GitHub Enterprise now includes Copilot for every licensed user, with AI billed by use

Microsoft CEO Satya Nadella and GitHub used Universe on Oct. 28 to announce that every licensed GitHub Enterprise user now gets Copilot as part of the platform, with AI usage metered against an enterprise budget. In a post on X, Nadella wrote: “We’re redefining the SDLC for the agentic era.”

The post points to a broad set of GitHub updates released the same morning. Its excerpt also says Copilot Fusion delivers frontier-level quality “40% faster and at up to 77% lower cost”. The GitHub pages we read do not state the 77% figure, so it should be treated as Microsoft’s own claim for now.

Copilot moves from add-on to platform feature

According to GitHub’s changelog, Copilot is included with GitHub Enterprise for each licensed user. Enterprise account users who do not hold a GitHub Enterprise license can get the same access through Copilot Standalone, priced at $10 per user per month. Both routes carry the same features and models as Copilot Enterprise, but without the monthly AI credits per seat that plan includes.

Instead, AI usage is metered from the first request against the enterprise budget, using existing policies, alerts and attribution. Code completion and next edit suggestions keep working when no budget is left. Copilot Chat, code review and Copilot in the cloud need available budget.

Availability is split. Customers with an account team on GitHub Enterprise Cloud can start now by contacting sales. Self-serve customers cannot yet; GitHub says access will open in stages, with dates promised “in the coming weeks”. Existing Copilot Business and Copilot Enterprise plans stay available and are not changed.

The rest of the Universe package

GitHub’s Universe post, written by Jay Parikh, lists several other items. The Copilot app connects agents and tools across projects and, per GitHub, handles specs, design discussions and telemetry analysis as well as code. Projects, which run agent work across multiple repositories in parallel sessions, are available in the app from today; we covered that in our report on the app’s Projects and computer use mode. Session continuity lets one cloud agent session be followed across the app, VS Code, the CLI and the web. Copilot automations, cloud-run repo workflows triggered manually, by events or on a schedule, can now learn from outcomes and propose improvements for approval. A new Copilot code review Max tier is billed as having the highest F1 score on ReviewBench for high-severity issues.

GitHub also reported infrastructure additions this year of 4 million CPU cores, 160 PB of high-speed storage and 50 Tb/s of network capacity. It is moving to a multi-region deployment in Azure, with completion expected early in 2027, and says it is building a new Git architecture, though it gave no technical detail.

Budgets replace seat credits, and the discount is temporary

The pricing change is the part most likely to touch finance teams. Under the older Copilot Enterprise model, each seat came with monthly AI credits. Under the new Enterprise inclusion, there is no allowance to track or top up, so the enterprise budget becomes the main control. GitHub’s own wording is that with Copilot included, usage against the budget “is the primary thing to manage”. For organizations with many light users, that could mean paying only for what runs; for heavy agent users, it makes spend less predictable than a flat per-seat credit pool.

Model costs are also in play. GitHub says Copilot Fusion, now generally available on all Copilot plans, bills each model it uses at that model’s standard rate, and for a limited time paid plan users get a 30% discount on those model costs. We looked at the routing mechanics in our Fusion report. Fusion has three effort levels (Standard, High, Max) and three workflows: Single, Cascade and Critique.

On benchmarks, GitHub cites the Artificial Analysis Coding Agent Index, where Fusion Max scored 60, within two points of Claude Code with Opus 5.5 Max. Against that setup, GitHub says Fusion Max had about half the average API cost per task, about half the tokens and roughly 40% less time. That is the source of the 40% figure; the “up to 77%” cost claim in the post appears to be a separate, best-case number that GitHub’s pages do not detail. All of these are company-reported results, not independently verified by us.

What the announcements leave open

Several details are missing. GitHub has not given a date for self-serve Enterprise access, nor said how the per-use rates for Chat, code review and cloud agents compare with the credits previously bundled per seat. The Fusion discount has no stated end date. And the headline benchmark compares Fusion with one competing setup, on one index. Teams weighing the switch will want their own cost-per-task data before moving budgets over.

For now, the practical effect is simple: large organizations already on GitHub Enterprise can turn on Copilot without buying a separate subscription, and pay as agents and chat are used. Microsoft’s pitch, in Nadella’s words, is a software development lifecycle rebuilt around agents; the budget line is where that pitch meets procurement.