GitHub's chief technology officer, Vladimir Fedorov, has offered a fuller account of the outage that hit the platform on 17 August 2026, acknowledging that the company "let down" developers relying on it to ship software. The incident, which lasted almost eight hours, disrupted Actions, pull requests, issues, Copilot and various APIs after the platform failed to scale with demand, marking the second such failure in a month following an earlier Actions outage on 6 August. Fedorov's comments underline growing concern that GitHub's infrastructure has not kept pace with its explosive growth in usage, despite the company having flagged reliability problems as far back as April.

GitHub said monthly commits have risen from 1.4 billion in April to 2.9 billion now, alongside 24 million new repositories and 130 million merged pull requests each month, with Microsoft Azure handling around 58 per cent of platform load and half of all Git operations. Fedorov stressed that neither outage stemmed from a code or configuration change, pointing instead to underlying scaling weaknesses such as retry storms and misconfigured limits. GitHub says it is accelerating migration to Azure, isolating critical systems to limit the blast radius of failures, tightening retry limits, and working towards a new architecture that scales read capacity linearly with the number of readers, starting with its largest monorepos.