An act to amend Section 25402 of, and to add Chapter 4.4 (commencing with Section 25345) to Division 15 of, the Public Resources Code, and to add Section 451.4 to the Public Utilities Code, add Section 25302.9 to the Public Resources Code, and to add and repeal Section 913.18 of the Public Utilities Code, relating to energy.
AB 222 would create a new state framework for tracking and evaluating the energy use of data centers and generative AI models, with a focus on electricity demand, efficiency, and potential cost impacts on utility customers. It requires developers of covered AI models to estimate and publicly post the energy used to develop and operate those models, including the share of that energy generated in California. It also directs the California Energy Commission to collect data center power usage effectiveness information, include data center energy consumption trends in its integrated energy policy reports, and provide a 2027 assessment of data center electrical load trends, peak demand, and possible grid mitigation measures.
The bill also adds a utility-rate component by directing the Public Utilities Commission to examine whether costs associated with serving new or expanded data centers are just and reasonable and to minimize cost shifting to ratepayers who do not directly benefit from those facilities. The PUC would be required to assess the extent to which data center-related load growth causes costs to be shifted to other customers, report its findings to the Legislature by January 1, 2027, and post the assessment publicly. That PUC reporting requirement would sunset on January 1, 2031.
In practical terms, AB 222 would affect the Public Resources Code and Public Utilities Code by creating new reporting obligations for data center owners, AI developers, the Energy Commission, and the PUC. It would also authorize the Energy Commission to report data center energy use only in aggregate form to protect confidential or proprietary information, while making clear that personally identifiable information and customer-specific energy data should not be disclosed. The bill is framed as an energy and grid-planning measure, but it also intersects with AI transparency and utility ratemaking.
The general sentiment reflected in the voting history appears cautiously favorable but not unanimous. The bill advanced through multiple committees and floor votes with majority support, including a strong Assembly third-reading vote, suggesting broad interest in addressing data center electricity demand and cost allocation. At the same time, the bill was amended several times and ultimately held under submission in committee, indicating that members continued to scrutinize its scope, implementation, and fiscal or policy implications.
The main points of contention appear to be privacy, proprietary business information, and who should bear the costs of data center growth. The bill itself includes constitutional findings to justify limiting public access to certain information in order to protect confidential and proprietary data, which suggests concern about disclosure requirements. Another likely area of debate is whether the new reporting and cost-shift analyses are sufficient to protect ordinary ratepayers from subsidizing data center infrastructure, versus whether the requirements could impose burdens on utilities, developers, and data center operators.
AB 222 would add new reporting, planning, and oversight requirements to California energy law. It would amend the Public Resources Code to require data center power usage effectiveness reporting and to require the Energy Commission to incorporate data center load trends into its integrated energy policy reporting, and it would add a new Public Utilities Code section requiring the PUC to evaluate and report on data center-related cost shifts. The bill would also impose new public disclosure obligations on developers of covered generative AI models regarding energy used to develop and operate those models. These changes would affect data center owners, AI developers, electric utilities, the Energy Commission, the PUC, and utility ratepayers.
The bill appears to have received generally supportive but measured treatment in the Legislature. It moved forward on several committee votes and passed the Assembly with a substantial margin, indicating that many lawmakers viewed the issue of data center energy use and cost allocation as important. However, the repeated amendments, referral changes, and eventual holding under submission suggest that there was also significant caution about the bill’s details, implementation burden, and potential effects on confidential business information and utility regulation.
The most notable contention centers on transparency versus confidentiality. AB 222 requires energy-use reporting for AI developers and data centers, but it also expressly limits disclosure of information that could reveal proprietary or customer-specific data, which implies concern from industry and privacy-focused stakeholders. A second major issue is cost responsibility: the bill seeks to prevent data center infrastructure and load-growth costs from being shifted to other ratepayers, but utilities and data center interests may disagree over how those costs should be measured, allocated, and mitigated. The bill’s multiple amendments and committee holds suggest unresolved debate over how far the state should go in regulating data center energy impacts while avoiding overreach.