Microsoft launches autonomous Copilot agent, shifts to usage-based pricing

The buried story here is not the modular rebrand but the billing change: moving Autopilot and Code tiers to consumption-based pricing effectively transfers infrastructure cost risk onto enterprise customers, a meaningful shift in who absorbs the expense of running long-horizon agentic tasks at scale.

This is largely disconnected from recent activity in our archive, as Modelwire has no prior coverage to anchor against. That said, this story belongs to a well-established pattern across the broader industry: hyperscalers initially subsidizing AI access to drive adoption, then repricing once stickiness is established. Microsoft is not the first mover here. The more interesting competitive question is whether Google Workspace and Salesforce Agentforce, both of which have experimented with similar tiered structures, will accelerate their own consumption models in response, or hold flat-rate pricing as a differentiator to pull enterprise accounts.

Watch whether Microsoft publishes per-task cost benchmarks for Autopilot within the next two quarters. If they do not, it likely signals that average consumption costs are high enough to deter transparent comparison shopping against competitors.