AI Demand Overwhelms Internal Cloud Plans: Microsoft Turns to AWS to Scale GitHub
Taylor Wilson
Microsoft is buying extra compute from rival Amazon AWS to keep GitHub running, after AI coding tools drove a 14× surge in code commits in a single year — the world's No. 2 cloud vendor going to No. 1 for capacity reveals an AI compute gap far larger than planned.
What exactly hit GitHub?
AI coding tools — software that auto-generates code for developers — collapsed the barrier to writing code. GitHub commits are projected to reach 14 billion in 2026, up from 1 billion in 2025 — roughly a 14× jump in one year.
This means → the data load multiplied more than tenfold in months; the existing infrastructure was never designed for that pace.
The direct fallout: GitHub suffered dozens of major outages in 2026. In plain terms = developers kept finding the platform down.
Why is Microsoft buying compute from a rival?
Microsoft acquired GitHub in 2018 and planned to migrate it fully onto its own Azure cloud by 2027.
AI-driven demand growth outran Azure's expansion schedule, forcing Microsoft to purchase overflow capacity from AWS.
This reflects an awkward reality: Microsoft is competing head-to-head with AWS for global cloud share, yet now relies on that same rival for emergency capacity — not a strategic choice, but a demand-driven stopgap.
What does Microsoft say officially?
A Microsoft spokesperson confirmed GitHub is onboarding multiple cloud providers but declined to comment on Amazon's specific role.
The official line: "The wave of agentic development that began late last year has pushed our infrastructure to its limits."
The stated strategy is to "accelerate migration to Azure while exploring a multi-cloud approach." In plain terms = borrow someone else's servers now, build your own capacity later.
Is buying compute from a competitor unusual?
It is not unique. This month SpaceX and Google disclosed a deal: Google will pay SpaceX $920 million per month for AI compute from October 2026 through June 2029.
Google Cloud itself sold AI compute to Anthropic just two months earlier.
This means → when AI compute is desperately scarce, competitive rivalry does not block deals — whoever has spare capacity becomes the supplier.
What other pressures is GitHub facing?
Insufficient compute is only half the problem; the other half is rivals poaching users — Cursor, Anthropic's Claude Code, and other AI coding tools are eating into GitHub's engineer base.
Per internal meeting audio obtained by Business Insider, a Microsoft executive said late last year that GitHub needs a comprehensive overhaul to meet the competition.
Microsoft projects $190 billion in capital expenditure for 2026, mostly for data-center expansion, but many projects face delays — whether GitHub's compute gap can be filled before Azure's own capacity comes online remains an open question.
Content is for reference only, not financial advice.