AI energy consumption strains power grid as data center demand drives costs up for manufacturers

US manufacturers' energy costs soar because of AI data center demand

US manufacturers in many Rust Belt cities and towns are paying significantly higher electricity costs as growing energy demand from data centers strains the largest power grid operator in the United States. The resulting squeeze on profit margins for steelmakers and brick factories could further undermine President Donald Trump's "Made in America" plan to revive US manufacturing, and it comes as Trump has simultaneously championed the tech companies behind the AI data center boom. Factory electricity bills are generally rising faster than those for other business customers or residential customers, according to a Reuters analysis. It highlighted the example of the Belden Brick Company, a 141-year-old brick manufacturer in Ohio, whose electricity bills have soared from $1,600 to $12,000 per month due to a higher monthly capacity charge in the 13-state region served by the grid operator PJM Interconnection. Meanwhile, the Steel Manufacturers Association warned that US steel companies concentrated in the Rust Belt region served by PJM Interconnection are paying tens of millions of dollars in higher power costs per year. Electricity accounts for 20 to 40 percent of the total production costs of making steel. Each electric arc furnace used in steelmaking has an operating power load between 40 and 200 megawatts, and the entire US steel industry draws up to 11 gigawatts of power at peak production across all facilities. US steelmakers have benefited from data center construction's requirements for an estimated 1 million tons of steel per year. But data center energy demand has also driven up operating costs for the US steel industry, according to the Wall Street Journal. The Ohio-based steelmaker Metallus described its electricity costs as having jumped by 70 percent since 2024, leading the company to pay an extra $15 million in energy costs annually. The higher electricity costs for manufacturers coincide with many states in PJM territory having attracted large AI data center projects with substantial electricity needs. That data center growth has driven up PJM's capacity prices -- paid to power generators according to supply and demand forecasts -- from $28.92 per megawatt-day in 2024 to $329.17 per megawatt-day in 2026, according to Reuters reporting. PJM has also forecast that electricity demand in its territory will surpass available supply by 6.6 gigawatts starting in 2027, which the Wall Street Journal describes as equivalent to more than six nuclear power plants. No easy fixes Some US manufacturers have raised the prices paid by customers to partially offset their own rising electricity bills, or are even considering relocation of their businesses, Reuters reported. The Wall Street Journal highlighted warnings from steel industry executives that production outages could become more likely if local power grids are overwhelmed by demand. Such results would likely undercut the competitiveness and viability of US manufacturing, which the Trump administration claims to have prioritized despite the loss of 83,000 manufacturing jobs in Trump's first year back in office. The White House has touted getting Big Tech companies to pay for new power generation and transmission infrastructure by signing a Ratepayer Protection Pledge, which happens to lack any meaningful enforcement mechanism. The Trump administration also joined state governors in pushing PJM to hold a one-time backstop auction for purchasing new power supply capacity. But the United States still faces huge challenges in building enough new power generation and transmission lines to support the energy needs of AI data center demand and US manufacturers, not to mention other businesses and residential customers. The Trump administration's efforts to stop renewable energy projects involving wind and solar power have also not helped. In 2025 alone, the United States saw the cancellation of power projects totaling 266 gigawatts of generation capacity -- equivalent to 25 percent of America's current electricity generation capacity and more than the total electricity generation of Texas, according to Michael Thomas, CEO of the Cleanview data platform that tracks renewable energy and data center projects. Clean energy projects accounted for 93 percent of those project cancellations. The Trump administration's cancellations of various wind power projects certainly represented one contributing factor. But other significant patterns included local opposition to renewable energy projects in states such as Ohio and Indiana that were also courting new data center development, along with a lack of new transmission lines leading to high interconnection costs for new clean energy projects, Thomas said. If US states and the federal government are hoping to support local manufacturing, they may need to start making different choices in addressing the rising energy costs of the data center boom.

