The next bottleneck for US AI computing power: not chips and electricity, but "licenses".
The real obstacles in the computing power race are shifting from silicon wafers and power grids to politics and regulation.
According to a recent research report by Barclays, as data centers face increasing community resistance, licensing delays, and regulatory fragmentation across the United States, "licensing" is becoming the next major bottleneck for the expansion of AI infrastructure, potentially as significant as computing power shortages or insufficient power supply.
A Gallup poll in March showed that 71% of American respondents opposed building data centers near their communities, a figure even higher than the 53% who opposed building nuclear power plants. This opposition transcends party lines, with Republicans, independents, and Democrats all holding majority views, and according to a more recent Politico survey, this sentiment continues to worsen. Barclays believes that today's political and licensing decisions will directly impact the supply of AI computing power in the latter half of this decade; if licensing challenges and community resistance continue to slow the deployment of hyperscale data centers, scarce computing resources will become even more valuable.
While the Trump administration positioned AI infrastructure as a strategic asset for economic and national security and actively promoted it through a number of policy tools, such as federal land opening, transmission expansion, electricity market reform, and nuclear energy deployment, many of the most critical decision-making powers remained decentralized at the local level, and federal influence was constrained by the highly fragmented US utility, licensing, and regulatory system.
Opposition is widespread and cross-party.
The politicization of data centers is becoming a major risk to the AI development boom. A Barclays report, citing Gallup survey data, points out that opponents' concerns are not focused on a single issue, but rather cover electricity consumption (18%), water use (18%), impact on quality of life (22%), rising electricity prices (20%), environmental pollution (16%), job losses (14%), and concerns about AI technology itself (14%).
This widespread opposition reflects a deeper societal perception: local communities bear the costs while the benefits flow elsewhere. In some projects, confidentiality agreements and limited public disclosure have further exacerbated the community's distrust of transparency. Meanwhile, general public concerns about AI technology, privacy, and the social influence of large tech companies are also reinforcing this resistance.
Political sensitivity has triggered substantial policy responses. The National Republican Senatorial Committee recently warned AI companies that public opposition to data centers could jeopardize a key Senate seat in Ohio. Texas Governor Greg Abbott has ordered state utility regulators and grid operators to review data center projects before proceeding with interconnection processes; Pennsylvania Governor Josh Shapiro issued an executive order requiring developers to meet state infrastructure standards and obtain local approval before permit applications are reviewed; New York Governor Kathy Hochul issued an executive order in July 2026 establishing the nation's first statewide halt to new hyperscale data center projects; and Virginia Governor Abigail Spanberger—of the world's largest data center market—requires data centers to pay for the dedicated power transmission infrastructure serving their facilities, rather than passing the costs on to other electricity users.
Electricity price increases have become the most politically damaging issue.
Of all the controversies, electricity affordability has become the most politically sensitive core issue. In the first quarter of 2026, the average residential electricity price in the United States rose by 12% compared to the same period in 2024, with more significant increases in areas experiencing rapid data center expansion. Washington state saw a 24% increase, Virginia and Pennsylvania both saw 17% increases, and Ohio, Louisiana, and Illinois all saw 14% increases.
Barclays points out that these increases cannot be entirely attributed to the expansion of hyperscale data centers; a significant portion reflects years of underinvestment in power grid infrastructure and spending on grid reinforcement to cope with extreme weather. However, the rapid growth in demand for data centers does put greater pressure on utilities and regulators to balance reliability, affordability, and investment incentives.
The report also expresses skepticism about the claim that "data centers can reduce electricity prices by amortizing fixed costs through increased electricity consumption." This logic relies on the premise of sufficient redundancy in existing power generation and transmission capacity, which is not the case in most AI data center regions. Data center cooling demands typically peak in the summer, coinciding with the period when the power grid is under the heaviest load due to air conditioning. Therefore, meeting incremental demand usually requires building new dispatchable power generation facilities and upgrading transmission lines, rather than utilizing existing idle capacity.
The regulatory landscape is fragmented, and the power struggle between local and state governments is intensifying.
Barclays categorized data center governance models across U.S. states into six types, revealing an increasingly complex regulatory landscape.
Under the state-led pre-occupation model, West Virginia passed the Electricity Production and Consumption Act in 2025, which restricts local government's regulatory power over eligible high-impact data centers, shifts the main approval authority to the state level, and provides supporting tax incentives and microgrid certification programs.
Under the direct state-level regulatory model, New York's statewide moratorium is the most typical example. However, a similar moratorium passed by the Maine legislature was vetoed by Governor Janet Mills, and similar proposals in Minnesota, New Hampshire, Oklahoma, and South Dakota have all failed to proceed. Under the hybrid state-local regulatory model, local governments in Texas, Ohio, and Pennsylvania retain primary zoning authority, but state agencies increasingly influence project outcomes through grid planning, interconnection requirements, and consumer protection policies.
This fragmented landscape means that licensing risks are becoming highly localized, and regulatory geography has become a significant variable in project site selection, development timelines, and the pace of AI infrastructure deployment.
The federal government has limited influence; key decision-making power remains at the local level.
