The promotional credits masking AI coding costs expire in September
GitHub Copilot Business plans carry an extra $30/user/month and Enterprise an extra $70/user/month in promotional credits through August 2026. When those lapse, teams whose usage has not changed will see their real baseline for the first time — on top of an Enterprise seat that already effectively costs $60/user before token spend.
A promotional credit is a loan against a future invoice, and the repayment date for a lot of AI coding budgets is 1 September. Nothing about anyone's usage needs to change for costs to jump.
The stated prices already understate the real ones. Copilot Enterprise lists at $39/user but requires a GitHub Enterprise Cloud seat at $21/user, making the effective base $60/user before a single token. Since 1 June 2026 agent mode and premium models bill through metered AI Credits at one cent each, with standard completions unmetered on paid plans. Cursor splits at $40/month standard and $120/month premium for 5x usage. Claude Code runs rolling five-hour windows plus weekly caps shared across terminal and web rather than a flat allowance.
Reported all-in spend of $200–$600 per month per developer for teams mixing inline completion with agentic work is where the credits have been doing their work. Agent mode is where the variance lives, because an agent's token consumption is a function of task difficulty rather than of typing speed.
The structural point is that per-seat pricing is being quietly abandoned in favour of metering while keeping the seat as a cover charge — and agent-seat pricing is the next step in the same direction. September is when a lot of teams find out what they have actually been buying.
daily.dev — What AI coding tools really cost in 2026: Cursor vs Copilot vs Claude Code pricing → · DX — AI coding assistant pricing and ROI guide (2026) →