AI servers will consume more power than all conventional data center hardware combined by 2027 -- global data center electricity consumption set to grow by 26% this year, Gartner forecasts

Gartner projects global data center electricity use hitting 565 TWh in 2026 and topping 1,200 TWh by 2030. Global data center electricity consumption will grow 26% in 2026 to reach 565 terawatt-hours (TWh), up from 447 TWh in 2025, according to a recent Gartner forecast that names power availability as a binding constraint on AI expansion. Worldwide power demand is set to rise 27% to 132 GW over the same period, up from 104 GW in 2025, with consumption projected to exceed 1,200 TWh by 2030. The gigawatt figure measures peak capacity that has yet to be built, permitted, and connected, while the terawatt-hour figure measures the electricity actually drawn over the year. Both, however, are climbing faster than utilities can add supply. "Surging demand for compute-intensive AI workloads is driving unprecedented data center power growth, while AI capacity is now constrained by power availability, making data center power security the new battle ground for scaling and protecting margins in the global AI race," said Gartner's Direct Analyst Linglan Wang. AI-optimized servers consumed about 95 TWh worldwide in 2025 and will draw 175 TWh in 2026, an increase of roughly 84%. Gartner expects that figure to reach 258 TWh in 2027, the point at which AI-optimized hardware will consume more electricity than conventional servers for the first time. By 2030, AI-optimized servers are forecast to account for close to half of all data center power consumption. Conventional servers are effectively flat by comparison. They grew less than 1% in 2025 and are projected to rise 1.2% in 2026 to around 195 TWh, reaching 200 TWh in 2027. Gartner estimates AI-optimized servers will make up 31% of total data center power consumption in 2026, up from roughly 20% a year earlier. Cooling, of course, represents a growing share of the total, with electricity used by cooling systems forecast to climb 22.6% in 2026 to 195 TWh, reflecting the thermal load of denser AI racks and continued capacity expansion. The U.S. accounts for about 204 TWh of the 565 TWh total in 2026, or 36% of worldwide consumption. Of that U.S. figure, dedicated AI data centers consume roughly 68 TWh, or one-third of the national total, while non-AI data center demand in the country has grown only marginally over the same period. Regional grids are already feeling the strain, and more than 75 data center projects worth $130 billion were blocked in the first months of 2026 amid opposition over power and water costs, while some operators have turned to on-site gas generators to bring capacity online without waiting for grid connections. In Virginia, one county asked employees to conserve power as data center demand pushed utility rates higher. In its report, Garner warns that grid supply will be insufficient to meet demand once consumption passes 1,200 TWh by 2030, a shortfall that will affect all data center users, not just AI operators. The forecast accounts for parts and supply shortages, delayed or cancelled projects, and geopolitical disruption, including conflict involving Iran. Wang said infrastructure and operations leaders should prioritize efficiency upgrades, secure grid access, and invest in high-efficiency cooling and edge computing to manage the constraint. Hyperscalers have moved in the same direction, with Meta having signed deals for more than 6GW of nuclear power to supply its upcoming data centers, and one firm repurposing retired U.S. Navy reactors for an AI site in Tennessee. Those projects will take years to deliver, with recommissioned nuclear plants and the earliest small modular reactors not expected online until 2028 or later, leaving power availability as a near-term limitation on the seemingly unstoppable AI build-out. Follow Tom's Hardware on Google News, or add us as a preferred source, to get our latest news, analysis, & reviews in your feeds.

