Power

Simply Wall St identified three utility stocks that have signed long-term power purchase agreements or capacity deals tied specifically to AI data center demand, framing them as beneficiaries of the sustained infrastructure buildout. The piece does not name specific dollar figures in the snippet but points to multi-year contracted revenue as the key investment thesis. Utilities securing long-term AI-linked contracts are increasingly seen as a distinct sub-sector within the broader power market.

The Tennessee Valley Authority says electricity demand across its service territory has set a 2026 record during the current heat wave, with the utility reporting it remains prepared to meet load. TVA serves approximately 10 million people across seven states in the southeastern United States. The authority has not disclosed the specific peak demand figure but confirmed grid stability has been maintained.

Amazon, Google, Microsoft, and OpenAI have endorsed a new power pledge aimed at addressing surging electricity demand from data centers. The commitment signals a coordinated industry effort to tie AI infrastructure expansion to clean energy procurement. Specific targets and timelines under the pledge were outlined as data center demand continues to strain power grids nationwide.

Transformer Magazine examines whether solid-state transformers can meet the power conversion demands generated by rapid AI data center growth, noting that traditional transformers face size, efficiency, and grid-compatibility limits at high densities. Solid-state designs offer faster response times and higher efficiency but remain more expensive and less proven at utility scale than conventional alternatives. Several pilot deployments are underway, with commercial viability expected to depend on cost reductions over the next three to five years.

Fervo Energy and Google have finalized what is being described as the largest-ever geothermal power purchase agreement to supply data centers, according to Factor This. The deal marks a significant expansion of firm, always-on renewable power sourced from geothermal resources for Google's data center operations. Specific capacity figures and contract terms were reported as part of the announcement. This agreement sets a new benchmark for geothermal procurement in the data center sector and could accelerate adoption of the technology by other hyperscalers seeking carbon-free baseload power.

Fortune reports that AI-driven electricity demand is arriving years ahead of the grid infrastructure needed to support it, creating a structural mismatch between data center buildout timelines and utility expansion schedules. Grid interconnection queues, transmission constraints, and permitting delays are cited as primary bottlenecks. Data center operators are pursuing alternatives including on-site generation and long-term PPAs to bridge the gap. The mismatch is expected to sustain elevated electricity prices and limit viable site options in constrained markets through the end of the decade.

Vice President JD Vance has publicly addressed growing American opposition to the Trump administration's data center expansion agenda, according to Newsweek. Vance defended the buildout, citing national competitiveness and AI leadership as justifications. The response comes as communities across the country have organized against data center siting, power demand, and tax incentive packages. The administration's direct engagement with public backlash signals that data center policy has become a politically visible issue requiring executive-level messaging.

The Natural Resources Defense Council has published an analysis arguing that AI data center energy demand does not inevitably strain electricity grids, and that demand flexibility, efficiency standards, and renewable siting choices can reduce grid stress. The NRDC identifies specific policy and operational levers, including time-shifting workloads and co-locating facilities with renewable generation. The report comes as utilities across multiple US regions report that data center load growth is outpacing planned generation additions.

As North Carolina faces rising electricity demand tied to data center growth, utilities, developers, and policymakers are examining microgrids as a partial solution to grid congestion and reliability concerns. Proponents argue that microgrids can provide on-site generation and storage that reduces the strain large facilities place on transmission infrastructure. The state has seen significant data center investment in recent years, particularly in the Research Triangle and Charlotte regions, which has accelerated interconnection queues. Microgrid deployment at scale would require regulatory changes to how utilities handle behind-the-meter generation.

SmartBrief's monthly infrastructure summary for August 2026 catalogued accelerating grid strain driven by data center power demand, identifying a growing gap between transmission capacity additions and load growth from large compute facilities. The roundup cited interconnection queue backlogs, delayed substation upgrades, and rising utility capital expenditure forecasts as compounding factors. The summary did not focus on a single company but reflected conditions across multiple U.S. regional transmission organizations. Grid planners are increasingly treating data center load as a long-duration planning variable rather than a short-term spike.

