The news
GitHub, owned by Microsoft (MSFT), is rebuilding the storage system behind its Git repositories to handle a surge in activity from AI coding agents. In an engineering blog post published on October 6, 2026, principal software engineer Brian Celenza described replacing Spokes, the architecture GitHub has used to store repositories.
GitHub's figures show how fast load has grown. Between September 2025 and August 2026, Git activity rose from 218.2 billion to 473.3 billion events a month. In September 2026, developers and agents made 7.38 billion commits, more than five times as many as a year earlier. Pushes grew 4.9 times, to 3.35 billion a month, and GitHub Actions ran 3.26 billion times in September, up four times year over year. The busiest single repository received roughly a billion requests in August.
Spokes keeps full copies of each repository on several file servers and uses a three-phase quorum commit, so every replica takes part in each write. GitHub said that creates a tradeoff: adding read capacity slows writes.
The new design writes data once to Azure Blob Storage, moves reads to lightweight compute workers, and shifts maintenance tasks such as compaction and garbage collection into the background. GitHub said the design reached up to 35 times higher write throughput in its benchmarks. The company said developer workflows for branching, review and merge are meant to stay the same.
The Register reported on October 9 that GitHub suffered 10 performance-degrading incidents in April 2026 and 9 in May, and that Celenza did not give a timeline for the migration. GitHub's post says more technical detail will follow in a later post.
The numbers
- Monthly Git events, Sept 2025 to Aug 2026
- 218.2B to 473.3B
- Commits in September 2026
- 7.38 billion (5x+ year over year)
- Monthly pushes
- 3.35 billion (4.9x)
- Write throughput gain in GitHub benchmarks
- Up to 35x
- Performance incidents, April / May 2026 (The Register)
- 10 / 9
Why CEOs should care
For CTOs and engineering leaders, GitHub is often a single point of failure for shipping software. The numbers show agents multiplying the load on it, and The Register's incident count shows that load has already affected reliability. Review what happens to your release process when GitHub is degraded: can builds, deploys and urgent fixes still go out? Consider read-only mirrors of critical repositories and documented fallback steps.
For leaders rolling out coding agents, your own agents add to this load. Agents that commit and push constantly can hit rate limits or slow shared repositories. Ask your platform team how many commits and Actions runs agents generate, and set limits before costs and contention grow. Since Actions usage is often billed, the 4x jump GitHub reports is also a budget signal.
For CIOs, the rebuild is a migration with no public timeline. Ask your GitHub account team about status updates, service-level commitments and whether enterprise customers will get notice before their repositories move to the new storage.
The bigger picture
AI coding agents do not just write code faster; they generate far more commits, pushes and automated runs than human teams. That shifts the bottleneck to shared infrastructure that was designed for human pace. GitHub's decision to rewrite a core system while it keeps running shows how large the change is.
Other developer platforms, from CI services to package registries, face the same pressure, and buyers should expect pricing and limits to change as usage per developer rises.
What’s next
Watch for GitHub's follow-up technical post, any published migration schedule, and whether incident counts on GitHub's status page fall in the coming months. Changes to Actions pricing or rate limits for agent traffic would also signal how GitHub plans to manage the load.
What “Fact-checked” means
Fact-checking means testing a story’s facts against the evidence before it is published. This story went through at least two separate checks before this version was published.
- What we checked
- Its names, figures, dates, job titles, quotes and who said what were checked against the story’s sources, including its main source where it could be opened. The headline was checked for accuracy and overstatement.
- How
- A first check reviewed the whole story. If it passed, a second, skeptical check went back to the sources to look for mistakes in the most important facts. If a check flagged the story, it was edited to fix the problems found, and a separate re-check then reviewed the whole story again.
- Who
- The checks are made by our newsroom, as steps kept separate from the writing, under rules set by our editor, Hussein Mukhtar. A story the checks still flag is held for the editor, who decides whether it is fixed, published or dropped.
- If something is wrong
- “Fact-checked” does not mean error-free. If a material error is found after publication, we correct the story and add a note saying what changed. Report an error