AI's gas-plant boom, and the fight to stop it

Clean-energy advocates cannot outbuild the data centres, so they are going after the regulators instead The AI build-out has done something the fossil-fuel industry could not do for itself. It has set off the largest-ever construction boom in natural gas-fired power plants, the Associated Press reports. Aging coal plants are being kept alive past their retirement dates too. Utilities, plant owners, and the federal government have all pushed to postpone the shutdowns. The reason is unglamorous arithmetic. Some data centres consume more electricity than a mid-size city, and wind and solar cannot be built at that speed. The states drawing lines Several states are trying to force the issue through law. A bill on New York Governor Kathy Hochul's desk would make large data centres hit renewable benchmarks from 2030, reaching at least 90% renewable energy by 2040. Its author, state Senator Kristen Gonzalez, thinks the targets are achievable. These are the wealthiest companies on earth, she argued, and firms able to spend billions on data centres can afford to build the power to run them. Michigan, Oregon, and Minnesota moved first. All three passed laws in the last 18 months to defend existing commitments to emissions-free electricity by 2040. Michigan tied it to money, requiring hyperscale data centres to reach 90% clean energy within six years to keep a lucrative sales tax exemption. Similar bills have appeared in California, Illinois, New Jersey, Pennsylvania, and Virginia. An honest admission from the other side The most useful quote in the story is not a triumphant one. Bob Jenks of the Oregon Citizens' Utility Board conceded the 2040 target was hard to meet with data centres, and hard to meet without them. That is the shape of the problem. The clean-energy goal was already stretching, and AI has arrived and pulled it further out of reach. Households are feeling it first. Electricity bills are climbing across many utility territories, and AI data centres are driving up power costs at Rust Belt factories. The regulatory back door Unable to outbuild the boom, advocates have gone after the rules instead. The tactic is to get regulators to let large power users build their own clean generation and plug it into the grid. Colorado ordered Xcel Energy to create such a programme. In an April filing Xcel accepted it could benefit customers, citing Google projects connecting 115 megawatts of geothermal in Nevada and 1,900 megawatts of wind, solar, and storage in Minnesota. Google's deal with NV Energy is seen as the first of its kind, and the company says similar arrangements are approved or pending in eight more states. The Corporate Energy Buyers Association struck a comparable deal with Georgia Power and is now working on North Carolina. The pitch to utilities is commercial, not moral. They gain a huge long-term customer who pays to expand the grid, rather than watching that customer build standalone generation and leave. Why this is the real fight Grid access is where the outcome gets decided, not the legislature. Regulators have been fast-tracking data-centre grid connections, and whoever controls that queue controls what gets built. Money is chasing the same bottleneck, with Nvidia-backed startups raising to solve data-centre power. Energy, not silicon, is now the binding constraint on AI. Communities are pushing back independently, having blocked 75 data-centre projects worth $130bn in a single quarter. Congress is circling too, with the House voting on a bill to push data-centre energy costs back onto the companies creating them. CEBA's policy chief reckons the decisions being taken now will set energy policy for two or three decades. That is probably right, and it is why a technical argument about grid interconnection is worth more attention than it gets. The gas plants are being poured in concrete while the rules are still being written. Concrete tends to win those races.

What Is the Smartest Way to Power the AI Boom?

