Data Centers and California Electricity Policy

Data Centers and California Electricity Policy Report #292 | March 2026 Milton Marks Commission on California State Government Organization and Economy www.lhc.ca.gov LITTLE HOOVER COMMISSION Dedicated to Promoting Economy Pedro Nava and Efficiency in California State Chair Government Anthony Cannella Vice Chair The Little Hoover Commission, formally known as the Milton Marks “Little Hoover” Commission on California State Government Dion Aroner Organization and Economy, is an independent state oversight agency. David Beier* Senator Christopher Cabaldon By statute, the Commission is a bipartisan board composed of five public members appointed by the governor, four public Assemblymember Phillip Chen members appointed by the Legislature, two senators and Gil Garcetti two assemblymembers. José Atilio Hernández* In creating the Commission in 1962, the Legislature declared Jason Johnson its purpose: Gayle Miller Senator Roger Niello ...to secure assistance for the Governor and itself in promoting economy, efficiency and improved services in the Assemblymember Liz Ortega transaction of the public business in the various departments, Janna Sidley agencies and instrumentalities of the executive branch of *Served on study subcommittee the state government, and in making the operation of all state departments, agencies and instrumentalities, and all expenditures of public funds, more directly responsive COMMISSION STAFF to the wishes of the people as expressed by their elected Ethan Rarick representatives... Executive Director The Commission fulfills this charge by listening to the public, Tamar Foster consulting with the experts and conferring with the wise. In the Deputy Executive Director course of its investigations, the Commission typically empanels advisory committees, conducts public hearings and visits government Krystal Beckham operations in action. Daniel Harris-McCoy Shara McAlister Its conclusions are submitted to the Governor and the Legislature for their consideration. Recommendations often take the form Gibran Maciel of legislation, which the Commission supports through the Jenna Waite legislative process. Cover Image Created by ChatGPT Contacting the Commission All correspondence should be addressed to the Commission Office: Little Hoover Commission 925 L Street, Suite 805, Sacramento, CA 95814 (916) 445-2125 | littlehoover@lhc.ca.gov This report is available from the Commission’s website at www.lhc.ca.gov. Letter from the Chair March 2026 The Honorable Gavin Newsom Governor of California The Honorable Monique Limón The Honorable Brian Jones President pro Tempore of the Senate Senate Minority Leader and members of the Senate The Honorable Robert Rivas The Honorable Heath Flora Speaker of the Assembly Assembly Minority Leader and members of the Assembly DEAR GOVERNOR AND MEMBERS OF THE LEGISLATURE: California’s electricity rates are among the highest in the nation, and affordability remains a top concern for residents. At the same time, the state is committed to some of the most ambitious clean-energy and climate goals in the country. Against this backdrop, the rapid growth of energy-intensive data centers presents both a serious challenge and a potential opportunity for California’s electricity system. If managed thoughtfully, these facilities could help catalyze investments that modernize the grid and strengthen reliability—without increasing costs for ratepayers or undermining the state’s clean-energy commitments. To help policymakers navigate these tradeoffs, the Little Hoover Commission examined the implications of data- center-driven load growth, with particular attention to ratepayer protection, regulatory structure, grid planning, environmental and community impacts, and innovation. In our report, Data Centers and California Electricity Policy, we offer recommendations to ensure data centers pay their fair share of infrastructure and grid-service costs, reduce the risk of stranded assets, improve interconnection and permitting predictability while maintaining strong review standards, protect communities from pollution—especially from backup generation—and support research and partnerships to advance cleaner, more efficient technologies. We respectfully submit this report and stand ready to assist as the state develops a policy framework that supports innovation while protecting Californians’ affordability, reliability, and clean-energy priorities. Sincerely, Pedro Nava, Chair Little Hoover Commission 3 | LITTLE HOOVER COMMISSION Table of Contents EXECUTIVE SUMMARY .......................................................................1 DATA CENTERS AND THE FUTURE OF CALIFORNIA’S ENERGY SYSTEM ..............................................................................................3 Introduction to Data Centers and Electricity Policy ...................................3 An Ethical Map for Navigating Data Center Risk .......................................4 Policy Approaches: Data Centers and Electricity ........................................6 DATA CENTERS AND ELECTRICITY POLICY IN CALIFORNIA ........8 Data Centers in California .............................................................................8 California Data Center Electricity Demand . ................................................9 The Disparate Pace of Energy Policy and the Consequences ..................12 Data Centers and Jurisdictional Policy Levers ...........................................13 Data Centers and California’s Regulatory and Planning Bodies. .............13 Current Data Center Rate Policy .................................................................15 Features of a New Very-Large-Load Tariff .................................................15 Improving the Interconnection Process .....................................................17 Mitigating Risk from Stranded Assets ........................................................18 Broadening Data Center Oversight .............................................................19 DATA CENTERS AND GRID PLANNING ..........................................21 Data Centers and Grid Reliability ................................................................21 Maximizing Existing Grid Capacity ..............................................................22 Demand Flexibility ........................................................................................24 BUILDING A CLEANER, SMARTER DATA CENTER FUTURE ........25 Data Centers and the Environment ............................................................25 Clean and Reliable Backup Generation ......................................................25 Incentivizing Data Center Energy Innovation ............................................27 Promoting Data Center Energy Research. ..................................................28 Addendum: Data Centers and the Water Supply ......................................30 Addendum: Data Centers and Air Pollution .............................................. 31 ENDNOTES ............................................................................................................33 EXECUTIVE SUMMARY California’s electricity rates are among the highest in These goals can be achieved through a well-designed the country and have risen considerably faster than very-large-load tariff that incorporates credit inflation in recent years, making affordability a major requirements, prepayment, full cost recovery through concern for residents. At the same time, the state is rates, minimum demand and term commitments, already committed to some of the most ambitious load flexibility, and meaningful exit penalties. Robust clean-energy and climate goals in the nation. data-sharing requirements should support these tariffs and inform ongoing policymaking. At the Against this backdrop, the rapid growth of energy- same time, the state should work to accelerate data hungry data centers presents both a serious center interconnection to support innovation, while challenge and a potential opportunity for California’s maintaining strong and consistent review standards. electricity system. Business and political leaders believe that data-center-driven computing—and SMARTER GRID PLANNING artificial intelligence in particular—has the potential California should accommodate data center to transform society by turbocharging service growth while avoiding unnecessary infrastructure delivery, accelerating scientific and medical discovery, investments by making more effective use of its improving governmental efficiency, and ushering existing electric grid through smarter planning. in an era of economic growth supported by major This includes incentivizing and prioritizing targeted productivity gains. transmission upgrades and strategic data center siting, encouraging the clean and flexible use of Unlocking this potential depends on California’s backup power resources, and leveraging load shifting ability to meet the enormous energy demands of to support the grid during periods of peak demand. these facilities. Doing so responsibly will require CLEAN ENERGY AND COMMUNITY enormous new investment without raising electricity PROTECTION rates or compromising the state’s clean energy California leads the nation in the pursuit of clean, commitments. zero-carbon energy, and the addition of new This challenge comes with opportunity. With the right data center load must support—rather than policies in place, data centers could help catalyze compromise—these goals. Communities must also investments that modernize California’s grid while be protected from added pollution, particularly from potentially lowering electricity costs for residents backup power sources that are often diesel-powered. and businesses. The report that follows outlines Achieving these objectives will require strong recommendations for how the state might achieve statutory protections alongside incentives to adopt this balance, with particular attention to rate design, technologies that reduce grid and environmental regulatory structure, affordability, environmental impacts and, consistent with the state’s culture protection, and innovation. of innovation, to develop new technologies and approaches that advance these goals. PROTECTING RATEPAYERS Above all, California must develop tariffs and policies Recommendations for data centers that protect consumers from rising To help ensure that the growth of data centers energy costs. This includes addressing cost pressures strengthens California’s electricity system without from increased electricity demand and ensuring increasing costs for ratepayers or undermining that ratepayers are not unfairly burdened with clean-energy goals, the Little Hoover Commission infrastructure costs, including the risk of stranded recommends the following actions for policymakers assets—investments built to serve data center loads and regulators: that are not ultimately paid for by those loads. 1 | LITTLE HOOVER COMMISSION 1. California should strive to develop a stable addressed with the same level of rigor as investor- regulatory framework for data centers that balances owned utilities. economic growth with clean, efficient, affordable, 8. California should ensure that new large data center and reliable electricity provision. It should make loads are integrated in a manner that preserves certain that costs associated with new large-load system reliability and does not disadvantage existing development are fairly allocated and do not shift customers. onto existing ratepayers. 9. California should maximize the use of existing 2. California should lead the nation on data-center grid capacity when considering data-center siting energy regulation by grounding its policies in the and new infrastructure, supported by transparent best ideas from researchers, state agencies, utilities, data and visualization tools that help developers and developers, and other jurisdictions (state, federal, policymakers identify locations where capacity is international). available or can be efficiently expanded. 3. California should enable regulators to 10. California should maximize load shift confidentially access essential facility-level electricity- opportunities to use existing load capacity and lower use data from large data centers, supported by costs for residential customers. structured data-sharing among agencies, to improve planning, assess localized impacts, and ensure fair 11. California should meet rising data-center cost responsibility while protecting sensitive business electricity demand without weakening its clean- information. energy or climate goals, including behind-the-meter generation. 4. California should ensure that data centers pay for costs imposed on the electric system through 12. California should require data centers to maintain tariffs and cost-allocation policies that recover a minimum level of clean backup power to support infrastructure and grid-services costs in a manner load shifting, enable curtailment during grid stress that benefits rather than burdens other ratepayers. (including extreme heat events), and reduce reliance on fossil-fuel backup generation. 5. While maintaining a rate structure that ensures shared costs and a high standard of regulatory 13. The state should limit pollution impacts on review, California should accelerate data-center nearby communities from data-center backup interconnection and permitting decisions to give generators. developers greater predictability in planning and 14. California should encourage data centers to investment. adopt technologies—such as advanced cooling, 6. California should establish consistent, statewide energy storage, or on-site clean power—that reduce safeguards that ensure large data-center customers their impact on the grid. bear appropriate financial risk, while making certain 15. California should establish a formal public- that major grid investments continue to serve the private partnership—provisionally the California public even if data center demand evolves differently Partnership for Advanced Research (CALPAR)—to than has been predicted. collaborate with industry, national laboratories, 7. The state should establish consistent statewide and universities to research, test, and promote oversight expectations for large data center technologies that improve the efficiency and developments in territories served by publicly environmental performance of AI-driven data owned utilities, municipal utilities, irrigation districts, centers. and community choice aggregators, ensuring that risks to reliability, costs, and local communities are 2 | LITTLE HOOVER COMMISSION Data Centers and the Future of California’s Energy System California’s electricity rates are among the highest in amounts of electricity and require costly energy the country and have risen considerably faster than infrastructure to connect them to the grid. To give inflation in recent years, making affordability a major a sense of scale, in July 2025, Pacific Gas & Electric concern for residents.1 At the same time, the state is (PG&E) reported that the amount of data center already committed to some of the most ambitious energy demand waiting to be connected to the grid clean-energy and climate goals in the nation. is equivalent to more than 10 percent of California’s current total generation capacity; enough electricity Against this backdrop, the rapid growth of energy- to power roughly 8 million homes.5 Comparably hungry data centers presents both a serious large upgrades will be needed—in the form of power challenge and a potential opportunity for California’s lines, substations, and transformers—to deliver this electricity system. This section explains what data energy. Meeting this demand responsibly will require centers are and how they relate to electricity policy; enormous new investment without raising electricity summarizes their benefits and risks; and outlines rates or compromising the state’s clean energy policy options for managing their impacts on energy commitments. affordability, reliability, and the state’s environmental goals. This challenge comes with opportunity. With the right policies in place, data centers could help catalyze INTRODUCTION TO DATA CENTERS AND investments that modernize California’s grid while ELECTRICITY POLICY potentially lowering electricity costs for residents As digital applications have become more complex— and businesses.6 The report that follows outlines from artificial intelligence and data analysis to recommendations for how the state might achieve streaming and social media—much of the computing this balance, with particular attention to rate design, power, storage, and software that support these regulatory structure, affordability, environmental applications has shifted from local devices to protection, and innovation. high-powered servers housed in data centers.2 By accessing these centralized resources over the Success is not guaranteed. The scale and Internet, devices such as personal computers and concentration of energy demand from data centers smartphones can perform tasks that far exceed their is unprecedented and has the potential to increase own hardware’s capacity. electricity prices while posing risks to communities and the environment.7 Policymakers may also face Many business and political leaders believe that data- pressure to defer to powerful technology companies center-driven computing—and artificial intelligence and