Fervo Energy and Google have signed a 396 MW geothermal power purchase agreement, expanding Google's clean energy portfolio with firm, around-the-clock geothermal generation. The deal represents one of the largest geothermal PPAs in history, adding to a growing push by hyperscalers to secure carbon-free baseload power. Fervo's enhanced geothermal technology is central to the agreement, distinguishing it from intermittent renewable sources.

Pennsylvania's Public Utility Commission forecasts that data centers will drive electric demand up nearly 19% per year over the next five years, according to a new PUC report. The projection covers utility service territories statewide and reflects the accelerating concentration of data center development in the commonwealth. The PUC's figures emerged shortly after Governor Shapiro signed an executive order aimed at managing data center growth, adding regulatory weight to an already contentious debate over who bears grid upgrade costs.

Utility Dive published an analysis laying out three principles for making AI data centers better participants in the broader electricity grid, addressing concerns about load flexibility, grid interconnection practices, and demand transparency. The piece comes as grid operators across the US face mounting pressure from surging data center power requests. Specific grid operators or utilities are cited in the analysis as examples of where coordination has broken down.

Texas regulators have moved to halt new data center connections to the grid, citing a pattern of 'ghost' demand where facilities reserve far more power than they actually use, distorting grid planning across the state. The Reuters report frames Texas as a case study for a broader US reckoning over how utilities and grid operators account for speculative load reservations. Overbooked capacity forces costly grid upgrades that may never be needed, while also blocking other users from connecting. The situation is pushing regulators in multiple states to reconsider how power reservations for large industrial customers are structured and enforced.

Google has signed a geothermal power purchase agreement with Fervo Energy for a Utah project that represents the largest geothermal deal ever completed, according to stocktitan.net. The project is expected to come online in 2028 and will supply firm, around-the-clock clean power to Google's data center operations. Fervo uses enhanced geothermal systems, drilling horizontal wells in hot rock formations to extract heat at commercial scale. The deal signals that hyperscalers are increasingly willing to fund frontier clean energy technologies to meet 24/7 carbon-free power commitments.

OSV News reports that concerns are intensifying among communities and utility customers over the energy consumption of data centers and the resulting pressure on electricity costs. Ratepayers in several states have raised questions about whether data center load growth is being cross-subsidized by residential and small commercial customers. The article cites ongoing debates in multiple jurisdictions about how infrastructure costs should be allocated.

A new analysis from Virginia Business finds that data center electricity demand in Virginia is growing at a pace that could exceed the state's power generation capacity within the coming years. Northern Virginia alone hosts more than 35% of the world's data center capacity, and load growth projections from Dominion Energy and regional grid operator PJM suggest supply constraints ahead. State lawmakers and utilities are under pressure to accelerate generation and transmission investments.

Data Center Knowledge reports that solid-state transformers are emerging as a viable alternative to conventional iron-core transformers for AI data center power distribution, offering faster response times, smaller footprints, and the ability to handle variable DC loads from GPU clusters. Traditional transformer lead times have stretched to two years or more amid surging demand, creating pressure to qualify alternative technologies. Several vendors are now in pilot deployments at hyperscale facilities.

OpenAI has received regulatory approval for a 3.2 gigawatt power agreement intended to supply a planned data center in Effingham County, Georgia. The deal is among the largest single power commitments tied to an AI operator and would require significant grid infrastructure to support. Georgia's utility regulators signed off on the arrangement, clearing a key obstacle for OpenAI's physical infrastructure buildout in the Southeast. The scale of the power commitment reflects the electricity appetite of frontier AI training facilities.