Few events in history have tested the U.S. energy grid like the AI revolution. As data centers proliferate across the country, their cumulative power demand is projected to double by next year, rising from 31 gigawatts to 66 gigawatts. In some parts of the country, the energy needs of AI are already outpacing available capacity, driving up consumer electricity bills, raising the risk of blackouts, and increasing reliance on high-emission energy sources. The aging grid clearly isn't equipped to handle AI's surging power demand, and while experts broadly agree that something needs to be done, they have differing opinions on the best path forward. For this Giz Asks, we asked various experts what they think is the smartest way to power the AI boom. They pointed to a diverse array of solutions -- from microgrids to geothermal energy -- underscoring the complexity of the challenge ahead. The following responses may have been lightly edited for length and clarity. Mohammad Shahidehpour University Distinguished Professor, Galvin Chair Professor, and Director of the Robert W. Galvin Center for Electricity Innovation at the Illinois Institute of Technology. Shahidehpour has been the principal investigator of over $80 million in grants and contracts on power system operation and control, smart grid research and development, and large-scale integration of renewable energy. The AI revolution presents one of the most consequential energy challenges of the twenty-first century. The global AI race is often characterized as a competition in algorithms, semiconductor manufacturing, and computational hardware. However, equally important is competition in energy infrastructure. Nations capable of delivering abundant, reliable, affordable, and low-carbon electricity will possess a decisive strategic advantage in attracting AI investments and sustaining long-term economic competitiveness. In this emerging landscape, access to high-quality electric power is becoming as strategically important as access to advanced computing technologies. The most effective strategy for powering the AI revolution rests on several complementary pillars. These include deploying a diversified portfolio of clean and firm generation resources, incorporating water efficiency and climate resilience into data center planning, and establishing market and regulatory frameworks that incentivize flexibility, reliability, resilience, and sustainability throughout the planning and operation of AI infrastructure. Ultimately, global leadership in AI will depend not only on breakthroughs in computer science and semiconductor technologies but also on the ability to build and operate an electric power system capable of supporting unprecedented computational demand. Consequently, AI data centers should be viewed not as passive electricity consumers but as intelligent, grid-interactive assets that actively enhance system flexibility, resilience, and reliability through coordinated demand response, distributed energy resources and energy storage, energy efficiency, and advanced methods for transmission, distribution, and delivery of electricity. The greatest opportunity for us lies in the co-design of AI and energy systems. Rather than planning digital infrastructure and electric power infrastructure independently, future investments should jointly optimize computational workloads, power system operations, communications, electricity markets, environmental sustainability, and water resources. Such an integrated framework will transform AI infrastructure from a rapidly growing electrical load into a strategic asset that strengthens grid performance while accelerating innovation and economic growth. Countries and industries that successfully integrate AI development with intelligent energy planning will not only lead the next generation of computing but also define the future architecture of resilient, sustainable, and secure energy systems. In the coming decades, the true measure of AI leadership will be determined not only by computational capability, but also by the intelligence, adaptability, and sustainability of energy systems that power it. Roland Horne Thomas Davies, Barrow Professor of Earth Sciences at Stanford University and Senior Fellow in the Precourt Institute for Energy. As a leading expert on geothermal energy, he is best known for his work in well test interpretation, production optimization, and analysis of fractured reservoirs. The smartest way to power the AI boom is through Enhanced Geothermal Systems (EGS). Data centers powering the AI revolution demand immense amounts of energy, but their most critical requirement isn't just capacity, it is continuous, 24/7 reliability. While wind and solar provide clean power, their intermittency requires massive battery storage infrastructure or fossil-fuel backups. EGS solves this by unlocking baseload, carbon-free energy almost anywhere on Earth. Traditional geothermal is geographically limited to natural volcanic or tectonic hotspots. EGS eliminates this constraint by applying advanced directional drilling and hydraulic stimulation techniques to access deep, hot basement rock that lacks natural permeability. By injecting fluid into artificially created fracture networks, EGS harvests heat from miles beneath the surface, driving turbines to produce continuous electricity. For tech infrastructure, EGS offers distinct advantages over other alternatives: By combining the predictable reliability of traditional baseload power with the geographic flexibility of modern drilling technology, EGS bridges the gap between massive computational demands and aggressive corporate net-zero targets. It is a scalable, sustainable solution capable of anchoring the next generation of computing infrastructure. Amin Khodaei