Usage-Based Copilot Pricing Turned On the Enterprise Tap
GitHub Copilot’s usage-based pricing is a billing model where companies pay for AI coding based on actual consumption of tokens and credits, rather than a fixed per-seat allowance, aligning AI coding tool costs with real developer activity and infrastructure demand across an organization in a more granular, metered way. GitHub switched Copilot to this usage-based billing model on June 1, moving away from a flat-rate per user for a fixed number of requests. Within weeks, CTO Vladimir Fedorov told employees that June was by far GitHub’s best month ever, directly tying the record performance to the new pricing. That is the headline: the meter changed, and enterprise AI adoption surged. This is not a minor tweak to GitHub Copilot pricing; it is the business model catching up with how companies now treat AI coding tools—as core infrastructure, not a side experiment.
From Flat Seats to AI Credits: Why the Old Model Broke
The old Copilot setup tried to treat all AI usage the same, and that broke as soon as agents started doing real work. GitHub’s previous premium request unit system meant a quick chat and a multi-hour autonomous coding session could cost a user the same amount, a mismatch that GitHub’s chief product officer flagged back in April. Once AI coding becomes daily infrastructure, flat-rate plans stop matching the cost or the demand. Heavy users effectively ran without limits while someone else picked up the inference bill. The new GitHub Copilot pricing introduces GitHub AI Credits consumed by token usage across input, output and cached tokens, with different costs depending on the model chosen. Copilot Business is listed at USD 19 (approx. RM88) per user per month with USD 19 (approx. RM88) in AI Credits, while Copilot Enterprise is USD 39 (approx. RM181) with USD 39 (approx. RM181) in credits. In other words, GitHub finally put a rational meter on AI.
What the Meter Means for Developers and Finance Teams
Usage-based billing is a double-edged upgrade for ordinary users. Some developers who treated Copilot like an all-you-can-eat buffet are now seeing projected monthly bills far above what they used to pay, with one Reddit user reporting an estimate of USD 847 (approx. RM3,930) for the next month. The subsidy phase is over; AI coding tool costs now follow actual workload. At the same time, GitHub is pooling included usage across an organization so unused credits from one developer can offset heavy usage by another, a design clearly aimed at enterprise AI adoption. Copilot Business still includes USD 19 (approx. RM88) in credits, while Enterprise includes USD 39 (approx. RM181), plus temporary extra credits through August. A USD 19 (approx. RM88) seat is simple to approve; a metered tool that can surge with model choice and token volume turns into a budget conversation. Admin caps help, but they reintroduce friction. The trade-off is clear: greater transparency, more oversight, and a sharper bill.
Record June Exposes GitHub’s Infrastructure Stress
When the new pricing went live and enterprise use followed, GitHub discovered the less impressive side of an AI boom: keeping the service up under real load. After switching Copilot to usage-based billing on June 1, GitHub saw a significant jump in customer usage and recorded its best month ever, according to Fedorov’s internal remarks. But that spike arrived on top of dozens of major outages this year, and commits are reportedly on pace to reach 14 billion in 2026, up from 1 billion in 2025. The result is infrastructure strain. GitHub has had to seek extra capacity from a leading cloud rival to support AI-driven traffic while continuing a planned migration to another cloud platform aimed at 2027. One founder’s complaint that GitHub is not a place for serious work if it blocks developers for hours is a warning shot. You cannot celebrate agent adoption and then act surprised when agent workloads behave differently from autocomplete.
The New Enterprise AI Bargain: Real Usage, Real Revenue, Real Pressure
GitHub’s record June is a signal: enterprise customers will use AI coding tools heavily when pricing matches how they work. Vendors can now charge closer to real consumption, and GitHub’s CTO has already said he does not think prices need to rise much based on the current spike, though he stopped short of firm plans. The deeper lesson is that usage-based billing is not a billing footnote—it is the business model catching up to the product. More AI usage means more revenue, more scrutiny, and more pressure on the pipes underneath. For any company selling AI into enterprises, GitHub’s experience should be a blueprint and a warning. Metered pricing can unlock spend from customers who are truly getting value, but it will also expose every weak assumption in your infrastructure and your budgeting. The Copilot meter has turned on the tap for enterprise AI adoption; whether vendors can keep the water flowing is the next real test.
