and counted 12 incidents in May, calculating an uptime of 87.26% over the past 90 days.

By month, availability was 78.33% in April, 93.86% in May, and 88.39% in June. These are low figures for a major cloud service and are particularly serious for organizations that rely on GitHub as the core of their development workflow. A month with uptime below 80% means more than six days of downtime.

Cost Adjustments and Future Outlook

In addition to availability issues, GitHub is also facing difficult decisions on costs. It temporarily suspended new subscriptions for Copilot and adjusted pricing to respond to policy changes by model providers. This indicates that the processing costs of AI services are squeezing revenue.

Microsoft CEO Satya Nadella previously denied a document about the company’s AI dependency strategy, but the current state of GitHub highlights the reality that “increasing dependence on AI is imposing a severe load on the company’s own infrastructure.”

Looking ahead, Oleksy SVP has promised “structural changes to permanently eliminate failure modes.” However, given that the pace of demand growth clearly exceeds supply, a short-term solution appears difficult. Particularly as automation of development by AI agents advances, traffic is likely to increase further.

Editorial Opinion

In the short term, GitHub’s availability issues will become a serious bottleneck for development teams using AI agents. CI/CD pipeline outages and pull request delays directly impact development velocity. Companies should consider diversifying to alternative platforms such as GitLab or Bitbucket. For mission-critical repositories in particular, hosting diversification as part of a multi-cloud strategy is unavoidable.

From a long-term perspective, GitHub’s reliability problems could trigger a tectonic shift in the developer tools market. The traffic patterns generated by AI agents are fundamentally different from those of human developers. If recognition spreads that concentrating on a single platform like GitHub carries risks, it could foster the rise of emerging services leveraging distributed version control and edge computing.

From an editorial standpoint, we note that the paradoxical situation—where the more AI proliferates, the greater the infrastructure demand—is emerging as a common challenge for SaaS platforms as a whole. GitHub’s problem is not merely a capacity issue for one company; it calls into question the very design philosophy of cloud services in the AI era.

References

• The Register: Holy git! Microsoft code-sharing site suffers downtime, despite move to Azure — Published 2026-06-12
• GitHub Availability Report (May 2026) — GitHub Official Status Page