Professor of electrical and computer engineering at the University of Denver, where he is currently on leave to serve as CEO and co-founder of Gridient, a start-up that offers AI-driven edge solutions for a decarbonized grid. He also serves as vice president of education at the IEEE Power & Energy Society and sits on its Governing Board. Khodaei's research focuses on power systems, microgrids, and grid resilience. The smartest way to power the AI boom is to stop treating it only as a question of how much electricity we can produce. We'll need more power plants and power lines, but the real problem is sharper than that: Can the grid deliver enough power at the exact hours when demand is highest, without risking blackouts or driving up costs for everyone else? A region might have plenty of electricity across a full day and still struggle during the busiest hours, especially in areas where the local grid is already stretched thin. It's like a highway: The issue isn't how many cars use the road over 24 hours, but how many show up at rush hour. The right strategy pairs building more power with two underused tools: flexibility and efficiency. Flexibility means data centers adjust to what the grid needs instead of pulling power at a constant, unchanging rate. When electricity gets expensive or the grid is under strain, some computing work can shift to a different hour, or on-site batteries and power sources can reduce how much power the facility pulls from the grid. But that flexibility only counts if it is actually verified and enforced, through pricing incentives, verified performance, and real penalties if a company does not follow through; otherwise, grid planners still must upgrade the grid as if every data center will demand full power at the worst possible time. Efficiency should also count as a way to add capacity, not just save energy, when it reduces demand at peak hours. Every megawatt a data center uses is competing with homes, electric cars, and businesses for the same constrained grid. Making data centers themselves run more efficiently matters, but reducing energy use in buildings, primarily residential and commercial, also matters because retrofits, smarter controls, and better energy management can free up grid capacity faster than waiting years to build new infrastructure. AI will still require major new investment in grid infrastructure. But treating this purely as a question of building more misses the fastest solution available: getting more out of the grid we already have through flexible data centers and smarter energy usage while new power sources catch up. Costa Samaras Director of the Carnegie Mellon University Scott Institute for Energy Innovation; Trustee Professor of Civil and Environmental Engineering; and an affiliated faculty member in the Department of Engineering and Public Policy. Samaras previously was the Chief Advisor for Energy Policy at the White House Office of Science and Technology Policy. His research focuses on the pathways to clean, climate-safe, equitable, and secure energy and infrastructure systems. The electric power system is the foundational economic, security, and environmental infrastructure of this century, but the grid is aging and is vulnerable to extreme weather events amplified by climate change. AI data centers are helping drive near-term electricity growth, but we have a generational moment to rebuild the electricity system for this century and to power the electrification of buildings, vehicles, and factories. Powering the AI boom in a way that is good for the economy, communities, and the climate requires bold actions across power generation, transmission, and demand. First, we need a grid infrastructure trust fund that helps us to double the capacity of the grid by 2040 -- not just to power AI, but to power the broader electrification needed to address climate change. This could be funded in part by data centers using a "Smart AI Fast Lane" connection fee to quickly connect to the grid when they bring their own clean power and enough additional power and energy storage to benefit surrounding communities. A larger, cleaner grid needs lots of solar, wind power, and batteries right now, as well as ramped up investments in geothermal, advanced nuclear, and broader energy innovation. Next, the wires and equipment that bring electricity to cities and neighborhoods need an upgrade to be resilient to climate change, move more power, and be ready for broader electrification. Data centers, firms, and governments can share in the reinvestment needed beyond the needs of a specific data center, which can help keep electricity costs affordable. The grid is an essential national infrastructure asset, and residential customers shouldn't be required to pay the entire bill for a 21st century refresh. Finally, a grid infrastructure trust fund can support energy efficiency and flexibility, distributed power and energy storage, and virtual power plants, which can reduce peak power demands and keep electricity reliable. In addition, transparency from data centers, a "miles per gallon" efficiency measure for AI, and the reporting of electricity use, emissions, and water use would incentivize best in class environmental performance. Communities are facing new economic and environmental risks, and at the same the grid needs a generational investment to get to a zero emissions future. AI has the potential to help drive better climate and infrastructure outcomes, but this will not happen without focused policy and deep partnerships with communities. Giz Asks is a recurring Gizmodo series in which experts answer big questions in their own words, offering a range of perspectives on the ideas, discoveries, and debates that affect our lives and shape our understanding of the world.