utilities.8 Yet, if approached thoughtfully, this in particular—has the potential to transform moment could offer California a chance to align the society. They are betting, with dollars and political interests of industry, government, and the public. capital, that it will turbocharge service delivery, accelerate scientific and medical discovery, improve Indeed, data center operators increasingly emphasize governmental efficiency, and usher in an era of that responsible development requires not only economic growth supported by major productivity careful management of electricity and water use, but gains.3 also that developers contribute directly to solutions. Microsoft promotes a “community-first” approach Unlocking this potential depends on California’s that includes paying the full cost of required energy ability to meet the enormous energy demands of infrastructure, reducing and replenishing water use, these facilities.4 Data centers consume extraordinary 3 | LITTLE HOOVER COMMISSION and delivering local benefits such as jobs, workforce some cases, they can support emerging applications training, and tax revenues for public services.9 Google of advanced computing and even help revitalize similarly highlights efficiency and transparency, underused commercial or industrial areas.17 publicly reporting the electricity and water intensity These benefits, however, are accompanied by of AI queries to help communities and policymakers real risks. Data centers often require a significant understand system-wide impacts as demand grows.10 physical footprint, which can affect land availability, At the industry level, the Data Center Coalition states ecosystems, and long-term development flexibility.18 that its members are committed to paying their full They can cause environmental problems relating cost of electric service, including direct payment for to water use, carbon emissions, noise, and air utility infrastructure needed to serve data center pollutants.19 If incentives or tax agreements are load.11 That said, voluntary leadership by a subset poorly structured, governments may receive fewer of developers is not sufficient—statutory mandates fiscal benefits than anticipated. Limited transparency, are required to ensure that these and other best community concerns about fairness, and perceptions practices are applied consistently, transparently, and of corporate favoritism can also create mistrust or over the long term. political backlash.20 Integrating data centers into California’s electric grid In terms of energy-related impacts, data centers have has become a politically sensitive issue, unfolding the potential to play a constructive role in California’s at a time when public skepticism toward artificial energy system. Their predictable, large loads could intelligence contrasts sharply with the tech industry’s help support investments in new clean energy central role in the state’s economy.12 Experiences resources and spread fixed system costs across more in other states show how backlash against data customers, potentially lowering overall costs. And centers can influence public opinion and shape if they participate in flexible-load programs, data elections, underscoring the stakes of getting policy centers may also contribute to grid reliability, support right.13 In California, this issue is playing out against demand response, and help integrate renewable an ongoing affordability crisis, heightening the need energy, while strategic siting and workload shifting for careful decision-making.14 This report outlines could permit the current grid to function more how policymakers can navigate these pressures, efficiently.21 identifying best practices in rate design, regulation, and implementation to help integrate data centers These opportunities again come with risks. Data- in a way that protects consumers while promoting center driven electricity demand has been widely innovation. cited as a potential driver of higher rates. New large AN ETHICAL MAP FOR NAVIGATING DATA loads may require substantial new generation, CENTER RISK transmission, and substation infrastructure, which Hosting data centers presents a complex set of also add cost, strain planning processes, and benefits and risks that span economic, community, crowd out other electrification priorities such as environmental, and energy-related issues.15 Starting transportation and buildings.22 And if demand with non-energy impacts, data centers can provide decreases or data centers relocate, Californians could meaningful economic benefits to local communities be left paying for unused energy infrastructure.23 and the state.16 They can reinforce California’s There are also important energy-related leadership in advanced technology, help retain environmental concerns.24 Reliance on diesel backup major AI and cloud firms, and support tax revenues, generation can harm nearby communities.25 And, investment, and research partnerships. At the local more broadly, if the state cannot keep pace with level, they can create construction jobs, support demand growth, data centers could complicate vendors and trades, strengthen property-tax bases, California’s pathway toward achieving its long-term and stimulate infrastructure improvements. In clean-energy and climate goals.26 4 | LITTLE HOOVER COMMISSION To navigate this complex issue, policymakers should ◊ Climate and Health: Policies governing data approach decisions about data centers and electricity centers should support California’s clean-energy policy with a clear sense of the values they want and climate commitments. Data centers must not to advance.27 The choices California makes will have a negative impact on public health. shape not only economic growth and technological ◊ Innovation in the Public Interest: Data centers leadership, but also how costs are shared, how should contribute to grid modernization and state communities are impacted, and whether the state policy goals in ways that benefit all Californians. can meet its clean-energy and reliability goals. The ◊ Transparency and Public Trust: Agreements, following values framework can help ensure that decisions, and outcomes related to data centers policy decisions are disciplined, transparent, and in should be transparent, understandable, and the public interest: accountable to the public. ◊ Affordability and Reliability: Californians should ◊ Predictability and Stability: California should have access to electricity that remains affordable commit to developing stable regulatory and dependable as data centers are added to the frameworks that enable data center developers grid. to invest and plan responsibly for the long term. ◊ Fairness and Cost Responsibility: Data centers should not shift electricity-related costs onto households or small businesses. Benefits and Risks: Non-Energy Impacts Benefits Risks Statewide Economic & Strategic Land Use & Environmental ◊ Reinforces CA tech leadership ◊ Large land footprint; long-term site impacts including ecosystem disruption ◊ Helps retain major AI/cloud firms ◊ Tax revenues & economic growth ◊ Significant water use for cooling ◊ Supports R&D and university partnerships ◊ Noise from cooling systems ◊ Air quality impacts from diesel backup Local & Community Economic Benefits Fiscal & Economic Risks ◊ Construction jobs & near-term stimulus ◊ Incentives may offset gains ◊ Potential local tax revenues ◊ Possible fiscal over-dependence ◊ Benefits without major traffic or population ◊ Uneven benefit distribution growth ◊ Weak transparency on costs vs. benefits ◊ Possible community benefit agreements Urban Development & Infrastructure Community, Governance & Public Trust ◊ Can drive broadband and local infrastructure ◊ Political backlash over corporate favoritism upgrades ◊ Confidentiality reduces trust ◊ Edge computing can support innovation/ advanced services ◊ Potential security/emergency target ◊ Can revitalize vacant commercial and industrial sites 5 | LITTLE HOOVER COMMISSION I Benefits and Risks: Energy-Related Impacts Benefits Risks Affordability & System Economics Affordability Risks ◊ Potential to spread fixed system costs ◊ Cited driver of rising electricity rates ◊ Supports long-term clean-energy financing ◊ Poor rate design could shift costs to households & ◊ Predictable load aids planning small businesses Clean-Energy & Climate Alignment Infrastructure & Planning Burdens ◊ Keeps load in CA’s cleaner energy mix ◊ Requires major new generation and transmission investment ◊ Supports CA climate leadership ◊ Aligns Western regional development toward ◊ Can crowd out other electrification needs cleaner power ◊ Heightens siting and permitting conflicts Grid Reliability & Innovation Environmental & Reliability Risks ◊ Strategic siting and workload shifting can ◊ Added load could strain grid capacity and increase reduce grid stress reliance on outdated energy sources ◊ Drives upgrades in grid infrastructure and ◊ Reliance on diesel backup generation innovation ◊ Locational inequities in community impacts Opportunity for Grid and Policy Innovation Stranded Asset & Policy Failure Risks ◊ Data centers could help finance grid upgrades ◊ If tech shifts or firms relocate, infrastructure may that benefit all Californians be stranded ◊ Rethinking energy-related policies and ◊ Ratepayers may pay for unused capacity processes could improve state utility ◊ Benefits materialize only if policy execution is regulation strong Policymakers should consider these values non- to government and can arise from law, regulation, negotiable as they navigate the evolving landscape regulatory or tariff direction or enforceable of data centers and energy, using them as an ethical agreement. map to ensure decisions remain fair, transparent, affordable, equitable, and aligned with the state’s RECOMMENDATION 1: California should strive climate commitments. By grounding policy in these to develop a stable regulatory framework for data principles, California can encourage innovation centers that balances economic growth with clean, and economic growth while protecting ratepayers, efficient, affordable, and reliable electricity provision. supporting communities, and strengthening public It should make certain that costs associated with new trust. large-load development are fairly allocated and do not shift onto existing ratepayers. To protect ratepayers from bearing costs associated with the creation of new energy generation for large POLICY APPROACHES: DATA CENTERS AND AI data centers, the owners, operators, and users of ELECTRICITY such facilities should be required to build or fully pay California policymakers may wish to look to other for the new capacity that serves their interests. This states for guidance on how to structure regulatory obligation should be transparent and accountable and rate policies for data centers. At the same time, 6 | LITTLE HOOVER COMMISSION this is an unprecedented moment, and California has Paraguay, for example, is leveraging surplus clean the opportunity not only to adopt proven approaches power from its hydroelectric resources to attract but also to innovate and design policies that reflect data centers and anchor a growing sector of its the state’s unique priorities around affordability, economy.30 In Europe, heat recycling programs in reliability, and climate leadership. Sweden, Finland, Ireland, and the United Kingdom are using waste heat from data centers to warm In January 2025, researchers at Lawrence Berkeley thousands of homes.31 A similar project is scheduled National Laboratory published a review of rate for apartments in the City of San Jose.32 design approaches that states and utilities across the country have either implemented or proposed We encourage policymakers to draw on the best of in response to the growth of data centers—here these approaches in shaping California’s approach and elsewhere referred to as “large loads” because to data centers and energy. Many of the major of the considerable electricity they draw from the players—state leaders, utilities, and technology grid.28 These approaches are organized according to companies—have signaled a willingness to work objective—e.g., defining which customers qualify for together on these challenges, even as residents specialized tariffs, reducing financial risk, developing remain skeptical about artificial intelligence, rising long-term contracts, or aligning energy sources with energy costs, and the pace of change. This effort environmental goals. The table below summarizes must therefore be undertaken thoughtfully, guided these approaches using simplified or slightly first and foremost by the interests of Californians. modified wording. RECOMMENDATION 2: California should States such as Ohio, Texas, and Oregon have led lead the nation on data-center energy regulation the nation in data center policy, and there are also by grounding its policies in the best ideas from innovative approaches emerging internationally researchers, state agencies, utilities, developers, and to reduce the energy impacts of these facilities.29 other jurisdictions (state, federal, international). Rate Design Approaches for Data Centers/Large Loads Category Purpose Example Provisions Eligibility Defines the conditions under which Minimum load requirement data centers may connect to the grid. Contract Size Specifies electricity purchase obligations Obligation for customers to pay for a defined and other provisions that reduce percentage of proposed demand financial risk to utilities and ratepayers. Contract Establishes the length and flexibility of Defined contract terms that balance risk to Duration agreements. ratepayers with ramp-up provisions and flexibility Energy Source Identifies requirements relating to the Clean energy sourcing requirements source of electricity. Other Captures additional design Marginal pricing mechanisms to reflect actual utility Provisions considerations. costs `` Source: Lawrence Berkeley National Laboratory 7 | LITTLE HOOVER COMMISSION Data Centers and Electricity Policy in California As California’s utilities add data centers to the grid, DATA CENTERS IN CALIFORNIA the state has the opportunity—and, indeed, the Data centers in California are currently concentrated obligation—to regulate them in a way that benefits in Santa Clara and Los Angeles.33 These locations all residents. This section describes where and how offer several advantages, including proximity to data centers are developing in California, how quickly major clients, access to fast fiber networks, local their electricity demand is growing, and why the governments accustomed to supporting large pace of development is putting stress on existing technology infrastructure, and utilities willing policy frameworks. It then explains the regulatory and able to provide the substantial electricity and levers available to California—across utilities, state water needed to support operations. In both areas, agencies, and local governments—and highlights municipal utilities offer lower electricity rates than how current rate structures and interconnection investor-owned utilities, providing substantial cost policies can shape costs, risks, and timelines. savings to data center operators. PG&E Data Center Project Pipeline as of June 2024 Source: Pacific Gas & Electric Co. 8 | LITTLE HOOVER COMMISSION More recently, however, development patterns proposed for California’s Central Valley, where land appear to be shifting. A growing number of new data availability and grid access may be more favorable.36 centers have been proposed in territory served by Data center ownership is highly fragmented, PG&E, one of California’s investor-owned utilities, with facilities operated both by major technology most of them concentrated in its Bay Area service companies and by firms that specialize in developing territory and San Jose in particular.34 and managing data centers. Several prominent In terms of size and function, California’s data centers companies with a significant national and global have tended to be smaller than some of the massive presence—including Meta, Apple, Google, Equinix, “hyperscale” facilities located elsewhere in the United and GI Partners—are headquartered in California. States. Many current and proposed facilities will Most other large operators maintain a substantial support relatively modest computing operations and corporate presence in the state and manage multiple are middling in size, although proposals for much facilities here, reflecting California’s important larger centers—including facilities up to roughly 600 position in the broader data center ecosystem.37 megawatts—have emerged.35 CALIFORNIA DATA CENTER ELECTRICITY DEMAND Nationally, hyperscale data centers have increasingly Electricity demand from data centers has grown been built in rural areas and in states with relatively rapidly and is expected to continue rising. A little direct connection to the tech industry. Reflecting January 2025 report by Lawrence Berkeley National that broader trend, some new facilities are being Individual PG&E Data Center Load Requests as of June 2024 Source: Pacific Gas & Electric Co. 9 | LITTLE HOOVER COMMISSION Data Center U.S. Market Share by Square Footage Source: CBRE Investment Management Laboratory estimates that, by 2028, data centers analysis, peak data center load could rise to between could account for between roughly 7 and 12 percent ~5.3 and 7.5 gigawatts (the chart below uses 2025 as of total U.S. electricity