Google Cloud lets you set a hard monthly limit on some AI spending

Google Cloud now lets you set a hard monthly spending limit on some of its AI tools. When the limit is reached, new requests to that service are blocked, and nothing is deleted. The feature is called spend caps. It is now generally available, which means it has left testing and is open to customers.

How it works

A spend cap is a budget with enforcement attached. You set it on one service in one project for a calendar month. When estimated spending hits your number, new requests stop. Requests already running finish and are charged. Resources that stay on, such as running servers, keep running. Billing administrators and project owners get alerts at 50, 80 and 100 percent of the cap.

What changed since the preview

Google ran a public preview in July. The full release adds three things: an API, which lets software set caps automatically, live cost estimates that trigger caps within minutes for AI services, and automatic monthly resets.

What it does not cover

• Only four services are covered: Gemini API, Gemini Enterprise Agent Platform, Cloud Run and Cloud Run functions.
• Compute Engine, Cloud Storage and networking are not covered, so a forgotten GPU server would not be stopped.
• Caps do not work across folders, organizations, labels or several services at once.
• Subscription charges such as Gemini Enterprise seats are outside the cap.
• Provisioned Throughput, reserved AI capacity, is never halted.

Not a perfect brake

Enforcement is not instant. Google's documentation says any cost overages are billed as normal and advises setting the cap slightly below your absolute limit. The estimate also uses list prices and ignores credits and discounts. One Hacker News commenter called the design "very crude." Developer Simon Willison argued that hard caps should be the default, because most people would prefer errors to a surprise bill of $10,000 or more.

What to do

If your business uses these Google services, set a cap below the most you could stand to pay. Then check that every service you rely on is on the covered list. Caps are set from the Budgets and Alerts page in the Billing console or through the Budgets API. Google also offers early anomaly alerts that flag unusual cost jumps.