The U.S. House of Representatives passed the Ratepayer Protection Act (H.R. 9340) on September 17 by a vote of 417 to 3 — a bipartisan bill that requires large data centers, including AI training and inference facilities, to cover the full cost of the electrical grid upgrades their demand triggers, instead of passing that cost on to other ratepayers' bills.
The Ratepayer Protection Act amends PURPA (the Public Utility Regulatory Policies Act of 1978), the federal framework that sets standards states must formally consider adopting for their utilities. The bill establishes a federal standard under which large data centers — at a 100-megawatt demand threshold — must cover the full incremental cost of new generation, transmission, or grid upgrades needed to serve their consumption, instead of that cost being spread across all electricity customers in the area.
Here's the detail that's gotten the least coverage: under the PURPA framework, federal standards like this one are "procedurally mandatory but substantively discretionary." That means state public utility commissions are legally required to hold a formal process to consider the standard, but aren't required to adopt it as state policy. The law forces the conversation in every state, but doesn't dictate the outcome.
The momentum behind this bill is voter anger in several states over rising electricity bills attributed, at least in part, to the accelerated build-out of data centers for training and running inference on AI models. When a utility builds a new substation or transmission line to serve a single large-demand customer — like a hyperscaler's data center — those costs have traditionally been spread across the general rate base. This bill aims to reverse that logic specifically for large compute consumers.
For companies in Guatemala and Latin America consuming AI compute via APIs from providers like OpenAI, Anthropic, or Google, the effect of this bill — if it also passes the Senate and states adopt it — would be indirect but real: if U.S. data center operators absorb the full cost of their electrical infrastructure instead of subsidizing it through general rates, that cost eventually flows through to API token pricing. It's one more reason for companies that depend heavily on generative AI to track mid-term compute pricing trends, and for those evaluating building their own infrastructure (on-premise or private cloud) to check whether the cost equation shifts if public cloud pricing starts reflecting the real energy cost of the data centers behind it.
Carlos Montiel is an enterprise AI solutions architect. He implements LLMs, Agents, RAG and orchestrators for companies across Guatemala and Latin America. Reach out for a consultation.
Contact Carlos Montiel