consumption—a dramatic its base year and does not reflect the ~1 gigawatt of increase from historic levels.38 current data center electricity load).40 The CEC’s mid- case projections are broadly consistent with forecasts In California, the state’s investor-owned utilities paint from the California Independent System Operator a steep growth trajectory based on interconnection (CAISO).41 requests from data centers. In February 2025, PG&E reported that data centers could add approximately The CEC’s forecasts are based on the understanding 5.5 gigawatts of new demand between now and that only some utility interconnection requests 2035. By May—just three months later—the figure will come to fruition. Thus the CEC evaluates these had climbed to 8.7 gigawatts. And by late July, PG&E requests using information about their status, stated that it is “proactively working to serve 10 capacity, and expected timelines, and then applies gigawatts of new electricity demand from data center a 67 percent “utilization factor” to reflect that even projects over the next ten years”—an amount that is completed projects are unlikely to operate at full equivalent to four times the generating capacity of capacity.42 Finally, these projections are grounded in the Diablo Canyon nuclear power plant.39 anonymized meter-level data from operating data centers, which provide insights into how facilities To help guide transmission-level and local reliability actually use electricity over time. planning, the California Energy Commission (CEC) provides data center load forecasts that are more This raises the larger issue of data reporting and tempered than PG&E’s investor- and customer- transparency around energy consumption by data oriented utility forecasts. According to the CEC’s centers. Utilities report electricity consumption to 10 | LITTLE HOOVER COMMISSION CEC Data Center Demand Forecast (2025-2040) Source: California Energy Commission the state in aggregated form, reflecting long-standing understand whether certain communities may face assumptions that usage patterns within and across disproportionate environmental or reliability impacts, sectors are relatively stable and predictable. This or call out individual facilities for under-utilizing the approach has typically worked well for most sectors, infrastructure built to serve them. At the same time, where no single customer materially alters statewide any discussion of improved reporting must balance trends. Utilities also provide anonymized 15-minute transparency and accountability with legitimate interval meter data, which allows analysts to infer privacy, competitiveness, and confidentiality data-center consumption patterns to some extent concerns. while preserving customer confidentiality. This issue is closely tied to the recommendations in The CEC has expressed confidence in its ability to this report, many of which call for solutions relating accurately forecast data-center loads based on to precise engineering, load shifting, and intelligent currently available information. However, enhanced use of backup energy generation and storage. reporting may support regulation, localized grid Achieving those objectives may require more fine- planning, and sound policymaking more generally.43 grained, facility-relevant data so that the state can design, target, and evaluate interventions effectively. For example, because the meter-level data the With this in mind, policymakers might consider CEC receives is anonymized, the agency cannot advocating for the following: know with absolute confidence whether this data actually belongs to a data center, which facility it ◊ Enhanced Facility-Level Reporting, With corresponds to, or where it is located. These gaps Privacy Safeguards: Require confidential, make it more difficult to identify localized grid stress, facility-specific reporting of relevant metrics— 11 | LITTLE HOOVER COMMISSION such as load, peak demand, utilization rates, state’s major data center clusters, and trillions of load shapes, and location—under strict dollars are being poured into new construction nondisclosure protections. More precise visibility nationwide.44 would strengthen regulatory oversight and Because access to electricity is widely viewed as the accountability while safeguarding proprietary primary bottleneck to growth, a data center site information. with a secured interconnection has become a highly ◊ Targeted Reporting for Grid Stress and valued asset—capable of serving immediate demand Community Impact: Create enhanced reporting and, potentially, being resold at a premium.45 requirements triggered by specific conditions— Companies also treat data centers as infrastructure such as regional grid congestion, high geographic investments that produce near-term revenue concentration of large loads, or environmental- and offer tax advantages through accelerated justice concerns. Facilities in higher-impact areas depreciation, especially in jurisdictions that pair these would provide more detailed data to help the benefits with targeted incentives.46 state better assess localized risks. ◊ Transparency Pathways That Respect Privacy: While developers hope to build data centers at Adopt a two-tiered transparency model in which breakneck speed and utilities are eager to add regulators receive detailed confidential data while load and upgrade infrastructure, identifying the public receives aggregated, non-identifiable policy solutions to large load grid interconnection reporting. This approach would promote public challenges is slower by nature.47 trust and accountability without compromising Intelligent siting requires careful analysis of competitiveness or security. existing substation and transmission capacity; selective upgrades require engineering studies and Enhanced data center reporting will provide insight environmental review; and the research needed into how individual facilities interact with the grid, to support long-term policy—including studies on affect local reliability, and contribute to infrastructure system impacts, rate design, and load forecasting— costs. Ensuring that regulators have access to takes time. Legislative reforms targeting data centers, essential, confidential facility-level information will still in an unsettled state nationwide, follow their own help them plan responsibly, protect communities measured timeline of querying stakeholders, settling and ratepayers, and guide thoughtful policy decisions on policy positions, and drafting and advocating for moving forward. legislation.48 This mismatch can lead to frustration on the part of developers and, if rushed, poor decision RECOMMENDATION 3: California should enable making on the part of regulators and policymakers. regulators to confidentially access essential facility- level electricity-use data from large data centers, Indeed, the data center “gold rush” has already led supported by structured data-sharing among to problems for energy regulators. For example, to agencies, to improve planning, assess localized preserve strategic flexibility, data center developers impacts, and ensure fair cost responsibility while sometimes overstate the expected energy protecting sensitive business information. requirements of proposed projects and file multiple interconnection requests for the same project or for THE DISPARATE PACE OF ENERGY POLICY several potential sites.49 Because the cost of filing is AND THE CONSEQUENCES relatively low, and because utilities may benefit from The extraordinary pace of investment in artificial the possibility of higher future load, these duplicative intelligence and the data centers that support it has applications can accumulate quickly.50 intensified pressure on California’s grid. Vacancy rates for existing facilities are low, especially in the 12 | LITTLE HOOVER COMMISSION Some states—such as Texas—have begun requiring. 3. Local Permitting, Water Use, & CEQA Review: developers to disclose overlapping requests and Cities and counties manage zoning approvals, to demonstrate ownership or control of land water-use permits, and building requirements. associated with each application.51 In Ohio, a new Most large data centers also undergo CEQA review, requirement to commit to purchasing a hefty which now routinely evaluates air emissions, noise, proportion of proposed new load has already led to water consumption, local ecological impacts, and— a marked decrease in the number of interconnection notably—energy use and the project’s likely demand applications from data centers—a sign, policymakers on the grid.56 say, that the tariff is achieving its intended goal.52 4. Statewide Regulatory Influence & Incentives: DATA CENTERS AND JURISDICTIONAL Although the state does not directly approve siting POLICY LEVERS decisions, it shapes the broader environment for The state does not typically determine whether data-center development through rates, cost- a data center may be built, but it can shape the allocation rules, data-reporting requirements, conditions under which a project moves forward. energy-efficiency standards, and statewide demand Prior to starting construction, developers must satisfy forecasting. Some states also offer financial a number of utility and governmental requirements. incentives such as sales-tax exemptions for These requirements represent potential policy levers construction and equipment purchases, typically tied through which state government can influence to minimum investment or job-creation thresholds.57 where, how, and under what conditions data centers are developed.53 The following categories summarize The state’s authority over whether individual data the primary regulatory and permitting interactions centers are built is thus relatively limited. Instead, its required for new development: influence is exercised indirectly through electricity rates, reporting requirements, load-forecasting 1. Utility Interconnection & Transmission processes, and regulatory mechanisms that guide Planning: Developers must submit interconnection utility planning and shape the costs and conditions requests to utilities, specifying expected load and under which new projects are developed. Together, triggering engineering studies and any required these touchpoints form the practical extent of infrastructure upgrades.54 Large projects that depend California’s policy leverage over data-center on new or expanded transmission may also intersect expansion. with CAISO’s Transmission Planning Process, even DATA CENTERS AND CALIFORNIA’S though CAISO is not a state agency. This differs REGULATORY AND PLANNING BODIES from the California Energy Commission’s role in California relies on several agencies—along with the statewide load forecasting, which informs—but does California Independent System Operator—to manage not replace—CAISO’s engineering-level transmission the policies, planning processes, and regulatory tools planning. that shape how large new electricity loads such as 2. Air Quality & Backup Generation: Backup- data centers are integrated into the grid.58 Each entity generator emissions must be permitted by local air plays a distinct role, from regulating utilities and districts, which enforce limits on diesel engines and overseeing rates to forecasting statewide demand related equipment. CARB sets statewide emissions and planning transmission infrastructure. standards and guidance, but it does not issue permits directly; day-to-day oversight occurs at the district level.55 13 | LITTLE HOOVER COMMISSION California Regulatory Agency Overview Core Policy Levers Entity Responsibilities Relevant to Data Centers California Regulates investor-owned utilities ◊ Designs and approves rate structures and Public Utilities (IOUs); oversees rates, tariffs, customer classes (including those affecting large- Commission cost-recovery mechanisms, utility load customers). (CPUC) procurement, capital investments, and implementation of statutory ◊ Approves utility interconnection policies (e.g., requirements. Rule 30). ◊ Approves IOU capital projects needed to serve large loads (substations, feeders, related upgrades). ◊ Can require reporting frameworks and data submission by utilities. ◊ Oversees programs that may shape costs (demand-response, load-management, etc.). California Conducts statewide demand ◊ Produces load forecasts that guide CAISO’s and Energy forecasting; oversees power-plant IOUs’ long-term transmission planning. Commission siting; sets building and appliance ◊ Reviews new power-plant siting that may support (CEC) efficiency standards; conducts system- grid capacity. level planning studies. ◊ Establishes efficiency standards affecting baseline building loads. ◊ Facilitates coordination with utilities (IOUs and POUs) to understand new large loads, including data centers. California Plans and operates most of the ◊ Determines whether and when large new loads Independent state’s transmission grid; conducts can be accommodated. System the Transmission Planning Process ◊ Manages interconnection processes for large Operator (TPP); oversees transmission-level projects requiring transmission-level service. (CAISO) interconnections. ◊ Uses CEC forecasts and utility data to prioritize planning decisions. California Air Sets statewide air-quality and climate ◊ Establishes overarching requirements that Resources standards; oversees statewide govern emissions sources relevant to data-center Board (CARB) emissions rules and the activities of air- equipment (e.g., generators, stationary engines). quality regulators. ◊ Issues climate-related regulations (e.g., reporting, diesel-engine standards) that shape what equipment data centers can use. 14 | LITTLE HOOVER COMMISSION CURRENT DATA CENTER RATE POLICY of rising electricity costs—including significant Under California’s current rate design, large data wildfire mitigation investments and other expenses centers generally pay lower electricity rates than traditionally embedded in distribution rates— most other customers because they typically are borne primarily by customers served at the interconnect at the transmission level and do not distribution level. Although transmission-level data rely on the local distribution system. This structure centers do contribute to some of these costs through is broadly consistent with the principle of “cost embedded charges, they do not shoulder the same causation,” under which “[r]ates for each class of level of distribution cost responsibility as other customer are proportionate to the costs of serving customers. that class of customer”.59 While new large loads may help spread certain fixed Most large data centers receive service at costs related to generation, transmission, and public transmission-level voltages. Because they do not purpose programs, they are unlikely to offset the use the lower-voltage distribution network, they distribution-level cost pressures that have been are not assigned the full suite of costs associated central to rising residential and small-business bills. with building, operating, and maintaining that This raises an important policy question for the system. However, as PG&E notes, transmission-level state: whether California is comfortable with this customers are not entirely exempt from distribution- distribution of costs, or whether a different allocation related charges. Within PG&E’s B-20T rate, for approach is warranted given today’s affordability and example, customers pay a customer charge that equity concerns. recovers certain metering and billing costs, as well FEATURES OF A NEW VERY-LARGE-LOAD as a distribution demand component that includes TARIFF some wildfire-related and demand-response costs.60 In the sections that follow, this report outlines several steps California can take to protect ratepayers Even with these protections in place, customers while managing rapid load growth due to data taking service at the transmission level still pay less centers. A primary element of this strategy is the than comparable customers receiving distribution creation of a very-large-load tariff tailored to the service. As of September 1, 2025, customers in unprecedented scale and financial risks associated PG&E’s B-20 “transmission firm” class pay roughly 10 with data centers and similar high-demand facilities to 18 cents per kilowatt-hour (depending on time of (for example, cryptocurrency mining).62 Here, we use), plus a monthly demand charge—a separate fee describe the critical elements that the Commission based on the highest level of power the customer believes should anchor such a tariff, which we believe draws from the grid during a billing period.61 the California Legislature should make statutory By contrast, industrial “secondary firm” customers— requirements. those served at the distribution level—pay ◊ Creditworthiness: A very-large-load tariff should approximately 12 to 21 cents per kilowatt-hour and require enhanced due diligence on the financial face a considerably higher demand charge. While this stability of data center developers and their difference may appear modest, it amounts to at least parent companies, including evaluation of long- a 14 to 17 percent discount for transmission-level term business viability, financing sources, and the service. Given the enormous electricity requirements likelihood that proposed loads will materialize and of large data centers, this results in a meaningful persist. This may include bonding requirements reduction in the overall revenue these customers or other forms of financial assurance designed to contribute to the system. protect utilities and ratepayers against default, This matters in the context of California’s broader abandonment, or significant downsizing of affordability challenges. Many of the main drivers planned facilities. 