Microsoft Leverages AWS to Address GitHub AI Capacity Crunch
Microsoft integrates AWS capacity to stabilize GitHub
Microsoft is adding Amazon Web Services (AWS) capacity to GitHub's infrastructure to prevent service outages caused by an unprecedented surge in AI-driven coding activity. This strategic shift occurs as "agentic development" workflows—where AI agents autonomously generate code, commits, and pull requests—have strained GitHub's infrastructure beyond the limits of its planned migration to Microsoft Azure.
While Microsoft originally intended to move GitHub fully to Azure by 2027, the volume of machine-generated activity has forced the company to adopt a multi-cloud strategy to ensure elasticity and scale. Microsoft has confirmed a broader shift toward multi-cloud infrastructure, though it did not explicitly name AWS in its official statement.
The impact of agentic development on infrastructure
GitHub is experiencing a growth curve that exceeds traditional organic scaling, driven specifically by AI agents that operate at machine speed rather than human speed.
Exponential growth in commits
GitHub COO Kyle Daigle reported that commits were on pace to hit 14 billion in 2026, a massive increase from 1 billion in 2025. This volume creates direct stress on the platform's ability to store code, process pull requests, update search indexes, and trigger automations.
Rapidly scaling capacity requirements
GitHub CTO Vlad Fedorov indicated that the platform's capacity needs shifted dramatically in a short window:
- October 2025: GitHub began a plan to increase capacity by 10X.
- February 2026: The company determined it needed to design for 30X scale.
This acceleration is attributed to agentic development workflows that spiked in late December 2025. Despite these efforts, GitHub's May 2026 availability report noted nine incidents that degraded services, including a major disruption on May 4 caused by a database schema migration that cascaded across pull requests, issues, and Git operations.
Azure capacity constraints and strategic risks
Despite Microsoft's massive capital expenditure—approximately $190 billion planned for calendar-year 2026—Azure's capacity is not an infinite pool. It is shared across OpenAI demand, Microsoft Copilot, security workloads, and external Azure customers.
Infrastructure as a bottleneck
Microsoft CFO Amy Hood stated that the company expects to remain capacity-constrained through 2026. This internal scarcity means that even strategic assets like GitHub must compete for GPU, CPU, and storage resources. The decision to use AWS is an admission that internal demand for AI infrastructure has exceeded Microsoft's own cloud deployment timeline.
The threat of AI-native competitors
Reliability issues have transformed from technical nuisances into product threats. High-profile developers, such as HashiCorp co-founder Mitchell Hashimoto, have publicly criticized GitHub's reliability, stating the platform was "no longer a place for serious work" after experiencing significant downtime.
This instability creates an opening for AI-native developer tools like Cursor and Anthropic's Claude Code. If GitHub cannot maintain a stable control plane for AI-assisted development, it risks losing the high-signal maintainers and enterprise users who form the core of its network effect.
Industry-wide trend of "bridge capacity"
The GitHub-AWS arrangement reflects a broader industry trend where hyperscalers purchase capacity from rivals to meet sudden AI demand.
- Google and SpaceX: Google has reportedly agreed to pay SpaceX $920 million per month from October 2026 through June 2029 for access to compute capacity.
These deals indicate that AI demand is currently outrunning the planning and construction cycles of the world's largest data centers, forcing even the primary cloud providers to seek external "bridge capacity" to maintain service stability.