The data center, the core pillar of the AI industry chain, is set to see its infrastructure spending hit new heights in 2026, with projections suggesting it will continue to break records at least through 2030. Amid this unprecedented construction surge, opposition to data centers across the United States has escalated dramatically this year, transforming from a niche local issue into a critical talking point ahead of the November midterm elections. Almost overnight, candidates from across the political spectrum have adjusted their campaign messaging to respond to a near-universal grassroots backlash against the AI computing infrastructure boom, which is backed by hundreds of billions of dollars in tech industry spending. Both major parties are scrambling to adapt to the shifting public mood and define their overall strategies toward the data center construction wave. However, President Donald Trump remains steadfast in his support for rapid development. "Other countries couldn't be happier about this anti-data center movement," he wrote in a social media post on August 31. His stance has put many Republicans in a difficult position, caught between supporting increasingly angry constituents and avoiding the ire of the sitting president, who is the absolute core of the party. Rising electricity costs are adding to the financial burden on American consumers and fueling discontent among lower- and middle-income voters. The race in Pennsylvania's 7th Congressional District, a crucial swing seat for both parties, may well illustrate the significance of this AI infrastructure issue in 2026. The constraints on data center construction have already become concrete. On August 3, Texas Governor Abbott called for a comprehensive review of all data centers proceeding through the state's grid operator's interconnection process, with projects paused until the review is complete. The state government disclosed that interconnection requests involved over 474 gigawatts of new electricity demand, with about 90% coming from data centers; these are application figures, not operational loads or firm orders. The scope of the review, covering electricity usage, water consumption, fiscal incentives, and community impact, shows how the bipartisan cost-of-living debate, closely tied to the midterms, can filter through to AI infrastructure investment via interconnection conditions and project timelines.
A research report led by a team headed by Michael Hartnett, the senior Bank of America strategist dubbed "Wall Street's most accurate strategist," integrates US policy conditions and capital costs into the valuation analysis of the AI computing theme. The report suggests that if Republicans fail to win the Senate and then lose both the Senate and the House, extended project approvals would delay the deployment of computing power and the realization of actual AI revenue and profits. A Democratic-controlled Congress, leading to higher financing rates, would increase data center construction costs and lower the present value of future cash flows. Under this valuation framework, global semiconductor firms, data center power chain equipment makers, and the "new cloud" AI computing rental companies would inevitably be affected by negative expectations, ultimately facing validation through actual orders, deployment pace, capital expenditure, and free cash flow. This selling pressure might first manifest as a contraction in valuations for AI-related stocks; if approval and financing constraints further cause cloud providers to postpone construction and equipment purchases, the impact would then transmit along the supply chain to supplier orders, revenue, and operator cash flow, turning policy concerns into earnings pressure. Hartnett identifies a Democratic sweep of Congress as a key stress scenario: increased market concerns over higher taxes, stricter regulation, and limited AI construction would pressure both earnings expectations and valuations. He accordingly proposes an asset price combination involving a potential US stock market decline of over 10%, a weaker dollar, and falling bond yields. Conversely, he views a Republican sweep as a risk-on scenario and a Republican Senate with a Democratic House as a "gridlock-style Goldilocks" that is mildly favorable for risk assets.
A trillion-dollar AI computing expansion faces a new test: voters, power supply, and financing costs. Wall Street giant Morgan Stanley, focusing on the latest key trends and judgments from the OpenAI Astra model, which Nvidia CEO Jensen Huang has hailed as the beginning of the "AGI era," notes that the significant improvement in large model capabilities makes more workloads economically viable, thereby tightening supply constraints on power, substrates, and memory manufacturing. In its scenario projections, the power capacity for compute deployment by hyperscale cloud providers is expected to expand from approximately 35 gigawatts in 2025 to about 145 gigawatts in 2028, roughly 4.1 times its original level. The release of Astra has heightened market expectations for Artificial General Intelligence (AGI), leading to a stronger trajectory of compute demand. In particular, the new growth model of "charging for outcomes" is expected to drive even stronger total compute demand, while more immediate evidence of demand comes from the AI R&D process itself—the "Recursive Self-Improvement (RSI)" trajectory where AI begins to "create AI". With advanced frontier models led by Astra driving ever-stronger AI compute demand, Morgan Stanley projects that the combined data center capital expenditure of the four major North American hyperscale cloud and AI application companies will rise from $917 billion in 2026 to $1.47 trillion in 2027 and $1.64 trillion in 2028, with deployed capacity expected to expand from 35 gigawatts in 2025 to 145 gigawatts in 2028. The sustainability of AI investment hinges on whether capital returns and operating cash flow can support expansion. Morgan Stanley's scenario calculations show that model companies offering API services using their own infrastructure achieve a Return on Invested Capital (ROIC) of approximately 46%, higher than the 31% for cloud providers renting out GPUs and the 25% for those relying on third-party infrastructure for API services. This supports the idea that companies mastering models, compute, and commercialization channels simultaneously can achieve higher returns, though these are estimates from quantitative models. Meanwhile, Morgan Stanley expects the combined operating cash flow of the four major companies to increase from $739 billion in 2026 to $1.23 trillion in 2028, with new debt financing needs falling from $238 billion to $90 billion.