Chevron's energy supply agreement with Microsoft for data center power drew more industry attention than the oil major's quarterly earnings, according to The Globe and Mail. The deal positions Chevron as a direct energy partner to hyperscale data center operations, a notable expansion of the company's commercial portfolio. Specific terms and volumes were discussed in the reporting. The arrangement illustrates how traditional energy companies are inserting themselves into the data center power supply chain.

Grid operator PJM has floated a proposal that would require new data centers to bring their own power sources rather than drawing solely from the shared grid. The rule would mark a significant shift in how large loads connect to the PJM network, which covers 13 states and the District of Columbia. The proposal comes as PJM faces mounting pressure from surging AI-driven electricity demand that threatens grid reliability across its footprint.

Missouri's power grid requires expensive infrastructure upgrades to accommodate rising electricity demand from AI data centers, according to a report by the Missouri Independent. Utilities serving the state are projecting load growth that outpaces current transmission and generation capacity, creating pressure on both ratepayers and grid operators. The situation mirrors dynamics playing out in Virginia, Texas, and Georgia, but Missouri has fewer existing regulatory frameworks specific to data center load.

Virginia legislators and utility regulators are actively weighing cost-allocation frameworks to determine whether data centers or existing ratepayers should bear the burden of grid expansion required by surging digital infrastructure load. The state hosts the largest concentration of data center capacity in the world, and grid upgrade costs are accelerating. Dominion Energy is central to the debate, as its capital spending plans directly affect residential and commercial electricity bills. A decision on cost allocation could come during the next state legislative session.

PJM Interconnection's projected power shortfall is intensifying scrutiny of data center load growth across its 13-state service territory, which spans major markets including Northern Virginia, Chicago, and Pittsburgh. Analysis from Data Center Knowledge points to the grid operator's capacity auction results and interconnection queue backlogs as signals that supply cannot keep pace with demand from hyperscalers and AI infrastructure operators. Officials have not yet specified how load-shedding priorities would be applied if shortfalls materialize. The situation raises direct questions about whether approved data center projects can realistically draw the power they have been allocated.

Axios published a detailed look at how the surge in data center construction is restructuring American electricity demand, transmission planning, and utility economics at a national scale. The piece traces the shift from incremental industrial load growth to concentrated, high-density demand clusters that stress local grid infrastructure in ways utilities were not designed to absorb. Specific markets including Northern Virginia, Texas, and the Carolinas are cited as focal points where grid investment is accelerating. The analysis arrives as multiple state legislatures and grid operators are simultaneously revising interconnection and cost-allocation rules.

The Union of Concerned Scientists published an analysis arguing that the lack of standardized, publicly accessible energy consumption data from data centers is hampering grid planners' ability to ensure reliability and keep electricity affordable. The organization calls for mandatory reporting requirements, citing the sector's accelerating demand as a systemic risk to grid planning. Without accurate load forecasts from operators, utilities and regulators are making infrastructure investment decisions with incomplete information.

ENGIE has signed a 48 MWac power purchase agreement with QTS Realty Trust to supply renewable electricity to QTS's data center campus in Irving, Texas. The deal adds to a growing list of utility-scale solar contracts executed by major colocation operators seeking to meet clean energy commitments while securing long-term power cost certainty. Terms of the agreement were not disclosed, but the Irving campus is one of QTS's larger operational footprints in the Dallas-Fort Worth market.

Emerald AI closed a $150 million Series A round at a $1.05 billion valuation, with its platform targeting more than 100 gigawatts of untapped U.S. grid capacity, according to Pulse 2.0. The company is positioning its software to identify and activate underutilized grid nodes, addressing one of the primary bottlenecks slowing data center interconnection. The round signals investor appetite for infrastructure intelligence tools as power queues lengthen across major data center markets.

A utility serving Kansas and Missouri is expanding natural gas generation capacity to accommodate data center load growth, and analysts warn residential and commercial ratepayers could face electricity rate increases of up to 10 percent. The deals involve unnamed data center operators securing large power contracts with the utility, which is adding gas infrastructure to meet the demand. State regulators in both states have not yet ruled on whether data center customers must bear a greater share of grid expansion costs.

