Microsoft owns GitHub and Azure, but the AI coding surge has pushed it toward Amazon anyway. That is the clearest sign yet that agentic software development is outrunning even the biggest cloud plans.
There is a particular kind of embarrassment in having to ask your biggest cloud rival for help. Microsoft owns GitHub, the developer platform it bought in 2018. It also owns Azure, the cloud business it has spent years presenting as the obvious home for serious AI workloads. Yet Business Insider reported on June 16 that Microsoft is adding extra GitHub computing capacity through Amazon Web Services after AI-driven coding activity strained GitHub's infrastructure.
The arrangement is not a small footnote in the cloud wars. According to Business Insider, Microsoft had planned to move GitHub entirely onto Azure by 2027, but the latest surge in AI-assisted development has forced a more practical answer: keep accelerating the Azure migration, while using multiple cloud providers where GitHub needs more elasticity. Microsoft confirmed to the outlet that GitHub is tapping multiple cloud providers, although it declined to name Amazon directly. Amazon said it does not comment on individual customers.
The most important number in the story is not a data center count or a capital spending target. It is GitHub's own code activity. Business Insider cited GitHub Chief Operating Officer Kyle Daigle, who wrote in April that GitHub commits were on pace to reach 14 billion in 2026, up from 1 billion in 2025. That is not normal platform growth. It is a step change in how software is being produced, and it explains why a migration plan drawn up before the agentic coding boom could suddenly look too small.
GitHub has already felt the strain in public. Business Insider reported that the platform has suffered dozens of major outages in 2026, and GitHub's own May availability report said it had nine incidents that caused degraded performance across GitHub services that month. For developers, this is not an abstract infrastructure story. When GitHub slows down or blocks work, repositories, pull requests, CI pipelines, code reviews, and AI agents all stack up behind the same bottleneck.
The shift from autocomplete to agents is the real pressure point. A simple Copilot suggestion in an editor is a bounded request. A coding agent can read a repository, plan a task, make changes, run tests, revise its work, and open a pull request. GitHub's product has moved in that direction fast. In February, The Verge reported that GitHub added Anthropic's Claude and OpenAI's Codex coding agents to GitHub, GitHub Mobile, and Visual Studio Code through Agent HQ for Copilot Pro Plus and Enterprise users.
That makes the AWS deal less strange than it first sounds. GitHub is no longer only hosting code written by humans. It is becoming one of the places where AI systems act on code directly, and each of those actions adds compute, storage, networking, and reliability demands. Microsoft can prefer Azure strategically and still need AWS operationally. The servers do not care about corporate positioning.
Copilot's pricing changed for the same reason
The economics are moving in the same direction as the infrastructure. GitHub said in an April 27 blog post by Chief Product Officer Mario Rodriguez that all Copilot plans would move to usage-based billing on June 1, 2026. The company replaced premium request units with GitHub AI Credits, calculated from token consumption, including input, output, and cached tokens. Base plan prices stayed the same, with Copilot Pro at $10 a month, Business at $19 per user a month, and Enterprise at $39 per user a month.
GitHub's explanation was blunt. Copilot is not the same product it was a year earlier. It has become an agentic platform capable of long, multi-step coding sessions across repositories, and GitHub said the old request model was no longer sustainable when a quick chat question and a multi-hour autonomous session could be priced the same way.
The user reaction has been sharp. Business Insider reported on June 3 that some Copilot users posted screenshots showing projected bills hundreds of dollars above previous months, with one Reddit user saying a $39 monthly plan could turn into an estimated $847 bill. Some users will change behavior, some will set stricter budgets, and some will look harder at Cursor, Claude Code, Amazon Q Developer, or cheaper model settings. The days when AI coding tools could feel unlimited for a flat fee are ending.
Microsoft is still spending as if it intends to own this layer. Business Insider noted that the company recently projected $190 billion in 2026 calendar-year capital expenditures, largely for data center capacity. But capital spending and usable capacity are not the same thing. Data centers take time, power, land, chips, and network buildout. GitHub needs capacity while developers are already using the product, not after the next wave of campuses comes online.
That is what makes this episode useful beyond Microsoft. GitHub should have been one of the best-positioned companies in the world to see AI coding demand coming. It owns the platform, sells the assistant, watches the workflows, and sits inside Microsoft. If even GitHub has to bring in a rival cloud to keep pace, other companies building agentic products should be careful about the forecasts sitting in their own infrastructure decks.
Also read: Coinbase is betting its Base App can do what no crypto company has done before • Cohere is turning the Anthropic export ban into its fastest customer acquisition moment yet • The Pentagon just ran the fastest enterprise AI rollout in history and it is only half done