Energy is AI's natural home - but right now, it's fumbling

Data centre electricity demand is set to roughly double by 2030, and AI is both a driver of that demand and one of the best tools available to manage it. When governments and technology leaders convened in Geneva in July for the first session of the UN Global Dialogue on AI Governance, one of the four themes on the table was bridging AI divides -- the digital foundations that decide who benefits from the technology. Few sectors illustrate the issue of access to AI more sharply than energy. And few expose a more uncomfortable paradox. The energy sector should be the natural home for AI. It generates more operational data than almost any other -- meters, sensors, SCADA systems and market signals streaming in real time. The economic prize for using that data well is now quantified: the International Energy Agency (IEA) finds that proven AI applications could cut energy costs in energy-intensive industries by 3 to 10 percentage points, and that well-documented use cases could save more than 13 exajoules of energy by 2035 -- about 3% of global final consumption -- if the barriers to adoption fall. Yet the sector is fumbling its own advantage. The same IEA work finds that energy is not taking full advantage of AI, with weak digital skills and limited data availability as the binding constraints. Both are largely self-inflicted, and both are fixable. The data exists, but it sits in proprietary silos and incompatible formats, locked inside operational systems never designed to share. The skills gap is just as real: utilities and grid operators lose the contest for data scientists to technology firms that pay far more, and few have built the in-house capability to turn raw telemetry into deployable models. This is the familiar shape of energy-transition bottlenecks. The headline ambition -- cleaner, cheaper, more competitive energy -- keeps colliding with the unglamorous plumbing beneath it. Just as connection queues and permitting throttle renewables, weak data foundations and thin digital skills throttle AI-driven efficiency. That readiness gap is why the "energy for AI" and "AI for energy" debates cannot be separated. Data centre electricity demand is set to roughly double by 2030. If the sector absorbs that load without deploying AI to offset it, the net effect on the energy system is negative -- more demand, no efficiency in return. If it adopts, the efficiency dividend plus AI-enabled flexibility can tip the ledger the other way. The flexibility piece matters. Operated more responsively, data centres could provide 50 to 60 gigawatts of flexibility to power systems over the next five years. This would help to integrate renewables and defer costly grid investment, and AI is what makes it possible. The technology creating the demand problem is, in other words, also one of the better tools for managing it -- but only for the systems equipped to use it. The net energy effect of AI is therefore not a property of the technology. It is a choice made in procurement decisions, hiring plans, data-sharing rules and regulatory design. In some places, the fix for this problem is being taken seriously. Take India. In 2025, the country's Ministry of Power launched the India Energy Stack -- a digital public infrastructure for the power sector, modelled explicitly on the Aadhaar identity system and UPI payments rails. Its premise is precisely the diagnosis above: data stays where it is generated, but a shared layer of unique identifiers, open APIs and consent-based exchange standardises the interface between fragmented utilities. It treats interoperable, AI-ready data as the foundation -- an attempt, in effect, to dismantle the silo problem at the root rather than paper over it with another pilot. Others are converging on the same insight from different directions. In June 2026 the European Commission published its Strategic Roadmap for Digitalisation and AI in the Energy Sector, pairing investment in grid-management AI models and a pan-European AI.grids initiative with measures on data governance and digital skills -- the two binding constraints, addressed together. In the US, the Department of Energy's Genesis Mission places AI at the centre of grid planning and nuclear deployment, drawing on decades of national-laboratory data. The Department expects decisions on grid planning, interconnection and operations to run 20 to 100 times faster, with electricity cost and reliability gains of up to 10%. It is separately turning 80 years of nuclear research into a secure, searchable database that future energy and security decisions can draw on. India's public-infrastructure model, Europe's sovereign-and-shared approach and America's laboratory-and-compute push are three different theories of the same fix, and their divergence is itself the lesson. A roadmap is not deployment, and the history of energy digitalization is littered with promising pilots that never scaled. The hard work is cross-cutting and slow: common data standards and protocols, interoperability, a workforce able to operate these tools and governance that earns trust in AI on critical infrastructure. None of it can stay national. Grids, supply chains and AI models are international; fragmented approaches will blunt the dividend and widen the very divides the Geneva meeting aimed to close. This is where a UN dialogue matters more than another national strategy. The energy transition will not wait for each country to solve interoperability alone, and the systems with the weakest data foundations are risk of being locked out of the efficiency gains. Turning national pilots into shared standards and transferable practice is unglamorous, multilateral work. It is also the difference between an AI dividend captured by whoever solves interoperability first and one shared across the global energy economy. Much of the conversation in Geneva rightly focused on managing AI's growing energy demand. That work is essential, but it is only half the story. The larger opportunity lies in how AI can reduce energy use across industry, buildings, transport and power systems, while enabling greater integration of renewable energy. The efficiency gains are proven and quantified. The barriers are known and addressable. What remains is execution: better data, stronger skills, institutional will and more effective international cooperation.

