Skip to content
News Incident

Microsoft pulled in AWS capacity to prop up GitHub after AI-driven outages overwhelmed the platform

· by Pondero Newsdesk

The short version

Business Insider reported that Microsoft provisioned compute from Amazon Web Services to support GitHub after a surge in AI coding agents pushed the platform to 275 million commits per week and triggered dozens of service incidents.

Microsoft pulled in AWS capacity to prop up GitHub after AI-driven outages overwhelmed the platform

Microsoft provisioned compute from Amazon Web Services to support GitHub after a wave of AI coding agents pushed the platform past its infrastructure limits, Business Insider reported on June 16, 2026, citing two people familiar with the plans. The arrangement is described as an operational stop-gap, not a strategic shift away from Microsoft's own Azure cloud.

What happened

GitHub has historically run much of its own data center infrastructure. It was on track to complete a full migration to Microsoft Azure by 2027. That timeline ran into a demand curve that no migration plan anticipated.

GitHub COO Kyle Daigle described the scale in an April 2026 post: "There were 1 billion commits in 2025. Now, it's 275 million per week, on pace for 14 billion this year if growth remains linear (spoiler: it won't)." GitHub Actions compute minutes followed the same curve, reaching 2.1 billion in a single week in early 2026, up from 500 million per week in 2023 and 1 billion per week as recently as 2025.

The driver is autonomous AI agents. Pull requests opened by AI agents grew from roughly 4 million in September 2025 to more than 17 million in March 2026, per The Information, as cited by Let's Data Science. Those agents do not browse GitHub the way a human developer does. They clone repositories, push branches, open PRs, and run CI pipelines at machine speed, around the clock.

The strain showed up in the status page. The Register reported nine service incidents in May 2026 and ten in April, with overall GitHub availability sitting at roughly 88.4 percent in June, per Let's Data Science's aggregation of that reporting.

A Microsoft spokesperson confirmed to Business Insider that GitHub uses multiple cloud providers. The company declined to comment specifically on Amazon's involvement.

Why it matters

Microsoft using a direct competitor's cloud for a flagship product is notable on its own. AWS and Azure compete for the same enterprise customers, and Microsoft has spent years positioning Azure as the default cloud for everything it touches. That the AWS arrangement was deemed necessary for GitHub tells developers something concrete about the gap between the Azure migration timeline and the actual pace of demand growth.

For developers and AI engineering teams, the practical concern is reliability. An 88.4 percent availability figure over a month means roughly 3.5 days of degraded or unavailable service. Teams running automated agent workflows that depend on GitHub's API or GitHub Actions infrastructure are absorbing those outages in failed jobs, retried pipelines, and delayed deployments.

The underlying math also signals a structural shift. AI agents are not occasional users of GitHub. They are becoming the majority of platform activity on some metrics. Infrastructure that was sized for human-scale usage now faces machine-scale demand, and the provisioning lag creates a reliability deficit until Azure capacity catches up.

Context

Daigle noted in April that his team was "pushing incredibly hard on more CPUs, scaling services, and strengthening GitHub's core features." AWS itself has flagged its own capacity constraints in 2026. The GitHub arrangement suggests that even with Microsoft's aggressive Azure buildout underway, specific product layers are hitting shortfalls before the broader infrastructure arrives.

What to watch next

Two near-term signals are worth tracking. First, whether GitHub's public status page shows a recovery in availability as the additional capacity takes effect. Second, whether GitHub introduces pricing or rate-limit changes targeting AI agent workloads, which would indicate how the company intends to manage demand commercially rather than purely through capacity additions.

The broader question is whether the Azure migration accelerates on a revised timeline or whether multi-cloud becomes a durable architectural choice for GitHub rather than a temporary fix.

Sources