At GitHub Universe 2026, GitHub announced that autonomous AI agents now generate over 40% of all code change pull requests submitted across the platform, establishing agentic coding as the primary development paradigm.

GitHub Universe 2026 Keynote Shock: 40% of All Pull Requests Generated by AI, Triggering Infrastructure Surge — Computer monitor in a modern development workstation displaying extensive source code and automated Git branch commit logs.

Thomas Dohmke, CEO of GitHub, revealed that AI-initiated pull requests surged from 17% earlier this year to 40.2% by September, propelled by developers assigning issue tickets directly to autonomous coding agents.

This explosive adoption drove GitHub's monthly commit volume to 7.4 billion—a sixfold increase year-over-year. To support the unprecedented surge in Git network traffic and automated CI/CD builds, GitHub emergency-provisioned 4 million CPU cores and 160 petabytes of storage over the past twelve months.

The keynote also showcased multi-agent orchestration frameworks where specialized autonomous agents collaborate across issue diagnosis, code refactoring, security patching, and automated documentation.

Dohmke asserted that the primary software bottleneck has shifted decisively from manual typing to architecture validation and code review, committing GitHub to building hyper-scale infrastructure for over 100 million developers collaborating with billions of AI agents.