GitHub added AI credit pools for cost centers on July 2, giving Copilot Business and Copilot Enterprise admins a way to cap how much of an enterprise’s monthly included AI credits a cost center can use.
The control is available through the REST API today. GitHub says management in the cost center settings UI is coming later.
The change matters because Copilot licenses pool included monthly AI credits across an enterprise. Without a cap, one cost center can consume included credits that were effectively funded by another group’s licenses before any metered spending begins.
The cap applies before overage budgets
GitHub draws a useful line between two controls. An AI credit pool governs the included usage pool. A cost center budget governs the metered phase after the included pool is exhausted.
That means the new control is not just another budget ceiling. It changes who can draw from shared included credits before the organization starts paying for additional usage.
GitHub says the pool limit is calculated automatically from the Copilot Business and Copilot Enterprise licenses assigned to that cost center. Admins do not set the number manually. When a cost center reaches its cap, the admin can block further included usage or let the cost center continue as additional spend if the enterprise allows overages.
AI spend controls are becoming operational controls
This follows a June 30 GitHub update that added per-user AI credit budgets for cost centers. Together with per-user credit metrics, managed settings, model policies, and agent-session visibility, Copilot is turning into an operating surface for enterprise AI governance.
That is the real story. Enterprise AI coding tools are no longer governed only by seat assignment. They now need controls for model access, feature availability, agent behavior, usage reporting, included credits, and overage spend.
For finance teams, the new pool reduces chargeback ambiguity. For platform teams, it gives a way to prevent one heavy agent workflow from consuming shared credits before other teams get value from the licenses they funded.
The UI gap matters
GitHub says the control is available through the REST API today, with UI management coming soon. That makes the first version more useful to platform teams that already manage GitHub Enterprise Cloud through automation.
It is less friendly for admins expecting a point-and-click budget workflow. Until the UI lands, organizations should treat this as an API-managed control and document who owns changes.