The Trump administration has consistently positioned AI infrastructure as a strategic economic and national security priority. The White House's AI Action Plan, released in July 2025, lists AI leadership as a dual priority for both the economy and national security, and dedicates a pillar to accelerating the deployment of AI infrastructure. Trump himself stated on Truth Social that he "absolutely does not want Americans to pay higher electricity bills because of data centers," and pushed hyperscale carriers, utilities, and state officials to sign the "Power User Protection Pledge," requiring data center developers to build, import, or purchase the necessary electricity and bear all related infrastructure costs. However, this pledge is voluntary, non-binding, and fails to address core issues driving local opposition, such as water resources, land use, emissions, noise, and community impact.
Regarding federal land use, the U.S. Department of Energy has identified several potential sites for AI infrastructure, including the PORTS-Pike Technology Park in Ohio with a maximum computing capacity of 8 gigawatts (in collaboration with SoftBank, OpenAI, Nvidia, and AEP Ohio), the former Padyuca Gas Diffusion Plant site in Kentucky with more than 1.2 gigawatts (led by Brookfield and NextEra Energy), and the 1-gigawatt project in Savannah River, South Carolina (undertaken by Amentum).
However, federal influence is fundamentally constrained by the highly decentralized nature of the U.S. utility and regulatory system. The most critical decisions, such as transmission investment, interconnection rules, cost sharing, reliability standards, and consumer protection, remain dominated by regional electricity markets and state-level regulatory procedures. While the Federal Electricity Regulatory Commission (FERC) has advanced initiatives such as Order 1920 (Long-Term Regional Transmission Planning) and Order 2023 (Interconnection Reform), the construction of a truly nationwide transmission network still requires addressing the complex coordination of interests across multiple states, utility territories, and regulatory jurisdictions.
Will the community public relations offensive of mega-operators be effective?
Faced with rising community resistance, mega-operators such as Microsoft, Amazon, Google, Meta, and OpenAI are increasing their community relations efforts, attempting to exchange economic benefits for social approval. Microsoft launched its "Community-First AI Infrastructure" initiative, promising not to increase electricity bills for existing users, minimize water consumption and replenish excess water usage, create local jobs, increase local tax revenue, and invest in AI education and digital skills training. Meta's massive investment in power infrastructure for its Hyperion campus in Louisiana is expected to save Entergy Louisiana users approximately $2.7 billion in electricity costs; the initiators of the PORTS-Pike campus pledged approximately $4.2 billion for grid and transmission upgrades.
However, Barclays argues that many of the issues driving local opposition are structural rather than reputational. Even with credible grid upgrades and emissions reduction commitments from hyperscale operators, electricity prices in many parts of the U.S. could continue to rise for reasons unrelated to their actions. Water disputes involve multidimensional trade-offs between direct water use, indirect upstream water use, and emissions from the power sector, making them far more complex than they appear. In the current AI arms race, the priority of "rapid access to electricity" has outweighed cost and emissions control, making it difficult for the industry to demonstrate substantial progress toward its long-term sustainability goals.
Self-supplied electricity and emerging solutions: mitigation, not cure.
Faced with licensing and grid bottlenecks, Bring-Your-Own-Power (BYOP) and Power-as-a-Service (PaaS) models are gaining increasing attention. By building their own dedicated power generation facilities, developers can reduce their reliance on transmission upgrades and lengthy interconnection queues, while also mitigating the risk of passing on grid upgrade costs to other electricity users. In the AI computing power race, the value of deployment speed has surpassed the cost of electricity itself—a single gigawatt of AI computing power can support hundreds of billions of dollars in revenue annually, making delayed deployment extremely costly.
However, Barclays emphasizes that self-sufficient power is not a panacea. Hyperscale operators typically still prefer grid connection when conditions permit, as the grid's economies of scale and diversified supply mix are more conducive to achieving the "five nines" (99.999%) reliability required for mission-critical loads. Most self-sufficient power solutions rely heavily on natural gas generation and associated pipeline infrastructure, making it economically difficult to achieve the same level of reliability. Furthermore, self-sufficient power projects also face risks related to air permits, pipeline capacity, fuel supply agreements, and community acceptance—the controversy surrounding the xAI Memphis facility is a prime example.
From a longer-term perspective, improved computing efficiency, advanced energy technologies (including small modular nuclear reactors), distributed computing architectures, and even space-based data centers are seen as potential pathways to alleviate constraints on terrestrial infrastructure. However, Barclays points out that most advanced nuclear energy technologies are still years away from commercial deployment and are unlikely to alleviate the industry's immediate power and interconnectivity pressures. Data centers can be migrated, but bottlenecks often remain—perhaps the most accurate description of the current predicament in expanding AI infrastructure.
~~~~~~~~~~~~~~~~~~~~~~~
The above content is from Zhuifeng Trading Platform .
For more detailed analysis, including real-time updates and firsthand research, please join the [ Trading Channel Annual Membership ].
~~~~~~~~~~~~~~~~~~~~~~~
The above content is from Zhuifeng Trading Platform .
For more detailed analysis, including real-time updates and firsthand research, please join the [ Trading Channel Annual Membership ].
Risk warning and disclaimerInvesting involves risk; please exercise caution. This article does not constitute personal investment advice and does not take into account the specific investment objectives, financial situation, or needs of individual users. Users should consider whether any opinions, views, or conclusions in this article are suitable for their specific circumstances. Any investment decisions made based on this information are at your own risk.