15 | LITTLE HOOVER COMMISSION ◊ Prepayment for Infrastructure: To reduce It should be noted that some of these concepts financial exposure for utilities and prevent shifting already appear in PG&E’s proposed Electric Rule risk onto other customers, the tariff should 30, which was approved by the CPUC on an interim require upfront infrastructure payments. These basis in July 2025 while its final form is still being payments would help fund the build-out of new considered.63 This tariff establishes requirements transmission, distribution, and substation facilities for transmission-level retail customers seeking to needed to serve data centers, ensuring that connect to the grid; historically, such customers utilities are not left with stranded assets if projects were handled on a case-by-case basis, leading to stall, downsize, or fail to materialize. uncertainty and long interconnection delays. It seeks to create a standardized framework for processing ◊ Cost Recovery: The tariff should establish a transmission-level interconnection requests while special rate structure tailored to very large incorporating measures to protect ratepayers.64 loads, recognizing that data centers impose unprecedented planning, infrastructure, and Under the currently-approved interim structure of reliability demands. This rate should be structured Rule 30, transmission-level customers—including to ensure meaningful contributions to overall data center developers—must provide upfront system costs, including transmission, distribution, funding for the transmission facilities needed to and reliability investments, while also ensuring serve them, including full pre-funding of network that these customers help fund California-specific upgrades if they wish to proceed during the interim policy costs such as wildfire mitigation and climate period. However, the CPUC has explicitly deferred resilience programs. decisions regarding any refunds, repayment of ◊ Minimum Commitments: The tariff should pre-funded amounts, and whether interest should include minimum payment obligations requiring accrue. As a result, developers must proceed customers to pay for a defined percentage of with the understanding that repayment is not their proposed demand regardless of whether guaranteed at this stage. Other tariff design elements their full projected load ultimately develops. Such and contractual protections also remain under provisions would align incentives, discourage deliberation and will be addressed in the CPUC’s final speculative requests, and ensure a stable revenue decision. base to support system investments made in anticipation of these loads. Policymakers may wish to build on lessons emerging from the interim implementation of Rule 30 and ◊ Load Flexibility and Curtailment: A very-large- the insights gained once it is finalized to guide the load tariff should require customers to design development of a statewide statutory framework facilities and operations so they can meaningfully for very-large-load customers. For example, the reduce or shift demand during periods of system Legislature could establish a two-year study period stress. This could include participation in demand- during which the CPUC evaluates the effectiveness of response programs, contractual curtailment Rule 30’s provisions in areas relating to cost recovery, obligations, or other flexibility mechanisms that risk allocation, infrastructure financing, customer help minimize the need for costly new capacity commitments, and overall system reliability. Findings and support grid reliability. from this study period would then inform the ◊ Exit Protections: Finally, a very-large-load tariff design of a statewide tariff that ensures very large should incorporate exit fees to address early loads contribute fairly to system costs and supports termination, significant downsizing, or withdrawal. California’s broader energy and climate goals. These provisions are essential to preventing cost shifts to other customers and ensuring that utilities can recover investments made in reliance on customer commitments. 16 | LITTLE HOOVER COMMISSION RECOMMENDATION 4: California should ensure California should similarly treat the interconnection that data centers pay for costs imposed on the stage as a structured policy moment—one that electric system through tariffs and cost-allocation requires developers to provide essential information, policies that recover infrastructure and grid-services commit to cost, reliability, and transparency costs in a manner that benefits rather than burdens obligations, and align their projects with the state’s other ratepayers. broader public-interest objectives before gaining access to the grid. IMPROVING THE INTERCONNECTION PROCESS In return, California should support utilities in California’s utilities are legally obligated to provide speeding up interconnection request review and service to customers who request it, but the recent approval. Here are some ways this could be achieved: wave of extraordinarily large interconnection requests—that is, requests to draw energy from the ◊ Streamlined Permitting & Environmental Review: Strengthen and coordinate permitting grid—from data center developers has presented a processes through state facilitation, improved unique challenge in terms of scale and complexity.65 agency alignment, and reduced duplication The interconnection stage is a critical policy and so projects can advance more quickly while planning checkpoint where developer needs and maintaining environmental protections. the state’s broader public-interest responsibilities ◊ Clearer, Faster Interconnection Studies: Expand can align. Properly designed, interconnection engineering capacity, adopt more standardized requirements—including those associated with Rule study approaches, and improve queue discipline 30—can both help utilities provide faster, more so credible projects move more quickly through coordinated review and ensure that developers make analysis. meaningful commitments before gaining grid access. ◊ Faster Transmission & Distribution Upgrades: California can look to other states for examples. Better coordinate state planning around expected In June 2025, Texas enacted Senate Bill 6, explicitly data-center growth and give utilities clearer directing regulators to process large-load permission and safeguards so they can start interconnection requests “in a manner designed building needed grid upgrades sooner without to support business development in this state putting ordinary customers at financial risk. while minimizing stranded infrastructure costs and ◊ Supply Chain & Equipment Support: Use state- maintaining system reliability.”66 In doing so, Texas led bulk procurement, incentives for in-state has made the interconnection stage a primary vehicle manufacturing, and permission for proactive for regulating large-load growth. purchasing so difficult-to-acquire equipment does not become a structural bottleneck. The Texas law closely parallels many of the ◊ Workforce & Construction Capacity: Invest in principles reflected in California’s proposed Rule accelerated training pipelines, partnerships with 30.67 Developers of facilities exceeding 75 megawatts unions and higher education, and standardized must cover study costs and fund needed grid designs to speed construction and reduce upgrades up front, with reimbursement available specialized engineering constraints. only once projects become operational or capacity is reassigned. In addition, SB 6 requires developers to disclose whether they have made additional By adopting these supports, California can strike interconnection requests inside or outside Texas a balance between setting responsible guardrails and demonstrate proof of site control. Finally, the for large-load interconnections and enabling data legislation authorizes curtailment provisions during centers to connect to the grid and begin operating periods of system stress.68 more quickly. 17 | LITTLE HOOVER COMMISSION RECOMMENDATION 5: While maintaining Technological innovation could also reduce the a rate structure that ensures shared costs and amount of energy required by data centers, a high standard of regulatory review, California potentially stranding investments in oversized or should accelerate data-center interconnection and unnecessary infrastructure. In terms of hardware, permitting decisions to give developers greater NVIDIA has reported a 4,000-fold performance-per- predictability in planning and investment. watt improvement in GPUs over a decade.75 And Google’s tensor processing unit (TPU) AI chips are MITIGATING RISK FROM STRANDED ASSETS reportedly three times more energy efficient than The tech sector’s economic strength has significant comparable models.76 fiscal consequences for California—a fact that has helped drive the Governor’s support for generative AI training has also become much more efficient. AI through a variety of initiatives and legislative China’s DeepSeek-R1 reportedly achieved high decisions.69 Despite uncertainty about AI’s long- performance with a small fraction of the hardware term economic value, firms continue to project and energy of comparable models;77 UC Berkeley’s rapid growth. California-based NVIDIA, for example, NovaSky team trained a highly capable reasoning reported roughly $32 billion in Q3 2025 revenue and model in 19 hours for roughly $450;78 and industry now exceeds $5 trillion in market value. And major leaders such as OpenAI, Google, and Anthropic now firms like Microsoft, Google, Apple, and Amazon routinely release “mini” models that deliver strong also report strong AI-linked profits and valuations.70 performance with significantly less compute.79 Analysts project up to $3 trillion in global data-center Some observers compare the present moment to investment by 2029—much of it financed through past infrastructure booms and busts, with railroads borrowing—and note that AI spending is propping and fiber optics being the most commonly cited up an otherwise sluggish economy.71 The frenzy to examples. Both were initially overbuilt but ultimately invest in AI and data centers has raised concerns produced lasting public benefits.80 Witnesses at the that utilities may overbuild grid infrastructure in November hearing similarly noted that data-center response to speculative demand, leaving ratepayers development could accelerate needed upgrades to responsible if anticipated load does not materialize. high-voltage transmission.81 However, should these One JP Morgan analysis estimated that achieving facilities close, relocate, or become obsolete as a 10 percent return on an expected $5 trillion in AI technology advances, ratepayers could be left paying investment would require $650 billion in annual for “stranded assets”—power plants, substations, or revenue, a level many view as unlikely.72 A major transmission lines that no longer serve customers.82 market correction could result in data center developers going bankrupt or a dramatic reduction in As California considers building significant new data center operations generally. energy infrastructure to support data centers, policymakers must therefore weigh benefits against There is also concern that data centers could leave longer-term financial risks. Some action has already the state, leaving behind long-term infrastructure been taken by the state to protect against this. investments that no longer serve their purpose. Prepayment for transmission infrastructure required Servers are portable, and the centers that house under PG&E’s Rule 30 is one example; another is the them are essentially storage facilities with CEC’s use of utilization factors and confidence levels sophisticated energy and cooling systems.73 While to develop a more realistic picture of which proposed Silicon Valley is the national—if not global—hub infrastructure projects are likely to be built. of the tech industry, its companies have often threatened to leave California for jurisdictions with more favorable incentives, tax structures, regulatory climates, or cheaper water and electricity.74 18 | LITTLE HOOVER COMMISSION A number of other states—including Ohio and also support broader reliability, clean-energy Texas, discussed above—have taken steps to protect integration, electrification, wildfire resilience, or against the risk that infrastructure built to serve other public benefits. This ensures that even if data centers could become stranded if projected data-center demand softens, Californians still load does not materialize. These actions have taken benefit from the infrastructure built. several forms that California policymakers may wish ◊ Clarify that prudence applies to both to consider. underbuilding and overbuilding: Regulators must explicitly evaluate the risk of unnecessary or For example, Georgia, South Carolina, Oregon, oversized infrastructure investments along with and Virginia have created or are considering very- the risk of insufficient capacity. large-load tariffs designed to better ensure that data centers bear the costs of the infrastructure required to serve them.83 These tariffs typically rely RECOMMENDATION 6: California should on mechanisms such as minimum demand charges establish consistent, statewide safeguards that or revenue commitments, longer contract terms, ensure large data-center customers bear appropriate and higher fixed charges, reducing the likelihood financial risk, while making certain that major grid that utilities will build long-lived assets based on investments continue to serve the public even if data speculative demand. center demand evolves differently than has been predicted. Other states, including Minnesota, have adopted BROADENING DATA CENTER OVERSIGHT more direct ratepayer protections.84 These While investor-owned utilities operate under direct approaches include statutory prohibitions on CPUC oversight, publicly owned utilities play an passing stranded-asset costs on to ratepayers, important role in serving complex, energy-intensive requirements for heightened regulatory review loads. However, current state policy tools do not when utilities seek to recover costs associated with provide the same level of transparency, planning large loads, and policies that assign upgrade costs to alignment, or cost-risk protections for customers in the customer that triggers the investment. these territories. As California faces unprecedented In addition to drawing on lessons from these states, electricity demand driven by rapid data center California has further options available to ensure growth, uneven standards across different utility that ratepayers are not left covering the costs of types may create avoidable risks: uncertainty infrastructure that ultimately does not serve new in statewide planning, inconsistent community load. Potential steps include: protections, and potential cost shifts if major projects fail to materialize as expected. ◊ Strengthen financial accountability for very large or mobile loads: Require stronger A recent example can be found in the City of creditworthiness reviews, collateral, minimum Imperial, which is served by the publicly owned load commitments, and/or exit fees so that Imperial Irrigation District.85 County officials are developers carry real financial risk if demand fails considering approval of a proposed 330-megawatt to materialize. Doing so will better align private data center, but city officials and local residents have incentives with the public interest and discourage stated that they were not adequately consulted and inflated energy requests. that environmental review was insufficient. While the developer has asserted that the facility would rely on ◊ Prioritize infrastructure that retains value renewable energy and recycled water, these claims even if AI demand slows or shifts: When have been disputed, and it is unclear whether such large investments move forward, they should commitments would be legally binding. 19 | LITTLE HOOVER COMMISSION The goal is not to displace local control. Rather, California should establish clear statewide expectations—particularly around reporting, reliability assurances, cost-recovery protections, environmental impacts, and coordination with agencies such as the CEC, CPUC, and CAISO—so that every community benefits from the same level of protection and every utility operates within a consistent framework. Potential minimum statewide standards could include: ◊ Financial Risk Protections: Extend versions of protections evolving under Rule 30 statewide— such as prepayment, financial security instruments, or clear cost-recovery assurances—to reduce stranded-asset risk regardless of utility type. ◊ Baseline Reporting Requirements: Require consistent, confidential reporting of critical data center metrics (e.g., anticipated and actual load, load shape, backup generation information) across all utility types to support statewide planning and risk assessment. ◊ Environmental and Community Impact Guardrails: Establish minimum standards for transparency relating to backup generation transparency and community engagement regardless of jurisdiction. California’s approach to managing very large loads should not depend on where a data center is located or which utility happens to serve it; residents across the state deserve the same level of financial protection, planning rigor, and environmental safeguards. Establishing clear, statewide standards will help guarantee that data center risks are handled consistently. RECOMMENDATION 7: The State should establish consistent statewide oversight expectations for large data center developments in territories served by publicly owned utilities, municipal utilities, irrigation districts, and community choice aggregators, ensuring that risks to reliability, costs, and local communities are addressed with the same level of rigor as investor-owned utilities. 