Home/Newsletter/Your AI Budget Has a Water Bill
Edition #24

Your AI Budget Has a Water Bill

Dan Toma·September 22, 2026·4 min read
Key Takeaway

The compute under your AI tools has always had an energy and water cost, absorbed by someone else. Regulation is now putting that cost on the operator and on the record, which makes the current price of your AI stack the cheapest it will be.


FAQ

Will AI tools get more expensive?

The underlying compute cost is being formally assigned to data centre operators through new utility rate classifications and infrastructure charges, which puts a floor under how cheap inference can get. Model efficiency will keep improving, but the energy and water inputs are now a priced cost rather than an absorbed one.

What do the new California data centre rules require?

The package requires a new utility rate classification for data centres, makes operators pay for local power grid and water system upgrades, and requires proposed facilities to disclose estimated water use, energy efficiency information and drought planning to local governments.

How should a company budget for AI cost increases?

Measure total real spend including API usage outside formal subscriptions, then calculate cost per useful output rather than cost per token. Test which parts of the operation would stop making commercial sense if that figure doubled, and move bulk workloads to smaller models before you are forced to.

Subscribe to The Weekly Vibe

Every Tuesday. 5-7 original takes on what matters in AI, Marketing, and Business Growth. No spam, no fluff, unsubscribe anytime.