Agents multiplied Git traffic
GitHub said it has entered a new architectural shift in its Git infrastructure as AI agents enter software development. According to the company, agents commit or checkpoint after nearly every action, turning the completion speed of a single push into a bottleneck.
The numbers show the scale
According to GitHub's data, monthly Git activity rose from 218.2 billion to 473.3 billion events between September 2025 and August 2026. In September alone, 7.38 billion commits were made, more than five times the figure a year earlier. Monthly pushes grew 4.9x, from 0.69 billion to 3.35 billion. GitHub Actions ran 3.26 billion times in September.
Writes are the real challenge
GitHub stresses that scaling reads with caches and replicas is relatively easy, but scaling writes is much harder. Every push must be stored durably and made consistently visible before the next agent or CI job can build on it. Thousands of agents working on their own branches in the same repository converge the write load onto a single point in the architecture.
The scale of repositories
As of August 2026, the busiest repository received roughly a billion requests a month. GitHub says the gap between a typical repository and the busiest ones is far wider than most people expect. The company says it is investing deeply in Git infrastructure to meet these workloads.