20 | LITTLE HOOVER COMMISSION Data Centers and Grid Planning This section explores how California can integrate daily life and pose serious risks to public safety.88 new data centers more efficiently and at lower California’s energy planners have therefore focused cost by making smarter use of the generation and on strengthening the grid’s ability to perform during transmission capacity it already has. It begins by these critical periods, and have recently expressed acknowledging two important realities: there are confidence in the system’s resilience—even as rising legitimate concerns about grid reliability as large new temperatures drive sharper and more frequent loads come online, and at the same time, there is demand spikes.89 meaningful untapped capacity within the system that Actions taken include adding over 20 gigawatts could help meet this demand if used wisely. of new energy supply; the addition of over 12 Recognizing both facts—and designing gigawatts of battery storage (including residential) to complementary policy responses to address risk complement variable energy sources like wind and while unlocking available capacity—will be essential solar; establishing a Strategic Reliability Reserve and to ensuring that data center growth does not coordinating with the Western Energy Imbalance undermine reliability. From that foundation, the Market to improve reliability; and extending section then examines three strategies: making operations at the Diablo Canyon nuclear power targeted upgrades to existing infrastructure, plant. This has allowed the grid to remain operational encouraging data centers to locate where the grid despite record setting heat events.90 can most readily support them, and promoting Significant progress has been made in strengthening demand flexibility to shift or shape energy use and the grid’s ability to meet peak demand. Looking relieve stress on the system. ahead, however, the rapid growth of data centers DATA CENTERS AND GRID RELIABILITY will add substantial new load, and the industry’s Planning for a reliable electric system requires strong emphasis on avoiding downtime creates preparing for the periods of highest demand— a natural tendency to maintain operations even the “peak” moments on the hottest summer during periods of system stress. While this approach days or during extreme weather events, when supports facility-level reliability, it can strain the air-conditioning and other essential uses place broader electric system when aggregate demand maximum pressure on the grid. These moments approaches—or exceeds—available peak capacity. represent a very small share of overall electricity As a result, concerns about grid reliability are use but nevertheless determine the scale of increasingly shaping data-center policy decisions in infrastructure that must be built and maintained.86 As other states facing rapid large-load growth. energy expert Sean Fleming has written, the “system Texas offers a clear illustration. During Winter is actually pretty under-utilized by design, with spare Storm Uri in 2021, electricity demand surpassed capacity more than 99.99% of the time. Over the last available capacity, triggering widespread outages and 5 years, California’s average load has been just ~50% underscoring the risks of unmanaged peak load. In of its [2022] peak”.87 response, the state enacted SB 6, which authorizes In other words, this excess capacity is a deliberate both voluntary and mandatory curtailment of data- feature of the grid. Electricity systems must be built center load during emergencies, giving grid operators to meet infrequent moments of peak demand, a backstop to protect systemwide reliability and because when demand exceeds available capacity, public safety.91 the result can be power outages that disrupt 21 | LITTLE HOOVER COMMISSION There are several ways to reduce the need for new loads, particularly as demand—and potential system infrastructure by making better use of the grid’s stress—grow in the years to come. existing capacity, which are discussed in the sections RECOMMENDATION 8: California should ensure that follow. At the outset, however, it is important that new large data center loads are integrated in a to emphasize that maintaining grid reliability must manner that preserves system reliability and does remain a core priority. Planners and policymakers not disadvantage existing customers. have meaningful opportunities to shape data center policy in ways that support system reliability while MAXIMIZING EXISTING GRID CAPACITY accommodating continued load growth. Policy One of the points the Commission heard repeatedly options could include: is that, most of the time, California’s electric system is not “at capacity”. This means that new large loads ◊ Mandatory Curtailment Authority During may be added to the grid without exceeding system Grid Emergencies: Require large data centers limits if development is planned thoughtfully and to reduce load when grid operators declare guided by clear principles.92 Several strategies were emergencies, with clear triggers, enforcement identified that make better use of existing grid mechanisms, and penalties to ensure compliance capacity, reducing costs by improving efficiency supports system reliability. rather than requiring new infrastructure. These are ◊ Reliability-Based Interconnection Conditions: outlined below. Condition approval of large-load interconnections based on enforceable reliability obligations, Targeted Infrastructure Upgrades: Research ensuring that data centers are prepared to conducted at Stanford University has identified respond to periods of system stress as a bottlenecks in the Western Energy Coordinating prerequisite for connecting to the grid. Council region—the interconnected electric grid ◊ Phased or Probationary Load Approval: Allow covering most of the western United States, regulators or utilities to approve large data including California.93 While transmission lines center loads in stages, with subsequent phases often operate below their rated capacity, their contingent on demonstrated performance and ability to deliver power is often constrained by large compliance with reliability requirements. power transformers, which regulate how much electricity enters and exits those lines. Many of these ◊ Centralized State Oversight of Large-Load transformers have met or exceeded their design life Reliability Impacts: Establish coordinated, and now represent a critical limiting factor. According statewide review of the cumulative reliability to the Stanford report, “[t]argeted replacement, impacts of large loads across utility territories to refurbishment, and added flexibility at these avoid fragmented decision-making and system- transformer interconnection points can therefore wide risk. unlock additional use of existing line capacity ◊ Clear Enforcement and Accountability and increase transfer capability.” It also finds that Framework: Define unambiguous authority, congestion and reliability risks are often driven by a reporting requirements, and penalties related relatively small number of overutilized lines. to reliability obligations so that curtailment and compliance are predictable, transparent, and Upgrades at these critical points in the grid could credible. expand—and are already expanding—overall system capacity and enable additional generation to be connected without requiring the large, Together, these policy actions reinforce the principle costly, and time-consuming expansion of the that grid reliability must remain a primary condition broader transmission network.94 The Commission for integrating new large data center therefore recommends that California require 22 | LITTLE HOOVER COMMISSION utilities to evaluate and prioritize targeted, asset- constraints, water availability, and other relevant level upgrades—such as transformer replacement, factors would allow developers, policymakers, refurbishment, or added operational flexibility— regulators, and community members to engage in before proposing large, costly transmission siting discussions from a common factual foundation. expansions. This expectation could be reinforced by Such a tool would not dictate outcomes, but would aligning cost-recovery rules so that efficient, capacity- support more informed, efficient, and transparent unlocking upgrades receive expedited or preferential decision-making across the many forums in which treatment, while major new infrastructure is subject data center siting is considered. to greater scrutiny when lower-cost alternatives have Cluster Studies: Most data center interconnection not been fully pursued. requests in California are concentrated in just a few Intelligent Siting: Developers, policymakers, and geographic areas. When multiple large load requests regulators share a common interest in locating data are in close proximity, they may be more efficiently centers in ways that control costs, make effective addressed through one “cluster” study that evaluates use of existing grid capacity, and limit environmental them together rather than through individual studies. and community impacts. Relevant considerations A cluster study can examine the combined demand include local grid conditions and electricity rates, of multiple facilities and identify coordinated, system- permitting and regulatory feasibility, grid reliability level solutions that can accommodate projected and flexibility, and access to resources such as load growth more efficiently and at lower cost. For water and fast fiber optic cables.95 Siting decisions example, in its original 2024 cluster study, Pacific are particularly consequential because the cost of Gas and Electric Company evaluated 740 megawatts upgrading the grid to serve data centers in congested of proposed electricity demand from multiple data areas can be substantial. For example, according center developers as a single, combined load.98 to the Public Advocates Office, recently-approved California could encourage more efficient batch transmission upgrades to accommodate 2.5 or cluster studies through a set of targeted, low- gigawatts of new load in the South Bay Area will cost disruption policy steps that focus on clarity and over $2 billion.96 predictability rather than mandates. The CPUC could This illustrates how alternative siting decisions— explicitly authorize and encourage utilities to use particularly in areas with readily available cluster studies when multiple large interconnection grid capacity—can reduce the need for costly requests emerge in close geographic or temporal infrastructure upgrades. One way developers and proximity, supported by objective triggers that signal policymakers are working to balance the many when a batch approach is appropriate. factors involved in data center siting is through At the same time, utilities could be directed to clearly data-driven mapping tools. For instance, the National communicate when cluster studies are likely, how Laboratory of the Rockies has developed maps that timelines and cost allocation would work, and what integrate infrastructure and resource constraints, developers should expect, reducing uncertainty and while Esri promotes the use of its ArcGIS platform concerns about ad hoc delays. Issuing this direction to help identify optimal data center locations by as guidance or a policy statement—rather than a evaluating a wide range of pertinent variables.97 rigid rule—would help normalize cluster studies as Building on these examples, California could play a a best practice while preserving flexibility for utilities constructive role by facilitating access to similarly and developers alike. comprehensive, high-quality siting data in a shared, RECOMMENDATION 9: California should user-friendly format—such as a statewide map or maximize the use of existing grid capacity when dashboard. Bringing together information on grid considering data-center siting and new infrastructure, capacity and congestion, electricity rates, permitting supported by transparent data and visualization 23 | LITTLE HOOVER COMMISSION tools that help developers and policymakers identify Demand flexibility can be especially valuable on a locations where capacity is available or can be seasonal basis during periods of peak demand. In efficiently expanded. California, the grid is often most stressed during hot summer months, when air-conditioning use drives DEMAND FLEXIBILITY electricity demand to its highest levels. In California, Demand flexibility—also called demand response incentive-based programs such as PG&E’s Flex or load shifting—refers to a data center’s ability to Connect Pilot are designed to encourage voluntary adjust electricity use in response to grid conditions, reductions in load during critical periods.103 particularly during periods of high demand or system stress.99 This may include temporarily Flexibility during peak periods may be achieved reducing electricity consumption or rescheduling through a combination of operational adjustments, energy-intensive activities to times when the grid scheduling decisions, and, in some cases, temporary is less constrained. Demand flexibility is important reliance on backup power resources. While backup because the electric system is built to meet peak generation can support grid reliability, its use also loads, meaning that substantial capacity is installed raises air-quality and environmental considerations, to serve relatively infrequent periods of highest highlighting the importance of aligning demand- demand. When large new loads can respond to these flexibility strategies with California’s broader conditions, they can often be integrated into the climate and community-protection goals discussed existing grid with fewer infrastructure upgrades than elsewhere in this report. would otherwise be required. In addition to shifting demand over time, some data- Demand flexibility can take different forms, ranging center operators may have opportunities to shift from voluntary, price-responsive reductions to more load geographically.104 Companies with networks of structured or contractually obligated curtailment. facilities across multiple regions or time zones may While not all data-center workloads can be shifted be able to route certain computing tasks to locations without affecting operations, even partial flexibility with available capacity or lower contemporaneous can meaningfully reduce peak pressure on the grid demand. In theory, this type of geographic flexibility and limit the need for costly new infrastructure. For could help smooth demand and reduce strain example, a Duke University study found that up to 76 on regions experiencing peak conditions. But in gigawatts of new electricity load—roughly 10 percent practice, it requires significant coordination, robust of current U.S. aggregate peak demand—could be infrastructure, and operational redundancy, and may accommodated if new loads are reduced, on average, therefore be more feasible for large, multi-regional for just 0.25 percent of their maximum annual firms than for smaller or single-site operators. operating hours. Moreover, in the vast majority of cases, only a 50 percent reduction in energy use is Taken together, time-based and geographic demand required.100 flexibility offer practical ways to align data-center electricity use with real-world grid conditions. These One important form of demand flexibility is time- approaches can improve utilization of existing based load shifting.101 On a daily basis, some data infrastructure, support system reliability, and reduce centers may be able to move certain computing the likelihood that the costs of serving new large tasks to off-peak hours. Shifting demand in this way loads are shifted onto residential and small-business can ease stress during the most constrained times customers. while improving utilization of grid capacity that would otherwise go unused. Improved load balance can RECOMMENDATION 10: California should also reduce operational challenges associated with maximize load shift opportunities to use existing load fluctuations in generation, which is particularly useful capacity and lower costs for residential customers. as California continues to integrate large amounts of renewable energy.102 24 | LITTLE HOOVER COMMISSION Building a Cleaner, Smarter Data Center Future California’s clean-energy and climate goals may come could pair stronger environmental standards with under new pressure as data centers expand rapidly streamlined processes for projects that meaningfully within the state. These facilities promise economic support decarbonization; ensure that communities and technological benefits, but they also raise serious hosting new infrastructure are protected and share concerns about emissions, local environmental the