Price certainty is part of the product.
A messaging, communications, and field-enablement response to GitHub Copilot's shift to usage-based billing. Written 18 August 2026, pre-decision, for Copilot GTM leadership. Revised 16 September 2026.
The transition landed. The invoice has not.
On 1 June, premium requests were replaced by metered AI Credits across every plan. Base prices held: Business $19, Enterprise $39 per user per month. Completions stay free; chat, CLI, agents, and code review are metered.
Promotional credits for existing customers replaced the standard allowance from June to August only; they were not added on top. The backlash so far is almost entirely individuals. Enterprise is still silent, which is the tell: the real exposure hasn't hit a finance team yet.
On 1 September, included usage falls by up to 44%: Business −37% (3,000 to 1,900 credits), Enterprise −44% (7,000 to 3,900 credits). Seat prices do not change.
We didn't raise prices. We removed the ability to forecast them.
is absorbed negotiated at renewal stays inside engineering · Quiet.
is escalated reaches finance, triggers procurement review ends in a spend freeze · Loud.
We produced the second and are being criticised for the first. Every fix that adjusts price solves the wrong problem. Credits burn on tokens at each model's API rate, and the person choosing is a developer mid-task with no budget context.
The safety net came out in the same release as the meter.
Allowance runs out Model falls back Work continues, degraded · Graceful. No decision required.
Allowance runs out No fallback exists Overage by default, or work stops · Binary. Overage unless an admin opts out.
What 1,000 Enterprise seats pay in September.
At full promotional usage: ~$31K of monthly overage, billed by default. Below 3,900 credits per seat September adds nothing; above it every credit is billed, and the customer can't see in advance which side of that line they land on.
From predictable seat to governed platform.
"One price per developer. Stop thinking about it."
→
"Spend you can see, attribute, and govern before it happens."
For the platform lead who now owns a variable AI budget they're accountable for, Copilot meters agentic work against a pooled organisation allowance, with budgets at enterprise, cost centre and user level, so spend is attributable, can be capped, and holds up in a budget review.
What it deliberately does not claim: that the bill is predictable. It isn't, and claiming otherwise breaks trust a second time.
One reposition. Three translations.
Developer · Sees the meter. Completions stay free; credit cost shows at the point you pick a model. · "See the cost before you commit."
Platform lead · Owns allocation. Pooled usage removes stranded capacity; budgets at enterprise, cost-centre, or user level. · "Allocate by cost centre, not by seat."
Finance · Sees the variance. On exhaustion the organisation decides: continue at published rates, or cap. · "You choose the ceiling in advance."
Everyone has this problem. Not everyone shows the customer.
Agentic coding consumes frontier inference. That cost is real for every vendor. The only question is who absorbs the variance, and for how long.
Predictable invoice, easy to buy Vendor absorbs variance, then rations quietly through throttles, fair-use limits and downgrades at peak Spend can't be attributed, because there's no spend to attribute
Spend is attributable and can be capped in advance The invoice moves The trust cost is paid up front, in public, which is where we're standing now
Land the cliff before it lands on them.
Admin notification · Every admin sees their own June–August consumption beside their standard September allowance.
Field briefed first · AEs and CSMs get the cap trade-off before customers get the email.
Guided configuration · CSM-led budget setup, ranked by projected gap rather than by ARR.
Say it plainly, in public · Changelog and docs state that promotional credits have ended. Silence converts a billing change into a trust story.
First unsubsidised invoice · Outreach to every account above forecast before they open it.
One ask. Marketing can't close this alone.
When the pool is exhausted and the organisation has chosen to cap, route to a low-cost model at reduced concurrency instead of stopping. The admin keeps a hard spend ceiling. The developer keeps a working tool.
Why this and not a discount. A taper postpones the conversation by one cycle and costs revenue in every cycle after. A floor is paid for once.
If it's declined. The claim weakens from governed to attributable. The fallback is alerting well ahead of the ceiling and a named approval path, which manages the stop rather than preventing it.
Four numbers, and what would prove me wrong.
Organisations with a budget set at any level, by 30 September.
Share of orgs whose usage approaches their own ceiling within 14 days. The only signal that arrives early enough to recalibrate.
Admin projection against actual October spend. The direct measure, and the number I'd be held to.
Non-renewals naming predictability. Lagging and contaminated. Read alongside 03, never alone.
Naveen Raj Kanagaraj · M.S. Marketing Analysis, DePaul University · Independent analysis, not affiliated with the companies named.