Who Pays for
California’s AI Boom?

California’s AI boom is creating a growing demand for electricity. According to Senator Jerry McNerney’s office, the state’s grid operator expects data-center load to grow by 2.3 gigawatts by 2030. Someone has to pay for the transmission lines, substations, and grid upgrades that new demand requires. The question is: who?

The debate involves two related but distinct questions: how much information data centers should disclose, and how the costs of serving them should be allocated. AB 222 brought the disclosure question into focus. SB 1168 addresses rate structures and the share of costs borne by data centers and other customers.

Signed into law on September 21, 2026, SB 1168 directs the California Public Utilities Commission to assess opportunities for rate structures that ensure data centers pay a reasonable share of transmission and distribution costs, account for their proportionate share of load increases and procurement needs, and alleviate pressure on residential ratepayers. The law starts a regulatory process; it does not automatically reduce anyone’s next electricity bill.

I’ve been following this fight closely because I’ve spent the past year building financial models, including net present value, internal rate of return, and depreciation schedules, to help small, independent businesses evaluate clean energy investments. That work has made the mechanics of a “cost shift” concrete to me in a way most coverage of this issue doesn’t capture.

When a utility builds new transmission capacity to serve a data center, the cost remains after the facility is built. Depending on the applicable tariff and regulatory decisions, some costs can be spread across a wider group of customers. A rate structure designed around cost causation asks which costs were created by new demand and how they should be allocated to the customers responsible.

A hyperscale operator and an independent hotel or medical clinic face very different financial pressures. Even a modest increase in electricity costs can matter to a small business operating on thin margins. That asymmetry is central to the debate. Disclosure helps the public understand a problem; rate design determines how its costs are distributed.

As AI infrastructure grows nationally, the question of who pays for the grid will recur wherever data-center demand expands. California has an opportunity to demonstrate an approach that supports innovation while making the allocation of infrastructure costs more transparent and accountable.

Now that SB 1168 has become law, the next test is implementation. Regulators will need to identify the costs associated with new loads, consider reliability and procurement needs, and assess protections for existing customers. The measure’s value will depend on how those principles are translated into workable rates.

I want California to support technological progress while taking affordability seriously. The goal should be a system in which the businesses driving new costs pay an appropriate share, instead of leaving households and independent businesses to absorb costs they did not cause.