benefits; and encourage innovation in flexible impacts, and staying true to the state’s climate load, cleaner backup systems, and more accurate commitments. This section outlines environmental carbon accounting. By doing so, the state can support challenges and policy solutions to help ensure economic growth while reinforcing its leadership data-center growth is consistent with California’s on climate, environmental justice, and clean-energy sustainability values. It concludes by highlighting transition. ways the state can support research and innovation RECOMMENDATION 11: California should to help reduce the negative impacts of data centers meet rising data-center electricity demand without on society. weakening its clean-energy or climate goals, including DATA CENTERS AND THE ENVIRONMENT behind-the-meter generation. California has committed to achieving 100 percent CLEAN AND RELIABLE BACKUP zero-carbon electricity generation (with offsets) by GENERATION 2045.105 These nation-leading goals are laudable, The question of backup power for data centers and the Commission believes that the Legislature presents a challenge both for grid stability and public should formally recommit to them even as it works health. Many top-tier facilities are designed to meet to integrate a growing number of data centers into extremely high availability standards—often referred the grid. Doing so will help ensure these facilities to as “five nines” reliability (99.999 percent), meaning rely on cleaner energy rather than the more carbon- only minutes of allowable downtime per year.106 intensive mixes common elsewhere, while showing To achieve this, operators rely on highly stable and other states that it is possible to expand data-center redundant power systems. capacity and still stay firmly on track toward climate goals. At present, diesel is the most common form of backup generation for data centers.107 These systems Looking more broadly, the chart that follows are typically designed to operate during power summarizes some of the environmental challenges outages or to provide relief for the grid during times associated with data centers, how they intersect with of high demand. The Environmental Protection electricity policy, and the questions policymakers Agency recently “clarified” that such generators could should consider as they craft legislation. Addressing be run for a certain number of hours under non- these issues proactively can help California harness emergency situations and for an unlimited amount of the benefits of data centers while reducing harms, time during emergencies.108 Backup generators must strengthening community protections, and advancing also be run periodically for testing and maintenance. the state’s broader sustainability goals. And yet, as Masheika Allgood and Linda Taub Gordon California should integrate data centers into the testified before the Commission, even limited grid in ways that accelerate, rather than dilute, the operation of diesel generators can contribute to state’s climate ambitions. For example, policymakers 25 | LITTLE HOOVER COMMISSION Data Centers and Environmental Challenges Environmental Connection to Electricity Policy Potential Policy Questions Challenge Impacts Energy infrastructure for data centers ◊ How should regulators assess environmental- on Nearby may affect communities through air or justice impacts when approving new Communities water pollution, land use, or noise. infrastructure for data centers? ◊ Could affected communities receive shared benefits such as local hiring or energy-efficiency investments? Where Data Siting data centers in areas with more ◊ Should siting policies or interconnection Centers Get clean energy capacity reduces total approvals favor utilities with cleaner electricity Their Power emissions and is more sustainable. supplies, or else require developers to offset emissions in dirtier areas? Transparency Public reporting of data-center energy ◊ Should large data centers be required to disclose About Energy use and emissions remains limited. annual energy use, energy source mix, and Use and emissions figures? Emissions ◊ Could this information feed into the state’s broader climate-tracking and planning tools? Building and New substations or transmission ◊ How can environmental review processes be Connecting to lines for data centers can trigger streamlined for projects powered by clean energy the Grid environmental review and permitting while still protecting communities? delays. Backup Generators used for reliability emit ◊ Should the state require cleaner backup systems? Power and Air carbon dioxide and local pollutants. ◊ Could utilities offer programs that incentivize Pollution data centers to provide clean backup power to the grid? Making Energy Data centers can help integrate ◊ Could the state create time-flexible rate options Use More renewable power by adjusting when for data centers that shift operations to periods Flexible they draw electricity or by participating of abundant renewable energy? in demand-response programs. ◊ Should this flexibility count toward grid- decarbonization goals? Cleaner Building materials and equipment ◊ Should low-carbon construction standards or Construction manufacturing for data centers have procurement rules apply to data center projects and Materials large carbon footprints. that benefit from state-supported infrastructure? 26 | LITTLE HOOVER COMMISSION local air pollution and raise public health concerns, curtailment during grid stress (including extreme particularly in communities already burdened by heat events), and reduce reliance on fossil-fuel poor air quality.109 backup generation. This leads to a complex challenge: backup generation RECOMMENDATION 13: The state should limit is necessary for data centers to function and to pollution impacts on nearby communities from data- support demand flexibility. But diesel-powered center backup generators. backup generators have harmful effects on health, INCENTIVIZING DATA CENTER ENERGY and data center operators are skeptical about the INNOVATION use and cost of variable generation resources (e.g., California’s data center sector is rapidly evolving, and solar, wind) and less proven technologies such a range of emerging technologies can help reduce its as battery storage, geothermal, or small modular impacts while supporting reliability, environmental reactors.110 goals, and community well-being. These innovations can make data centers more efficient, less resource- The Little Hoover Commission believes that data intensive, better aligned with clean energy, and center developers and the tech industry should more responsive to grid needs. Together, they offer drive solutions to dirty backup generation, and that a practical pathway for enabling economic growth California can and should compel them to do so. while limiting stress on the electric system and Specifically, the state should require data centers reducing local environmental concerns. to transition their backup power systems toward progressively cleaner, lower-emission technologies, ◊ More efficient computing equipment: Using aligned with the state’s broader 2045 zero-carbon servers, chips, and power systems that deliver goals. more computing for each unit of electricity used. Performance would be assessed at regular intervals ◊ Smarter energy management: Using software to shift non-urgent computing to times when the using measurable emissions or technology grid is less stressed or when cleaner energy is benchmarks. Facilities that fail to meet interim available. standards would face consequences—such as higher electricity rates, restricted eligibility for ◊ Advanced cooling systems: Moving beyond new interconnections, or, in persistent cases, traditional cooling to methods that use less energy suspension of operations—while those that meet and/or water while keeping equipment safe and or exceed targets could receive incentives such as stable. rate discounts, expedited permitting, or eligibility for ◊ Better heat reuse: Capturing waste heat from state grants and pilot programs. California has an data centers and using it for nearby buildings or opportunity to set national standards by ensuring other community needs where feasible. that backup systems support grid reliability without ◊ On-site clean power generation: Generating sacrificing the health of nearby communities. By cleaner electricity at the facility, such as solar or pushing the data center sector towards progressively other low-emission power sources, to reduce cleaner technologies, the state can protect public pressure on the broader grid. health, maintain dependable electricity service, and demonstrate that grid reliability, public health, and ◊ Cleaner and quieter backup power: Replacing environmental stewardship can advance together. or supplementing diesel generators with cleaner backup options that reduce air pollution and RECOMMENDATION 12: California should community impact. require data centers to maintain a minimum level of ◊ On-site energy storage: Installing large batteries clean backup power to support load shifting, enable so facilities can rely less on the grid during peak periods and help smooth demand. 27 | LITTLE HOOVER COMMISSION ◊ Flexible demand: Designing operations so data PROMOTING DATA CENTER ENERGY centers can voluntarily reduce or adjust power use RESEARCH during emergencies or when the grid is strained. The State of California should establish a formal ◊ Better siting and design choices: Locating and private/public partnership with industry, the federal designing facilities in ways that align with cleaner national laboratory system (including Lawrence energy resources, minimize local impacts, and Livermore National Laboratory and Lawrence reduce the risk of stranded infrastructure. Berkeley National Laboratory), and relevant academic institutions including the University of California system, Stanford University, and other private California can encourage adoption of advanced universities. This initiative—provisionally named efficiency, clean energy, and grid-support the California Partnership for Advanced Research technologies by aligning incentives with measurable (CALPAR)—would research, test, validate, and public benefits. Incentives could be adopted to promote cutting-edge technologies to improve AI data reward facilities that are more efficient, flexible, center efficiency and environmental performance. and cleaner—for example, lower rates or credits Potential priority working group areas could include: for centers that reduce their contribution to peak demand, participate in demand-response programs, ◊ Grid Reliability, Flexibility, and Infrastructure or demonstrate strong energy and environmental Readiness: Focuses on ensuring rapid growth in AI performance. data centers strengthens rather than destabilizes California’s electric system. Examines demand Streamlined interconnection and permitting pathways flexibility, interconnection planning, transmission could also be offered for facilities that meet high expansion, and backup solutions that enhance standards, reducing uncertainty and timelines overall grid resilience. while protecting communities and environmental safeguards. A voluntary state designation for “high- ◊ Efficiency, Technology Innovation, and performance” data centers, paired with meaningful Environmental Performance: Aims to accelerate performance reporting, would create recognition, research, validation, and real-world deployment transparency, and potential market value for of technologies that reduce energy and water use. operators. Works with national laboratories, universities, and industry to test cooling innovations, improve server Financial tools and partnerships can complement efficiency, and explore waste-heat reuse and these steps. Incentives or grants could help offset related advancements. early costs for emerging technologies such as ◊ Transparency, Accountability, and Public advanced cooling, large-scale storage, or cleaner Trust: Develops privacy frameworks for sharing backup power. A state-led research partnership meaningful operational and environmental data with national laboratories, universities, utilities, so policymakers and the public can understand and industry—as recommended below—could true system impacts. Emphasizes standardized test new technologies, validate results, and share metrics, responsible disclosure, and clear public best practices. And state procurement and leasing communication to reduce confusion and build preferences for high-performance facilities can use confidence. public purchasing power to encourage environmental ◊ Community Impacts, Siting, and Environmental leadership. Justice: Seeks to ensure that data center siting and RECOMMENDATION 14: California should infrastructure decisions avoid disproportionate encourage data centers to adopt technologies—such harms while providing tangible local benefits. as advanced cooling, energy storage, or on-site clean Addresses pollution, land use, water concerns, power—that reduce their impact on the grid. clean backup power expectations, and community 28 | LITTLE HOOVER COMMISSION engagement practices that foster predictability and trust. ◊ Workforce, Economic Development, and Competitiveness: Aligns California’s technological leadership with strong workforce pipelines and long-term economic strategy. Evaluates training pathways, partnerships with higher education, job quality standards, and policies that help retain investment while advancing efficiency and environmental performance goals. There is precedent for such an organization. Governor Newsom recently formed a California Innovation Council, consisting of a blend of tech industry leaders, academics, and state workers. This Council is divided into working groups focused on modernizing government service delivery, economic and workforce development, promoting online safety, and countering fraud. RECOMMENDATION 15: California should establish a formal public-private partnership— provisionally the California Partnership for Advanced Research (CALPAR)—to collaborate with industry, national laboratories, and universities to research, test, and promote technologies that improve the efficiency and environmental performance of AI- driven data centers. 29 | LITTLE HOOVER COMMISSION Data Centers and the Water Supply Data centers are resource-intensive, placing closed-loop liquid-based cooling systems significant demands not only on electricity typically use more electricity but less water.117 As systems but also on the local water supply. a result, regions with ample clean or renewable This water is used to manage the significant electricity can potentially minimize water use heat generated by high-performance by data centers without compromising their computing equipment and, in the case of environmental standards.118 thermal generation, to produce the electricity Emerging technologies may also help mitigate that powers the facility.111 It should also be data center water consumption. These include noted that chip fabrication itself is a water intensive process.112 A “typical” data center advanced high-efficiency cooling systems, direct-to-chip liquid cooling and immersion can consume as much water as approximately cooling technologies, lower-heat-emitting chips, 1,000 homes, while a large hyperscale data center may use water comparable to that of a the substitution of recycled or graywater for town with roughly 50,000 residents.113 While the potable water, and smarter cooling and water Commission recognizes the importance of data management.119 centers’ water use as an issue for policymakers, Lack of data has hindered meaningful analysis this report focuses on the centers’ impacts of this issue, however. Data center operators on electricity systems and does not offer have been reluctant to disclose information recommendations regarding water policy. about their water use, including how efficiently In California, the primary concern is often water is used relative to electricity consumption, not the water demand of any single facility, and have opposed legislation that would require but the cumulative impact of multiple data such reporting.120 centers clustered within the same area. As a result, mandating water-use reporting Such concentrated development can strain by data centers has emerged as a prominent local water supplies.114 In addition, increased policy theme. In California, for example, AB 93 water withdrawals may stress aging water (Papan—2025) would have required data center infrastructure and wastewater treatment developers to provide estimates of water use facilities—which are themselves energy- prior to receiving a business license or permit intensive and may therefore contribute to and to report their annual water consumption higher overall electricity demand due to added thereafter.121 The bill was ultimately vetoed load from data centers.115 by Governor Newsom, who cited concerns Tangible impacts are already emerging. Data about its potential impact on the growth of the center water demand appears to be accelerating artificial intelligence sector.122 the depletion of the Potomac Aquifer, forcing Other states, however, have moved forward some communities to shift from groundwater to with similar—and in some cases more more constrained surface-water supplies.116 stringent—requirements. Minnesota’s HF 16 There is also an important tradeoff between requires detailed water-use reporting as part of