Ireland is reconsidering nuclear power as a long-term energy source in direct response to electricity demand growth driven by data centers, according to a report cited by Data Center Dynamics. The country has historically opposed nuclear energy, making any policy reversal a significant shift. Data centers now account for a substantial and growing share of Ireland's national grid consumption, straining supply and prompting broader energy strategy reviews.

IPM Newsroom reports that surging data center energy consumption is generating concern among utility planners and consumer advocates about long-term electricity costs and grid reliability. The piece highlights that large-load additions from data centers are complicating capacity planning for utilities in multiple regions. No single company or dollar figure is named, but the analysis points to systemic risk as AI infrastructure build-outs accelerate.

QTS Realty Trust has signed a 48 megawatt solar power purchase agreement to supply renewable energy to its Texas data centers. The deal adds to a growing list of corporate PPAs as operators face pressure from regulators, investors, and local communities to reduce carbon intensity. Texas remains a key growth market for data center operators drawn by available land, deregulated power markets, and fiber infrastructure.

Ars Technica reports that the surge in energy demand from AI data centers is accelerating the development of new power transformer technologies designed to handle higher loads faster and in smaller footprints. Traditional transformer lead times have stretched to two years or more, creating bottlenecks for new data center projects. Manufacturers are exploring advanced materials, modular designs, and faster production methods to close the supply gap.

Data Center Knowledge examines how early grid insight platforms are being used to assess interconnection feasibility, transmission constraints, and queue positions before operators commit capital to site development. The approach aims to reduce the risk of stranded investment when power access proves harder or more expensive than initial projections suggested. Several developers are now incorporating grid analysis into the earliest stages of site selection.

Equinix has signed its fourth solar power purchase agreement in Singapore with Flo Energy, adding renewable capacity to its Southeast Asian data center portfolio. The deal continues Equinix's strategy of stacking multiple PPAs in a single market to build toward its renewable energy targets in a country with limited available solar generation. Singapore's tight land and energy constraints make each additional PPA materially significant for operators seeking to decarbonize there.

Data centers account for 55% of total US electricity demand growth, according to reporting by the Las Vegas Sun. The scale of this concentration reflects accelerating AI infrastructure buildouts by hyperscalers and colocation providers across multiple US regions. Utilities and grid operators are revising long-term load forecasts upward as a result, with some planning cycles compressed from decades to years.

Republican leaders have warned that Ohio's data center energy policy could cost the party its Senate seat, as Lieutenant Governor Jon Husted publicly defended the state's approach to meeting surging power demand from data centers. The political friction centers on how Ohio balances industrial electricity load growth against residential ratepayer costs and grid reliability. President Trump has continued to champion data center expansion nationally, even as the issue becomes an electoral liability in key midterm states.

A Texas legislative hearing on transmission line expansion drew sharp disagreement between residents concerned about costs and land use and industry representatives arguing that new lines are essential to support data center and broader load growth. Witnesses on both sides testified before lawmakers weighing how to fund and site new transmission infrastructure. The dispute reflects broader tensions across ERCOT as power demand from large industrial customers accelerates.

A power utility serving Alabama and other states has established a dedicated rate category for data centers, separating their billing structure from standard commercial customers. The move reflects growing recognition among utilities that data center load profiles, characterized by high and relatively constant demand, require distinct rate treatment. Specific rate terms and the utility's name were not disclosed in available reporting, but the action applies to facilities in Alabama among other states.

Data Center Dynamics published an analysis arguing that the traditional supply-and-demand framework no longer adequately describes how data centers secure power, as grid constraints, regulatory queues, and bilateral utility negotiations have fundamentally altered the process. Developers now face interconnection timelines measured in years, power purchase agreements with complex conditions, and utility reluctance to commit capacity without load guarantees. The analysis draws on recent grid operator data and developer experiences across multiple U.S. markets.