Microsoft shifts to usage-based AI billing across enterprise software spurring new revenue streams

Microsoft is integrating metered AI charges with traditional software licenses, blending predictable entry prices with usage-based costs in products like GitHub Copilot and Microsoft 365 Copilot, signalling a strategic move that could reshape enterprise AI pricing models.

Microsoft is increasingly blending fixed software licences with usage-based AI charges, a shift that gives customers predictable entry pricing while allowing bills to rise as they use more computing capacity. The change is most visible across GitHub Copilot, Microsoft 365 Copilot and Dynamics 365, where the company is pairing per-user subscriptions with metered credit consumption for heavier workloads. TechInformed reported that the model is already feeding through to Microsoft’s results, even though the company has not disclosed how much of the new revenue comes from consumption rather than seats.

GitHub Copilot is the clearest example. According to GitHub’s billing documentation, Business still costs $19 a user a month and Enterprise $39, but those plans now include AI credit allowances rather than unlimited use. Once the shared pool is exhausted, organisations can buy more capacity at published rates unless an administrator sets a spending cap. GitHub says the system measures input, output and cached tokens, reflecting the longer coding sessions and agent-style tasks now common in developer workflows.

Microsoft has also extended usage billing into Microsoft 365 Copilot and Dynamics 365. Microsoft’s own documentation says Copilot Credits are designed to complement fixed subscriptions, while Dynamics 365 Sales now supports both prepaid capacity and pay-as-you-go billing for Copilot and agent capabilities. In both cases, administrators can monitor usage, set budgets and impose hard limits, which is meant to keep AI spending closer to procurement and finance controls than traditional cloud add-ons.

The company has not given a full revenue split, but the signs of adoption are visible. TechInformed reported that Dynamics 365 Customer Service credit consumption rose fourfold quarter on quarter, while GitHub Copilot generated what Microsoft described as significant consumption revenue after its June billing change. Microsoft also said it now has more than 30 million paid Microsoft 365 Copilot seats, with net additions more than doubling from the previous quarter, and that GitHub continues to add Business and Enterprise customers.

Microsoft chief executive Satya Nadella has framed the approach as a compromise between budget discipline and flexibility. Customers want predictable spend, especially for procurement, he said, while also leaving room for extra use above the included allowance to be billed separately. That position mirrors a wider shift in enterprise AI pricing, as firms search for models that match volatile model costs with usage that can vary sharply by user, team and task.

The industry is still settling on workable rules. A June paper from the FinOps Foundation said token cost and consumption management were now among the biggest pain points for finance and operations teams, citing opaque billing and the difficulty of allocating AI usage cleanly across departments. Salesforce has taken a similar seat-plus-usage path with Agentforce, although it also offers separate pricing structures. Microsoft, for its part, has not said how often customers exhaust their included credits or whether hybrid billing raises total spend versus fixed licensing alone.