water use and electricity use in data center a project application, while Kansas’ SB 98 links cooling. Cooling systems that rely more heavily data center tax exemptions to, among other on water—such as evaporative cooling—can things, the development of a sustainable water- reduce electricity consumption, while air- or use plan.123 30 | LITTLE HOOVER COMMISSION Data Centers and Air Pollution THE DANGERS OF DIESEL generators—operating for only 20 hours To meet stringent reliability standards—of- per year—would emit approximately the ten allowing only a few minutes of down- same amount of air pollutants as 428 time per year—data centers typically rely gasoline-powered cars driven for an entire on on-site diesel backup generators. These year.130 These regions already experience generators provide emergency power significant air quality challenges, and during grid outages and ensure service con- additional emissions may further amplify tinuity during peak demand.124 public health burdens. POTENTIAL SOLUTIONS Emissions from diesel backup generators Technological and policy options exist to raise significant public health and mitigate these impacts. Alternative backup environmental justice concerns, particularly power solutions—including hydrogen for communities near data center clusters. fuel cells, lithium-ion battery storage Diesel generators emit diesel particulate systems, renewable generation paired matter, a toxic air contaminant associated with energy storage, and other lower- with increased cancer risk as well as emission technologies—can substantially adverse cardiovascular and respiratory reduce or eliminate local air pollutant health outcomes, including asthma.125 emissions.131 Despite these alternatives, Researchers from the University of data centers continue to rely heavily California, Riverside estimate that pollutant on diesel generators because they are emissions from data centers nationwide generally perceived as more cost-effective could contribute to more than one-third of and reliable.132 asthma-related deaths by 2028.126 Other states have begun to consider CALIFORNIA IMPACTS legislation addressing air quality and These risks are especially concerning in climate impacts from data centers. In California, which is home to the third- Oregon, House Bill 2816 (Marsh, 2023) highest concentration of data centers in the proposed requiring high-energy-use United States.127 Public health impacts are facilities, including data centers, to keep worsened by the geographic concentration associated greenhouse gas emissions of data centers around San Jose and Los below specific target levels.133 In New Angeles. In 2025, the American Lung York, Senate Bill S9144 (Krueger, 2026) Association ranked Los Angeles and San would impose a temporary moratorium Jose the fifth and sixth most polluted cities on permits for new data centers while the in the nation, respectively, for year-round Department of Environmental Conservation particle pollution.128 conducts a statewide environmental impact One proposed hyperscale data center review and develops updated regulations in San Jose plans to operate 39 diesel to mitigate environmental and air quality backup generators.129 According to the impacts.134 Although the Oregon bill died in California Energy Commission, these committee and the New York bill remains 31 | LITTLE HOOVER COMMISSION under review, these proposals reflect or carbon-free energy generated onsite or growing concern among policymakers procured to power their operations.135 And about the air pollution impacts of data SB 887 (Padilla, 2026) would require data center development. centers to meet peak demand using on- site zero-carbon energy storage and rely CALIFORNIA POLICY as much as feasible on behind-the-meter California legislators have similarly zero-carbon generation.136 As data center introduced measures aimed at reducing air development accelerates, California has pollution from data centers by encouraging an opportunity to shape national policy by cleaner energy and backup power systems. linking data center reliability requirements Proposed legislation such as AB 1577 with stronger protections for local air (Bauer-Kahan, 2026) would require data quality and public health. centers to disclose the share of renewable 32 | LITTLE HOOVER COMMISSION Notes 1. U.S. Energy Information Administration, “Retail of the Utility Reform Network Addressing Proposed Electricity Prices Closely Tracked Inflation Over the Rule 30 Ratepayer Impacts and Related Issues [CPUC Last 10 Years” (Sep. 11, 2024). M. Baldassare, D. Proceeding A.24-11-007] (Jun. 30, 2025). Bonner, L. Mora, and D. Thomas, PPIC Statewide 8. See, for example, n.d., “How Big Tech Won Big Survey: Californians and Their Economic Well-Being, Against Regulation in California This Year”, GovTech Public Policy Institute of California (Dec. 2024). Industry Insider [Tribune News Service] (Oct. 22, 2. S. Susnjara and I. Smalley, “What Is Cloud 2204). K. Johnson, “California Lawmakers Wanted to Computing?”, IBM Think (n.d.); Geeks for Geeks Get Tough on Data Centers. Here’s What Survived”, “History of Cloud Computing” website. CalMatters (Sep. 25, 2025). 3. For a good review article on predictions of AI’s 9. B. Smith, “Building Community-First AI potential impact on the economy, see D. Björkegren, Infrastructure” [blog post], Microsoft (Jan. 13, 2025). “AI is Transforming the Economy—Understanding 10. A. Vahdat and J. Dean, “How Much Energy Does its Impact Requires Both Data and Imagination”, Google’s AI Use? We Did the Math” [blog post], Nature (Dec. 15, 2025). For a searchable collection Google Cloud (Aug. 21, 2025). of AI business use cases, see Deloitte’s “AI Dossier” website. 11. Data Center Coalition “Frequently Asked Questions about the Data Center Industry” website. 4. R. Li, “The Bay Area is Losing the Data Center Race. Economists Say That Could be a Good Thing”, San 12. I. Klaus et al., Carnegie California AI Survey, Francisco Chronicle (Dec. 18, 2025). Carnegie Endowment for International Peace (Oct. 9, 2025). Cf. E. Gorelick, “Why Do Americans Hate A.I.?”, 5. PG&E, “PG&E Data Center Demand Pipeline Swells New York Times (Jan. 2, 2026). to 10 Gigawatts with Potential to Unlock Billions in Benefits for California” [press release] (Jul. 31, 2025). 13. N. Portuondo, “Data center boom creates political According to the California Energy Commission’s conundrum for the GOP”, Politico (Dec. 14, 2025). most recent reporting, the state’s generation capacity is 89,283 megawatts or ~8.9 gigawatts. 14. In a recent PPIC survey, 37 percent of California residents named “cost of living, economy, [and] 6. For example, PG&E has claimed that for every inflation” as the most pressing issue in the state. gigawatt of new data center load, customers can See M. Baldassare, D. Bonner, L. Mora, and D. expect to save between one and two percent on Thomas, PPIC Statewide Survey: Californians and Their their electricity bills (see “PG&E Data Center Demand Government, Public Policy Institute of California (Jun. Pipeline”, op. cit.). For a balanced analysis, see S. 2025). Borenstein, “What Will Data Centers Do to Your Electric Bill?”, Energy Institute Blog [UC Berkeley] (Sep. 15. Cf. A. Barth et al., “The Data Center Balance: 29, 2025). How US States Can Navigate the Opportunities and Challenges”, McKinsey & Co. (Aug. 8, 2025). 7. The impact of data centers on electricity prices is hotly debated. For the view that data centers lower 16. A. Schaap, “How Data Centers are Transforming customer bills, see above. For the opposite, see Local Communities”, Fast Company (Mar. 20, 2025). J. Saul et al., “AI Data Centers Are Sending Power Bills Soaring”, Bloomberg (Sep. 29, 2025). For an in- depth argument that data centers will raise rates in California, see The Utility Reform Network, Testimony 33 | LITTLE HOOVER COMMISSION 17. B. Violino, “Computing and Storage are Moving to 27. For an interesting, in-depth discussion of “data the Edge, and IT Needs to be Ready”, CNBC (Jun. 11, center ethics”, though not specifically related to 2024). A. Steele, “The Rise of Vertical Data Centers”, energy, see the Climate Sustainability Directory “Data Colliers Knowledge Leader (Jun. 24, 2024). Center Ethics” website. 18. J. Roundy, “The Increasing Concern of Data Center 28. A. Satchwell, N. Mims Frick, P. Cappers, et al., Land Acquisition”, TechTarget (Aug. 21, 2025). Electricity Rate Designs for Large Loads: Evolving Practices and Opportunities, Lawrence Berkeley 19. For an overview with links to scientific studies, National Laboratory (Jan. 2025). see A. Kadasamy, “More Data Centers, More Environmental Problems?”, National Wildlife 29. The aforementioned Lawrence Berkeley study Federation (Sep. 24, 2025). contains numerous examples. See also M. Fitzpatrick, Data Center Energy Impacts and State Responses, Office 20. A. Garcia and K. Norton, “Next Data Center of Legislative Research [Connecticut] (Oct. 14, 2025). Roadblock for Google, Microsoft, Oracle, Meta is On Main Street”, Investor’s Business Daily (Dec. 20, 2025). 30. J. Chambers, “Paraguay – the Silicon Valley of South America?”, British Broadcasting Corporation 21. S. Nadel, “Data Center Efficiency and Load (Oct. 12, 2025). Flexibility Can Reduce Power Grid Strain and Costs” [blog post], American Council for an Energy-Efficient 31. M. North, “Here’s How Data Centre Heat Can Economy (ACEEE) (Oct. 7, 2025). Warm Your Home”, World Economic Forum (Jun. 18, 2025). To this list, we could add how China is 22. For example, T. Panadreou, “Who Wins the Energy developing a wind-powered data center off the Race: AI Data Centers, EVs, or Bitcoin Mining?”, Forbes coast of Shanghai that is submerged underwater, (Aug. 30, 2024). taking advantage of natural cooling and fewer land constraints. Companies are also exploring orbital 23. The issue of stranded infrastructure is addressed data centers, which could be powered by constant in K. Hieta and E. Rodriguez, “How Will Data Center solar energy and face no land limitations. See Y. Growth Impact California Ratepayers?”, Public Xiaoxing, “China Is Putting Data Centers in the Ocean Advocates Office (Oct. 28, 2025). to Keep Them Cool”, Scientific American (Jul. 16, 2025) 24. It is unclear how often data centers relocate or and D. Rao, “Data Centers Could Soon Be Orbiting in for what reasons. For an overview of the concept, Space”, The Week (Dec. 23, 2025). see C. Tozzi, “Streamlining Data Center Relocation: A 32. A. Peters, “These Silicon Valley Apartments Will Checklist for Success”, Data Center Knowledge (Sep. 24, Get Their Heat from Nearby Data Centers”, Fast 2023). A number of companies provide specialized Company (Nov. 15, 2024). data center relocation services. 33. See the Data Center Map “California Data 25. S. Ren and A. Wierman, “Mitigating the Public Centers” website for an overview. Cf. M. Petersen, Health Impacts of AI Data Centers”, Harvard Business “Power-hungry AI Data Centers are Raising Electric Review (Nov. 5, 2025). Bills and Blackout Risk”, Los Angeles Times (Aug. 12, 26. A. Lazo, “Data Centers are Putting New Strain 2024). on California’s Grid. A New Report Estimates the 34. Links to PG&E’s increasing 2025 predictions of Impacts”, CalMatters (Nov. 14, 2025). new data center demand can be found here (Feb.), here (May), and here (July). 34 | LITTLE HOOVER COMMISSION 35. The two “Data Center Project Pipeline” charts are 43. Efforts to require more detailed disclosure have from PG&E, 2024 Investor Update (Jun. 12, 2024), pp. so far stalled. For example, 2025 legislation (AB 222, 20-21. Bauer-Kahan), which would have required electricity- use reporting from data centers and assessment of 36. D. Bergstrom, “Could Fresno Soon be Home potential cost shifts, did not advance. to Thirsty AI Data Centers?”, Fresnoland (Aug. 26, 2025). Cf. F. Barringer, Thirsty for Water and Power, 44. See T. Tsoneva and J. Affleck, Data Center AI-Crunching Data Centers Sprout Across the West, Investment, op cit.; n.d., North America Data Center Bill Lane Center for the American West [Stanford Trends H1 2025, CBRE (Sep. 8, 2025); and J. Noffsinger University] (Apr. 8, 2025). et al., The Cost of Compute: A $7 trillion Race to Scale Data Centers, McKinsey & Co. (Apr. 28, 2025). 37.See the Data Center Map “California Data Centers” website as well as T. Tsoneva and J. Affleck, Data 45. N.d., What the Real Estate Industry Needs to Know Center Investment: Decoding Opportunities, CBRE About Data Centers, McKinsey & Co. (Oct. 15, 2024). (Jul. 17, 2024). The pie chart indicating data center 46. T. Hylton, E. Herald, and P. Werner, “Taxing the ownership is from this second report. Digital Backbone: What You Need to Know About 38. A. Shehabi et al., 2024 United States Data Center Data Centers” [blog post], Price Waterhouse Coopers Energy Usage Report, Lawrence Berkeley National (Jul. 25, 2025). Laboratory (Dec. 2024), p. 12. 47. See, for example, S. Rwejuna and A. Bryant, “DOE 39. Links to PG&E’s increasing 2025 predictions of Directs FERC to Accelerate Interconnection of Data new data center demand can be found here (Feb.), Centers”, White & Case (Oct. 27, 2025). here (May), and here (July). At the same time, PG&E 48. California has struggled to pass major data center has emphasized that substantial portions of its legislation. For an analysis of the politics involved, existing grid are underutilized and that additional see J. Wolman and L. Kashinsky, “Big Tech’s Next large-load customers could distribute fixed costs Major Political Battle May Already be Brewing in Your more broadly, potentially lowering residential bills by Backyard”, Politico (Aug. 10, 2025). one to two percent per gigawatt. See PG&E, “Surging Data Center Growth to Help Lower Energy Costs for 49. B. Giacobone, “Phantom Data Centers are PG&E Customers” [press release] (May 27, 2025). Flooding the Load Queue”, Latitude Media (Mar. 26, 2025). N.d., Uncertainty and Upward Bias are Inherent 40. California Energy Commission, 2025 IEPR in Data Center Electricity Demand Projections, London Forecast—Updated Results [DAWG Meeting] (Jan. 5, Economics Institute (Jul. 7, 2025). 2026). The mid case forecast appears on p. 8. The high case forecast appears on p. 9. 50. B. Martucci, “A Fraction of Proposed Data Centers Will Get Built. Utilities are Wising Up”, Utility Dive (May 41. CAISO “Large Load” website. CAISO predicts that data centers will add 2.3 gigawatts of load by 20230 51. Overviews of the contents of SB 6 include: n.d., and 3.3 gigawatts of load by 2035. “Texas Senate Bill 6 Significantly Expands Regulatory Oversight Over Large Loads in ERCOT”, McGuire 42. H. Javanbakht, “California Energy Commission Woods (Jul. 10, 2025). D. Chernicoff, “Texas Senate Bill Data Center Demand Forecast” [presentation at the 6: A Bellwether On How States May Approach Data Little Hoover Commission hearing on Data Centers Center Energy Use”, Data Center Frontier (Jul. 2, 2025). and California Electricity Policy, pt. II] (Dec. 11, 2025), Slides 4 and 6. 35 | LITTLE HOOVER COMMISSION 52. Z. Skidmore, “AEP Ohio Slashes Data Center 61. Links to PG&E’s current and historical rates, Pipeline by More Than Half—Report”, Data Center including B-20, can be found at its “Electric Rates” Dynamics (Oct. 1, 2025). website. 53. Identifying “policy levers” vis a vis state regulation 62. This was the goal of SB 57 (Padilla, 2025) in its of data centers was inspired by n.d., Advancing original form. Affordable Electricity in California: Policy Levers to 63. The following details are from California Public Address Rising Rates, Public Advocates Office (Dec. 13, Utilities Commission, Decision (D.) 25-07-039, Decision 2024). Partly Granting and Partly Denying Pacific Gas and 54. K. Hieta and E. Rodriguez, “How Will Data Electric Company’s Motion for Interim Implementation of Center Growth Impact California Ratepayers?”, Electric Rule Number 30 [A. 24-11-007] (Jul. 24, 2025). Public Advocates Office (Oct. 28, 2025). See also the 64. PG&E, Electric Rule No. 30 – Retail Service California ISO “Interconnection Request and Study” Transmission Facilities (Interim Implementation) [Advice website. Letter 7772-E; D.25-07-039] (Dec. 4, 2025). 55. See the California Air Resources Board “Air 65. Technical details about the load interconnection Quality and Emissions Resources” website. process can be found on PG&E’s “Transmission and 56. See City of Santa Clara “Environmental Review/ Distribution Interconnection Handbooks” website. CEQA” website, as well as Office of Senator Steve 66. Quoted text is from Tex. SB 6 § 2(b) (2025). Padilla, “Senator Padilla Calls for Public Input and Review After Data Center Project Granted Exemption 67. Overviews of the contents of SB 6 include: n.d., from Environmental Review and Previews Plans to “Texas Senate Bill 6 Significantly Expands Regulatory Introduce Legislation Setting Data Center Energy Oversight Over Large Loads in ERCOT”, McGuire Standards” [press release] (Dec. 3, 2025). Woods (Jul. 10, 2025). D. Chernicoff, “Texas Senate Bill 6: A Bellwether On How States May Approach Data 57. See, for example, J. Remington and R. Carter, Center Energy Use”, Data Center Frontier (Jul. 2, 2025). “An Overview of State Data Center-Related Tax Incentives”, NAIOP Commercial Real Estate 68. N.d., “ERCOT ‘Kill Switch’ for Data Centers Passes Development Association (Winter 2024/2025). in Texas House”, GovTech Industry Insider [Tribune News Service] (May 27, 2025). 