Unprecedented Growth Spikes
GitHub's Chief Operating Officer, Kyle Daigle, has publicly acknowledged the staggering volume of activity the platform is currently managing, largely
driven by the rise of AI agents. This surge represents a growth rate far beyond anything previously experienced. To put it into perspective, while the entire year of 2025 saw one billion commits, GitHub is now processing an astonishing 275 million commits every single week. If this linear growth trajectory continues, the platform is on pace to handle an incredible 14 billion commits this year alone. Furthermore, GitHub Actions has experienced a dramatic escalation, leaping from 500 million minutes per week in 2023 to over 2.1 billion minutes in just one week this current year. When questioned about the colloquial term 'shit code' being used to describe the output of these AI agents, Daigle offered a concise and characteristic response, stating his disinclination to comment given his own history with code creation.
Designed for Humans, Not Bots
A fundamental challenge facing GitHub is the inherent mismatch between its original design and the current wave of AI agents. As pointed out by a software developer, GitHub has been operational since April 10, 2008, a time when autonomous coding agents were not a consideration. The platform's infrastructure, including API rate limits, was conceived with human users in mind, not the constant, high-volume demands of autonomous agents operating around the clock. This foundational disparity is now manifesting in real-time operational issues. GitHub's Chief Technology Officer, Vlad Fedorov, confirmed that agentic development workflows have seen a rapid acceleration since late December 2025. This has concurrently driven repository creation, pull request activity, API usage, and workloads involving large repositories to record highs. Initially, the company had planned for a tenfold increase in capacity back in October 2025. However, by February 2026, it became evident that a thirtyfold expansion of current scale was necessary to accommodate the demand.
Major Incidents Emerge
Two significant incidents have highlighted the strain on GitHub's infrastructure. On April 23, a critical bug in the merge queue caused previously merged code to be silently reverted across 658 repositories and affecting 2,092 pull requests. This was followed by another major disruption on April 27, when a botnet attack overloaded GitHub's Elasticsearch cluster, rendering search-dependent features across pull requests, issues, and projects unusable. While neither incident resulted in data loss, both severely impacted developers' ability to complete and deploy their work. The developer community reacted strongly, with Mitchell Hashimoto, an early GitHub user and developer of Ghostty, announcing his departure due to nearly daily outage disruptions logged over a month. This departure served as a prominent focal point, prompting a public apology from Fedorov and a separate acknowledgment from Daigle, who confirmed the team's focus on infrastructure upgrades amidst escalating usage. Despite these challenges, Daigle offered a sliver of optimism, noting that pull request performance has seen improvements of up to 40 percent in certain instances, with further enhancements planned, though the ultimate impact on user retention remains uncertain.