Data center opposition is expanding the constraints on AI investment from just chip and power supply to include building permits and community acceptance. An Annenberg survey published in August, based on polling conducted in June and July, shows the proportion of adults opposing new data centers in their local area has risen from 49% to 61%, although there has been no significant change in public perception of AI's overall impact. Concurrently, Data Center Watch has confirmed that at least 75 projects with a total value of around $130 billion were blocked or delayed in the first quarter alone. For the AI industry chain, the direct impact is that the timing for bringing new computing power online becomes more uncertain, potentially delaying equipment delivery, cloud service expansion, and the AI-related revenue recognition that the market is focused on, which could lead to a pullback in the AI computing sector due to compressed valuations. The economic substance of the dispute revolves around who bears the cost of new power generation, transmission, and water supply. Even after tech companies signed the "Electricity Bill Payer Protection Pledge," disagreements persist over cost-sharing. Reports in early September indicate that Microsoft is appealing new regulations in Virginia requiring data center developers to pre-pay for transmission infrastructure costs. This case illustrates that implementing "corporate responsibility for their own power costs" involves complexities around payment timing, infrastructure scope, and risk allocation. If more expenses require developers to pay upfront, initial capital investment and financing needs will increase; if costs are passed on to residents, it could continue to provoke political resistance. How the pledge is incorporated into power agreements and construction conditions will directly impact data center investment returns. Hartnett's warning signal of a "bond-dominated bubble," combined with local construction resistance, points to the cash recovery cycle of AI projects. Persistently high long-term Treasury yields will undoubtedly significantly raise financing costs for tech giants' new AI infrastructure investments and suppress the present value of future cash flows; approval delays will postpone the start of revenue generation. Together, even if long-term compute demand remains extremely strong, annual project returns could decline.
How does the public view data centers? Public dissatisfaction is rapidly increasing. A survey conducted by the University of Pennsylvania's Annenberg Public Policy Center in August found that approximately 61% of American adults oppose building new data centers in their areas, up 12 percentage points from a survey conducted in February and March. In an increasingly polarized country, this has become a rare bipartisan issue: according to a Reuters/Ipsos poll conducted in June, 75% of Democrats and 63% of Republicans said they do not want data centers built in their communities. The Reuters/Ipsos survey also showed that 47% of registered US voters cited the cost of living as the most important factor in determining their midterm vote, and 71% of American adults disapprove of Trump's handling of the cost-of-living issue, with only 33% approving of his overall performance in a parallel survey due to high energy costs and resurgent inflation. These are preliminary media polls and cannot be directly translated into congressional seats or election win probabilities, but they largely indicate that energy, electricity prices, consumer burden, and opposition to data center construction have become essential policy backdrop considerations. Amid widespread public dissatisfaction with the cost of living and the economy, voters say they are increasingly concerned about the impact of data centers on their electricity bills. Monitoring Analytics, an independent market monitor for the nation's largest grid, which covers 13 eastern and central states, reported that overall electricity prices on that grid rose a record 76% in the first quarter due to data center demand. A Bloomberg News analysis found that wholesale electricity costs increased by as much as 267% between 2020 and 2025 in areas with high data center activity. While US tech giants have pledged to pay separately for electricity needed for AI data center development and accelerate the construction of dedicated power plants not yet connected to the grid, how these commitments will be implemented in practice remains unclear as compute demand continues to grow in the short term. Some critics also raise concerns about data centers' water usage, though industry leaders argue these concerns are overstated. Others are unhappy with the loud hum from these facilities or the generous tax breaks they receive. Additionally, broader public anxiety and resistance to AI often manifests as opposition to data centers, with many believing AI poses a threat to jobs and privacy.
