GitHub Expands Infrastructure with Four Million Cores to Support AI Agents
Massive compute upgrade addresses service outages caused by surging AI-driven code commit volumes.
1Infrastructure Under Pressure
GitHub recorded 7.4 billion code commits in September 2026, marking a fivefold increase over the previous year. This rapid surge, driven by developers deploying AI agents, overwhelmed existing systems. Consequently, the platform experienced frequent service outages, with availability dropping to approximately 84 percent on May 7, forcing an urgent infrastructure overhaul.
2Scaling Compute Capacity
To stabilize operations, GitHub is adding four million CPU cores and 160 petabytes of high-speed storage to its roadmap. These resources specifically target GitHub Actions, the company's CI/CD platform, which struggled under the recent traffic load. This expansion aims to provide the necessary headroom for sustained growth in developer activity.
3Architectural Redesign Strategy
GitHub is actively decoupling compute and storage layers within its technical architecture. This fundamental redesign is projected to enable 35 times more throughput across the platform. Furthermore, the company is migrating to a multi-region deployment within Azure, a transition expected to reach completion early next year to ensure global service reliability.
4Introducing Copilot Fusion
GitHub announced the general availability of Copilot Fusion, formerly known as HydraFusion. This model routing service dynamically selects the optimal AI model for specific tasks, such as debugging or code generation. By automating model selection, it improves task-completion rates while significantly reducing token consumption costs for developers using the platform.
5Performance and Efficiency Gains
According to the Artificial Analysis Coding Agent Index, the Fusion Max service demonstrates high efficiency. It scored within two points of Opus 5.5 Max while operating at half the cost. Additionally, the service completes tasks in 40 percent less time, proving that intelligent model routing can optimize both speed and expenditure.
6Expanding Enterprise Accessibility
GitHub is broadening access to its Copilot suite for all Enterprise users without requiring additional per-seat purchases. While the base access is now included, customers remain responsible for token usage costs. This move aims to increase Copilot adoption across workplaces by removing barriers to entry for existing enterprise-level development teams.
7Enhanced Security Features
The platform is launching AI Scans to identify vulnerabilities in code that rule-based systems often miss. This includes scanning merged code and utilizing reachability analysis to prioritize exploitable issues. Additionally, a new package firewall will enforce enterprise policy settings before fetching npm packages, further hardening the software supply chain for users.
8Future Operational Stability
These infrastructure investments represent a critical pivot for GitHub as it transitions from a code hosting service to an AI-agent-driven ecosystem. By separating compute from storage and optimizing model routing, the company seeks to move past the service instability that plagued its operations throughout 2026, ensuring a more resilient developer experience.