GitHub routes traffic through AWS as AI coding agents overwhelm capacity and break enterprise SLAs

Microsoft confirmed on June 16, 2026 that it is routing some GitHub traffic through Amazon Web Services, its largest cloud rival, after a surge in AI coding agent activity pushed the platform past the reliability thresholds it guarantees to enterprise customers. The confirmation is the latest turn in reliability problems GitHub has been working through for much of 2026.

GitHub logged nine service-degrading incidents in May 2026 alone, and June availability is tracking well below the 99.9% "three nines" standard. GitHub Chief Technology Officer Vlad Fedorov acknowledged the platform also failed in February and March 2026.

The load is steep. GitHub Chief Operating Officer Kyle Daigle said the platform processed 275 million commits per week in April, putting 2026 on pace for 14 billion total, up from 1 billion across all of 2025. GitHub Actions compute minutes grew from 500 million per week in 2023 to 2.1 billion in a single week in early 2026. AI agent-opened pull requests rose from about 4 million in September 2025 to more than 17 million in March 2026, a 325% increase in six months, GitHub said.

GitHub began a plan in October 2025 to increase capacity tenfold, then revised the target to 30 times by February 2026 as agentic usage grew faster than its models predicted, Fedorov said. The root cause is architectural coupling in a near-two-million-line Ruby on Rails monolith built in 2008, where one localized failure can cascade across all services. Planned fixes include moving performance-critical paths from Ruby to Go, adding load-shedding for misbehaving clients and a multi-cloud strategy.

Microsoft said it is "both accelerating our move to Azure and continuing to explore a multi-cloud strategy." As of the May 2026 report, 40% of GitHub monolith traffic was served from Azure, up from 8% in February. The scale and duration of the AWS routing arrangement were not detailed.