How are major political figures responding? Progressive figures like Vermont Senator Bernie Sanders and New York Representative Alexandria Ocasio-Cortez were among the first prominent politicians earlier this year to call for a pause on data center construction. Since then, lawmakers from various political persuasions, including MAGA Republicans and centrist Democrats, have followed suit, calling for a range of policy measures to ensure local residents are not affected by rising energy costs or environmental disruption. Even politicians who were once staunch supporters of data centers, such as Democratic Pennsylvania Governor Josh Shapiro and Republican Texas Governor Greg Abbott, have shifted their positions on the issue. Last year, Abbott called Texas the "center of AI development" when announcing a $40 billion investment from Alphabet's Google. By mid-August 2026, he had enacted measures to pause new data center construction. In Pennsylvania, Shapiro, who initially helped expedite data center development, signed an executive order in August imposing some of the strictest regulations on their construction in the nation. In states like Wyoming, Ohio, Florida, and Pennsylvania, numerous campaign ads feature opposition to data centers as a central theme. As of August, campaigns had cumulatively spent $31 million on ads mentioning data centers. In Florida, Republican gubernatorial candidate Byron Donalds is running ads pledging to ensure data centers don't drive up energy bills. In Ohio, Democratic Senate candidate Sherrod Brown is attacking Republican incumbent Senator Jon Hirst for his previous support of these facilities.
Does this favor Democrats or Republicans going into the midterms? Republicans are clearly on the defensive. Democrats have moved faster than Republicans in sharpening their anti-data center messaging, while the political right is still grappling with Trump's pro-industry stance. Some Republicans have fully embraced moratoriums on data center construction, while others are attempting to craft more nuanced strategies. For instance, Florida's Donalds has proposed a plan echoing Trump's "Electricity Payer Protection Pledge," a voluntary agreement signed by tech companies aimed at protecting American consumers from electricity price increases caused by data center power consumption. But there is growing concern within the GOP that they are losing the messaging war on data centers. In a memo sent to AI companies in August, the National Republican Senatorial Committee warned that Brown is making opposition to data centers the "centerpiece" of his campaign attacking Hirst. The memo cautioned: "Nothing in this race is more of a drag on Hirst's campaign than data centers."
Which 2026 election races could be affected by data centers? The Ohio Senate race is the most prominent example dominated by the data center messaging war, but there are several other similar contests across the country. In Wisconsin, Republican gubernatorial candidate Tom Tiffany is spending on attack ads linking his Democratic opponent to data centers. In Michigan, Republican Senate candidate Mike Rogers has announced his support for a one-year statewide moratorium on new data center construction. In this key swing state election, his stance on the issue aligns him with his Democratic opponent, Abdul El-Sayed. It is difficult to gauge how many voters will prioritize a candidate's views on data centers when casting their ballots, but anti-data center messaging has become an important way for candidates to demonstrate their populist and anti-establishment credentials.
How are core AI companies and data center operators affected by the wave of opposition? A report released on August 26 by the investment firm Kimmeridge Energy Management suggests that up to half of the proposed large-scale AI data centers in the US are at risk of delay or cancellation, with political opposition being the primary reason. According to Data Center Watch, in the first three months of 2026 alone, at least 75 data center projects were blocked or delayed due to local opposition, with a total value of approximately $130 billion. Data center projects face significant delays and cancellations for many reasons, including some grids being unable to handle the massive power requirements and difficulties in sourcing critical electrical components. But political opposition is becoming an increasingly important factor slowing construction. Even when projects proceed, tech companies like Google, Amazon, Oracle, and Microsoft are investing significant resources to win over local communities to avoid backlash. So far, this remains a daunting task. AI companies and their supporters have pledged to spend hundreds of millions of dollars in this year's midterm elections. "Building America's AI," an advocacy group supported by the pro-AI super PAC "Leading the Future," committed on August 31 to spend millions of dollars on pro-data center advertising and publicizing grid plans that isolate data centers from residential power systems in swing states such as Ohio, Kansas, and Wisconsin.
Comments