Aspend cap in Gemini Enterprise now pauses an agent's requests the moment a project hits its monthly limit, and Google Cloud has documented exactly how to turn the meter back on.
Google Cloud published the change on August 26: a pay-as-you-go option beside the existing per-user seat subscription for Gemini Enterprise, project-level spend caps that pause an agent's calls automatically once spend hits the ceiling, and Flexible Savings Plans that cut token costs 10 percent for a one-year commitment and 20 percent for three years, with no minimum or maximum spend required (Google Cloud Blog, "FinOps for the AI era," 2026). Google Antigravity and Android Studio AI now draw quota from the same subscription for eligible customers, pooled project-wide instead of assigned seat by seat, though Google says both the pay-as-you-go tier and the Antigravity inclusion are live for select customers first, with broader rollout still coming (Google Cloud Blog, 2026).
Seats are not going away.
Per-user subscriptions still carry the daily quota pool a fixed monthly fee buys, and Google positions pay-as-you-go as an option to mix with seats rather than a replacement for them (Google Cloud Blog, 2026). What has changed is the assumption underneath the pricing: a chatbot session is one request and one response against a license a company already bought, while an agent loop calls a model, evaluates the result, calls a tool, and can spawn a subagent that repeats the cycle without a person approving each step. Consumption tracks the complexity of the task, not the number of people holding a license. That gap between seat count and agent consumption is why Google built pay-as-you-go, and it is also why a hard cap can now stop a task mid-loop.
Google Documented the Billing Reset, Not the Task Reset
When a project hits its spend cap, the agent's requests pause and the rest of the production environment keeps running (Google Cloud Blog, 2026). Administrators get email alerts at 50, 80, and 100 percent of the budget, and once a cap trips they can resume work with a single click in the console, or turn on overages so usage keeps flowing at consumption rates, which can draw against an existing Flexible Savings Plan to keep the overage discounted (Google Cloud Blog, 2026). That is a documented, well-designed billing workflow.
None of it says what happens to the specific task the agent was running when the pause hit.
A multi-step workflow that calls subagents to research, draft, and verify a report does not fail the way one interrupted chatbot request fails. Resuming billing with a click restarts the meter. It does not say whether the agent picks the task back up where it stopped, restarts it from the beginning, or leaves a partial draft sitting there with no signal that the verification step never ran. Google's own materials describe the billing recovery in detail and leave the task recovery unaddressed.
Antigravity Now Shares the Ceiling It Used to Have to Itself
Folding Antigravity and Android Studio AI into eligible Gemini Enterprise subscriptions gives administrators one console for spend, security, and usage across coding agents and business agents, with project-level caps live today and per-user and team-level controls still rolling out later this year (Google Cloud Blog, "Expanding Google Antigravity for Enterprise Customers," 2026). Until those finer controls arrive, a developer running a long autonomous coding session and a sales team running an agent-assisted forecast draw from the same monthly cap if they sit in the same project. Developer overage can pause business agents that had nothing to do with it, unless finance teams split the workloads into separate projects now.
What Google Has Not Said
Google has not published whether an agent task resumes from where it stopped or restarts from scratch once a cap is lifted, and it has not said whether the tokens spent on the interrupted run count against the cap that stopped it. Per-user and team-level spend thresholds for Antigravity are confirmed as coming later this year but carry no date. Pay-as-you-go pricing and Antigravity inclusion remain limited to select customers, so the caps and quota pooling described here will not reach every Gemini Enterprise account on the same timeline.
Gerstenhaber, Michael, and Pravir Gupta. "FinOps for the AI Era: New Flexible Billing and Cost Controls for Agents." Google Cloud Blog, 26 Aug. 2026, cloud.google.com/blog/products/ai-machine-learning/flexible-billing-and-cost-controls-for-agents-on-google-cloud.
Densmore, Scott, and Damith Karunaratne. "Expanding Google Antigravity for Enterprise Customers." Google Cloud Blog, 21 Aug. 2026, cloud.google.com/blog/products/ai-machine-learning/expanding-google-antigravity-for-enterprise-customers.
Evans, Scarlett. "Google Rolls Out Flexible Billing, Cost Controls for AI Agents." CIO Dive, 26 Aug. 2026, www.ciodive.com/news/google-rolls-out-flexible-billing-cost-controls-ai/828832/.
"Google Adds Pay-As-You-Go Gemini Pricing as Enterprises Seek Control over AI Spending." CIO, 26 Aug. 2026, www.cio.com/article/4214185/google-adds-pay-as-you-go-gemini-pricing-as-enterprises-seek-control-over-ai-spending.html.