58. Cf. the background document to the July 24, 2025 Little Hoover Commission hearing on California 69. For a succinct overview, see the Governor’s Electricity Costs, pp. 1-2 and n. 2, which includes “GenAI for California” website. relevant bibliography. 70. B. Casselman and S. Ember, “The A.I. Boom Is 59. Quoted text is from the “What are Customer Driving the Economy. What Happens if It Falters?”, Classes?” tab of the CPUC “Electric Rates” website. New York Times (Nov. 22, 2025). 60. See PG&E’s written testimony, submitted in 71. R. Singh and J. Jose, “Citigroup forecasts Big advance of the Little Hoover Commission’s Dec. 11, Tech’s AI Spending to Cross $2.8 Trillion by 2029”, 2025 hearing, which provides specific information Reuters (Sep. 30, 2025). C. Metz, “Why Debt Funding regarding costs paid through transmission rates (pp. Is Ratcheting Up the Risks of the A.I. Boom”, New York 5-6 and n. 10). Times (Nov. 10, 2025). 36 | LITTLE HOOVER COMMISSION 72. As cited in M. MacCarthy, “Policymakers Have to Initiative. Slides from his presentation to the Prepare Now for When the AI Bubble Bursts”, Tech Commission can be found here. Policy Press (Nov. 24, 2025). 82. C. Chen, “China Built Hundreds of AI Data Centers 73. It is unclear how often data centers relocate or to Catch the AI Boom. Now Many Stand Unused”, MIT for what reasons. For an overview of the concept, Technology Review (Mar. 26, 2025). see C. Tozzi, “Streamlining Data Center Relocation: A Checklist for Success”, Data Center Knowledge (Sep. 24, 83. Georgia Public Service Commission, “2025 Data 2023). A number of companies provide specialized Center Fact Sheet” (Dec. 1, 2025); J. Holdman, “SC’s data center relocation services. State-Owned Utility Enacts Higher Rates for Data Centers, Large Users”, South Carolina Daily Gazette 74. See, for example, G. Radauskas, “California (Apr. 25, 2025); Z. Skidmore, “Oregon House Passes Dreamin’ No More: State Facing Tech Jobs Exodus”, Bill Shifting Power Infrastructure Costs to Data CyberNews (Apr. 12, 2024). Centers”, Data Center Dynamics (Jun. 9, 2025); Z. Skidmore, “Virginia Regulators Approve New Rate 75. The International Energy Agency has documented Class for Data Centers and Other Large Loads”, Data continuing, if more modest, efficiency gains. See Center Dynamics (Nov. 27, 2025). M. Offutt and L. Zhu, “Data Centers and Their Energy Consumption: Frequently Asked Questions”, 84. A. Levenson-Falk, “CUB Spearheads Nation- Congressional Ressearch Service (Aug. 26, 2025). Leading Legislation to Protect Ratepayers from the Costs to Serve Data Centers”, Citizens Utility Board 76. J. Quigley, “Load Growth Irrational Exuberance (Jul. 9, 2025). Crashes into DeepSeek”, Kleinman Center for Energy Policy Blog [University of Pennsylvania] (Jan.31, 2025). 85. K. Suzuki, “Imperial Valley City Sues to Force For an overview of the benefits of Google’s new AI Environmental Review of Massive Data Center chips, see D. Patel et al., “TPUv7: Google Takes a Project”, KPBS (Dec. 12, 2025). Swing at the King”, SemiAnalysis (Nov. 28, 2025). 86. For a clear discussion, see S. Borenstein, “Can 77. J. Quigley, “Load Growth Irrational Exuberance Data Centers Flex Their Power Demand?”, Energy Crashes into DeepSeek”, Kleinman Center for Energy Institute Blog [UC Berkeley] (Apr. 14, 2025). Policy Blog [University of Pennsylvania] (Jan.31, 2025). The in-quote hyperlink is original to this source. 87. S. Fleming, “The Grid is Built for the Peak” [blog post], Clean Energy Review (Oct. 3, 2024). 78. NovaSky Team, “Sky-T1: Train Your Own o1 Preview Model Within $450”, Github (Jan. 9, 2025). 88. D. Ai, C.L. Crago, J.T. Mullins, “Heat, Power Outages and Mortality in the United States”, Journal 79. For a general introduction, see n.d., “The Rise of of Environmental Economics and Management (Dec. 20, Mini Models in Artificial Intelligence: A New Era of 2025). Efficiency”, VisioTech (Sep. 24, 2024). For a more in- depth discussion, see R.D. Caballar, “What are Small 89. California Energy Commission, “California Energy Language Models?”, IBM Think (n.d.). Leaders Report Progress on Grid Reliability Ahead of Summer 2025” [press release] (May 2, 2025). 80. J. Bushnell, “Is There Really an Energy Silver Lining California’s highest ever peak demand was 52,061 to a Datacenter Bubble?”, Energy Institute Blog [UC MW on Sep. 6, 2022. Berkeley] (Oct. 13, 2025). 90. For analysis of grid performance, see Public 81. See, in particular, the testimony of Dr. Liang Advocates Office, Preliminary Analysis of California’s Min, Managing Director of Stanford’s Bits & Watts Resiliency During the September 2022 Heat Wave (n.d.). 37 | LITTLE HOOVER COMMISSION 91. B. Martucci, “Texas Law Gives Grid Operator 101. C. Knittel, J.R. Senga, and S. Wang, “Flexible Power to Disconnect Data Centers During Crisis”, Data Centers and the Grid: Lower Costs, Higher Utility Dive (Jun. 25, 2025). Emissions?”, MIT CEEPR [WP 2025-14] (Jul. 2025). 92. For example, in written testimony provided in 102. See U.S. Government Accountability Office, advance of the Little Hoover Commission’s Dec. 11, “Electricity: Information on Peak Demand Power 2025 hearing (p. 6, n. 11), PG&E pointed to its Flex Plants” [GAO-24-106145] (May 21, 2024). L. Kearney, Connect program, which can rapidly interconnect “Insight: AI Data Centers are Forcing Dirty ‘Peaker’ new EV charging ports if they commit to lowering Power Plants Back into Service”, Reuters (Dec. 23, load during times of peak use. Cf. PG&E, “New 2025). PG&E Service Offering Makes It Easier and Faster to 103. PG&E, “New PG&E Service Offering Makes It Connect EV Chargers, EV Fleets and Big Batteries to Easier and Faster to Connect EV Chargers, EV Fleets the Grid” [press release] (Apr. 28, 2025). and Big Batteries to the Grid” [press release] (Apr. 28, 93. R. Prabha and L. Min, U.S. Transmission System 2025). Utilization Study Phase 1: WECC, Bits & Watts Initiative 104. For an overview, see n.d., “How ‘Load Shifting’ [Stanford University] (Dec. 12, 2025). May Help Improve Data Center Sustainability”, Data 94. PG&E, “PG&E and Smart Wires Enhance Grid Center Knowledge [Bloomberg News] (Feb. 26, 2024). Reliability, Capacity for Data Centers in San Jose” For a scientific analysis, see I. Riepin, T. Brown, and [press release] (May 28, 2025). V.M. Zavala, “Spatio-Temporal Load Shifting for Truly Clean Computing”, Advances in Applied Energy 17 (Mar. 95. H. Parmar et al., How to Find the ‘Sweet Spots’ to 2025). Build All Those Data Centers, ICF (n.d.). 105. SB 100 (De León, 2018) requires that 60 percent 96. K. Hieta and E. Rodriguez, “How Will Data Center of California’s retail electricity sales be supplied Growth Impact California Ratepayers?”, Public from renewable sources by 2030 and 100 percent Advocates Office (Oct. 28, 2025). from renewable or other zero-carbon resources by 2045. AB 1279 (Muratsuchi, 2022) requires that 97. B. Roberts, Data Center Infrastructure in the United statewide greenhouse gas emissions to be reduced States, 2025, NREL (2025). K. Cooke, “The Hidden at least 85 percent below 1990 levels by 2045 and Complexity of Data Center Siting” [blog post], Esri that remaining emissions be offset to achieve overall (Sep. 8, 2025). carbon neutrality. 98. PG&E, “PG&E Accelerating Connection of New 106. For an accessible technical overview of the “five Data Centers throughout Northern and Central nines” concept, see R. Reynolds, “Achieving “Five California” [press release] (Feb. 13, 2025). Cf. PG&E’s Nines” in the Cloud for Justice and Public Safety” [blog written testimony to the Little Hoover Commission (p. post], Amazon Web Services Public Sector Blog (Mar. 17, 6) for details. 2020). 99. For an overview, see W. Kirchstetter, Data Center 107. For an overview, see C. Tozzi, “Comparing Load Flexibility Workshop Summary, Lawrence Berkeley Data Center Backup Power Systems”, Data Center National Laboratory (Jan. 2025). Knowledge (Nov. 25, 2205). 100. T. Norris et al., Rethinking Load Growth Assessing 108. United States Environmental Protection Agency, the Potential for Integration of Large Flexible Loads “EPA Issues Clarification to Help Power Data Centers, in US Power Systems, Nicholas Institute for Energy, Ensure U.S. Is the AI Capital of the World” (May 1, Environment & Sustainability [Duke University] 2025). (2025), p. 2. 38 | LITTLE HOOVER COMMISSION 109. Gordon and Allgood’s written testimony to 116. M. Willson, op. cit. For a discussion of drought- the Commission in advance of its November 20 and wastewater-related concerns, see County of hearing can be found here and here, respectively. Fairfax (VA), 2024 Annual Report on the Environment Cf. S. Ren and A. Wierman, “Mitigating the Public [Prepared by the Environmental Quality Advisory Health Impacts of AI Data Centers”, Harvard Business Council] (Dec. 2024). Review (Nov. 5, 2025). For the politics behind these 117. A. Setmajer, “How Data Centers Use Water, and generators, see C. Paullin, “State Regulators Weigh How We’re Working to Use Water Responsibly” [blog Expanded Use of Data Centers’ Diesel Backup post], Equinix (Sep. 19, 2024). Cf. Lawrence Berkeley Generators”, Inside Climate News [National Public National Laboratory, Center of Expertise for Data Radio] (Dec. 17, 2025). Center Energy “Water Efficiency” website. 110. R. Quint, J. Zhao, and K. Thomas, An Assessment 118. See M. Ashtine and D. Mytton, op. cit. of Large Load Interconnection Risks in the Western Interconnection, Western Electricity Coordinating 119. See, for example, P.J. Campbell and R. Schwaab, Council (Feb. 2025), p. 18. The U.S. Department of “How to Use Water Wisely in Data Centers”, Ramboll Energy has directed its staff and the national labs to (Dec. 2, 2025). engage in research and data gathering around this issue, and we believe California could assist in this 120. M. Willson, “States Push to End Secrecy Over process, for example, by providing grants targeted Data Center Water Use”, E&E News by Politico (Dec. 8, towards the study and development of backup power 2025). sources that are clean and capable of supporting demand flexibility. See Secretary of Energy Advisory 121. The text of AB 93 (Papan—2025) can be found Board, Recommendations on Powering Artificial here. Intelligence and Data Center Infrastructure, U.S. 122. Gov. Gavin Newsom, AB 93 Veto Message (Oct. Department of Energy (Jul. 30, 2024), pp. 8-10. 11, 2025). 111. M. Ashtine and D. Mytton, “We are Ignoring 123. The text of Minnesota’s HF 16 (Davids; Huot-- the True Cost of Water-Guzzling Data Centres”, The 2025) can be found here. The text of Kansas’ SB 98 Conversation (Oct. 19, 2021). I. James, “Amid the AI (Sen. Comm. on Commerce—2025) can be found boom, California Looks to Address Water Use by here. Massive Data Centers”, Los Angeles Times (Sep. 23, 2025). 124. Joint Legislative Audit and Review Commission, Data Centers in Virginia [RD206] (Dec. 9, 2024). 112. M. Yañez-Barnuevo, “Data Centers and Water Consumption”, Environmental and Energy Study 125. UC Newsroom, “California Data Center Health Institute (June 25, 2025). Impacts Tripled in 4 Years” [Press Release] (Nov. 21, 2025). 113. J.W. Kane, “AI, Data Centers, and Water”, Brookings (Nov. 20, 2025). 126. Y. Han, Z. Wu, P. Li, A. Wierman & S. Ren, “The Unpaid Toll: Quantifying and Addressing the Public 114. K. Silverstein, “America’s AI Boom Is Running Into Health Impact of Data Centers” [Working Paper, An Unplanned Water Problem”, Forbes (Jan. 11, 2026). arXiv:2412.06288] (Dec. 9, 2024). 115. R. Ahmad, “Engineers Often Need a Lot of Water 127. A. Foote & C. Wilkie-Rogers, “Employment in to Keep Data Centers Cool”, American Society of Civil Data Centers Increased by More Than 60% from 2016 Engineers (Mar. 4, 2024). to 2023 but Growth Was Uneven Across the United States”, U.S. Census Bureau (Jan. 6, 2025). 39 | LITTLE HOOVER COMMISSION 128. American Lung Association, State of the Air 2025 (Apr. 23, 2025). 129. For details, see the CEC “Great Oaks South Generating Facility” website. 130. A. Keating, “California Needs More Data Centers, but Builders Are Using a Legal Loophole to Power Them with Diesel”, Fast Company (May 2, 2023). According to CARB’s Airborne Toxic Control Measure for Stationary Compression Ignition Engines [Sec. 93115.6] (Dec. 11, 2007), diesel backup generators are required to run a minimum number of hours per year for maintenance and testing purposes. 131. CARB “Emergency Backup Power Options— Commercial” website. 132. U.S. Environmental Protection Agency, The Role of Distributed Generation and Combined Heat and Power (CHP) Systems in Data Centers (2015). 133. Click here for a link to the text of HB 2816. 134. Click here for a link to the text of S 9144 135. Click here for a link to the text of AB 1577 136. Click here for a link to the text of SB 887. 40 | LITTLE HOOVER COMMISSION Little Hoover Commission Members PEDRO NAVA | Santa Barbara GIL GARCETTI | Los Angeles Appointed to the Commission by Speaker of the Assembly Appointed to the Commission by Governor Gavin John Pérez in April 2013 and reappointed by Speakers Newsom in November 2021. Professional photographer Rendon in 2017 and 2021 and Rivas in 2024. Government and author of ten books. Former Los Angeles County relations advisor. Former State Assemblymember from District Attorney, teaching Fellow at Harvard University’s 2004 to 2010, civil litigator, deputy district attorney and Kennedy School, and president of the California Science member of the state Coastal Commission. Elected chair of Center Foundation’s Board of Trustees. the Commission in March 2014. JOSÉ ATILIO HERNÁNDEZ | Burbank ANTHONY CANNELLA | Ceres Appointed by Speaker of the Assembly Anthony Rendon Appointed to the Commission by the Senate Rules in April 2023. Founder and CEO of IDEATE California. Also, Committee in March 2022. Civil engineer and principal Founder and Board Chairman of ideateLABS non profit. with Northstar Engineering Group. Former State Senator from 2010 to 2018. Previously served on the Ceres City JASON JOHNSON | Napa Council and was twice elected mayor of that city. Elected Appointed by Governor Newsom in June 2023. Member Vice Chair of the Commission in July 2023. of the Land Trust of Napa County Board of Trustees and Honorary Commander of Travis Air Force Base. Former DION ARONER | Berkeley Managing Partner at Founders Den. Founder and former Appointed to the Commission by the Senate Rules CEO at August Home Inc. Committee in April 2019. Partner for Aroner, Jewel, and Ellis. Former State Assemblymember from 1996 to GAYLE MILLER | Sacramento 2002, chief of staff for Assemblymember Tom Bates, Managing Director of Transition, Institutional social worker for Alameda County, and the first female Relationships and Investments for Brookfield Asset president of Service Employees International Union 535. Management and Vice-Chair of the Delta Stewardship Council. Previously served as Senior Counselor on DAVID BEIER | San Francisco Infrastructure and Clean Energy Finance for Governor Appointed to the Commission by Governor Edmund G. Newsom and Chief Deputy of Policy at the California Brown Jr. in June 2014 and reappointed in January 2018. Department of Finance. Appointed to the Commission by Managing director of Bay City Capital. Former senior Governor Gavin Newsom in January 2025. officer of Genentech and Amgen, and counsel to the U.S. House of Representatives Committee on the Judiciary. SENATOR ROGER NIELLO | Fair Oaks Elected in 2022 to represent the 6th Senate District. SENATOR CHRISTOPHER CABALDON | West Former President and CEO of the Sacramento Metro Sacramento Change of Commerce and University of California Center Elected in 2024 to represent the 3rd Senate District. Sacramento Governance Fellow. Previously served on Former Mayor of West Sacramento for two decades. the Sacramento County Board of Supervisors and the Served as Vice Chancellor of the California Community California State Assembly. Appointed by the Senate Rules Colleges and director of the State Assembly Higher Committee in February 2025. Education Committee. Previously appointed by President Obama to the National Advisory Board of America’s ASSEMBLYMEMBER LIZ ORTEGA | San Leandro College Promise. Appointed to the Commission by the Elected in November 2022 to represent the 20th Senate Rules Committee in February 2025. Assembly District. Executive Secretary-Treasurer of the Alameda Labor Council. Former Statewide Political ASSEMBLYMEMBER PHILLIP CHEN | Yorba Linda Director for AFSCME Local 3299. Appointed by Speaker of Appointed to the Commission by Speaker of the Assembly the Assembly Rendon in March 2023. Anthony Rendon in October 2021. Elected in November 2016 to represent 55th District. Represents portions of JANNA SIDLEY | Los Angeles Los Angeles, Orange and San Bernardino counties and Appointed to the Commission by Governor Edmund the cities of Brea, Chino Hills, Diamond Bar, La Habra, G. Brown Jr. in April 2016 and reappointed in February Industry, Placentia, Rowland Heights, Walnut, West Covina 2020. Partner at Ichor Strategies and appointed to the and Yorba Linda. Board of the Los Angeles City Employee Retirement System (“LACERS”). Former general counsel at the Port of Los Angeles and city attorney at the Los Angeles City Attorney’s Office. 41 | LITTLE HOOVER COMMISSION “DEMOCRACY ITSELF IS A PROCESS OF CHANGE, AND SATISFACTION AND COMPLACENCY ARE ENEMIES OF GOOD GOVERNMENT.” By Governor Edmund G. “Pat” Brown, addressing the inaugural meeting of the Little Hoover Commission, April 24,1962, Sacramento, California Milton Marks Commission on California State Government Organization and Economy www.lhc.ca.gov