AI data centres are driving up power bills at America's Rust Belt factories

Capacity charges have soared across the PJM grid, and manufacturers from Ohio to Pennsylvania are paying for the AI boom. For years, electricity costs at the Belden Brick Company in Sugarcreek, Ohio, barely moved. Last year they jumped by 90%, driven largely by the data centres multiplying across the region to feed the AI boom. The 141-year-old manufacturer, whose bricks feature in landmarks including the Alamo and the University of Notre Dame, traced most of the pain to one line on its bill. Its monthly capacity charge climbed from $1,600 to $12,000, part of the same strain now showing up as households across Europe are asked to cut power use and US utilities line up $1.4 trillion in grid spending. Belden Brick is one of many manufacturers across the heartland facing the squeeze, according to a Reuters review of energy data and interviews with nearly a dozen firms. Factory power bills, a core cost, are rising faster than those for most homes and businesses. The pressure is concentrated in the 13-state region run by grid operator PJM Interconnection, which runs from New Jersey to northern Illinois and south to Tennessee. A single server warehouse there can use as much power as a mid-sized town, and five of the eight states seen as emerging data centre hubs sit in the Rust Belt, per Synergy Research Group. Capacity charges, which pay generators to keep supply on hand for peak demand, have soared. PJM's price leapt from $28.92 per megawatt-day in 2024 to $329.17 now, a rise of roughly 1,038%, driven mainly by data centres, which made up about 40% of the record $16.4 billion in costs from its most recent auction. Those charges filter down unevenly. Average industrial electricity prices rose 31% in Pennsylvania and 26% in Ohio in the year to December 2025, against a 7% national rise, per Reuters calculations of Energy Department data. Households in the two states saw gentler increases of 14% and 9%. Supply is not keeping pace. Data centres "can be built faster than the generation needed to serve them," PJM spokesperson Jeff Shields said. Last week the operator asked some users to curb consumption to avoid rolling blackouts, as a heatwave pushed peak demand to a record. Data centre advocates counter that the surge is forcing overdue grid upgrades. Aaron Tinjum of the Data Center Coalition also points to plant retirements and transmission limits as culprits. For owners on thin margins, even a small rise bites. Belden has lifted brick prices by 4% and still seen profits shrink. "There are going to be some companies that are on the razor's edge," said company president Brad Belden. Others are improvising. Plaskolite, a plastics maker, saw annual capacity charges across its Pennsylvania and Ohio plants jump to $1.2 million from $200,000, and is weighing a direct natural gas feed. Tosoh SMD, an electronics-materials firm in Grove City, Ohio, is mulling production on the graveyard shift, when power is cheaper. Regulators are responding, though not always in factories' favour. Manufacturers sit in the same rate class as data centres, so rules meant to shield households can catch them too. The Federal Energy Regulatory Commission wants firms with onsite generation to pay transmission charges on that power, which manufacturers are appealing, and at least 10 states have their own pending data centre rules. The fight mirrors a recent House vote on who pays for AI data centre energy. The stakes are political. Rising bills threaten some plants just as President Donald Trump pushes domestic manufacturing. The White House said he has hosted tech firms signing a "ratepayer protection pledge" and ordered new PJM power plants, funded by the tech companies. Industry, meanwhile, is betting on nuclear, including a record run of SMR deals. "Manufacturers are not data centres," said Paul Cicio of the Industrial Energy Consumers of America. For now, the build-out promising America an industrial revival is quietly raising the cost of keeping its oldest factories running.

Stymied datacentre projects threaten global AI revolution

Large-scale datacentre projects around the world are being challenged or cancelled, as infrastructure's energy demands ramp up Datacentre planning proposals face all kinds of hurdles, from securing energy supply to high construction costs. But the 2,000 acre Prince William Digital Gateway site in the US state of Virginia had another problem: its proximity to a Civil War battlefield. "If the development is allowed to proceed, the solemn nature of this historic site would become marred by sitting in the shadow of the monstrous datacentres, along with their associated electrical infrastructure," said one legal brief against the plans. The Gateway project is now in doubt after a local court ruling halted the project and a key backer pulled out. It is one of hundreds of large-scale datacentre projects around the world that are in various states of development, from chancier attempts at riding the AI boom to the more committed projects that have the support of tech behemoths like Microsoft. But while models produced by cutting-edge AI companies like OpenAI, Anthropic and Google are improving rapidly, the central nervous systems behind their technology - datacentres - are being built at a much slower pace. The Uptime Institute, which inspects and rates datacentres, has identified 250 global datacentre projects exceeding 100MW in energy demand - equivalent to around 300,000 homes - that have been announced between 2021 and 2024. It said approximately half of those projects will either not happen, or their completion will be delayed. Even if the cancellations and delays came to fruition, there will still be an "unprecedented and rapid" increase in the power required over the next five years, according to Uptime. Mega-projects cancelled last year include Project Range in the US state of Arizona and the Cyberjaya campus in Malaysia. The Prince William Gateway is also on the cancelled list. This backlog poses problems for AI firms that need datacentres to train and operate their models. Google has admitted its cloud business - which uses datacentres to provide AI services like chatbots to companies and users - is "compute-constrained", as demand for ever more powerful AI models and services increases. Jay Dietrich, a research director at Uptime, says a number of factors are working against proposed datacentre projects. Those include: proposals from developers without datacentre experience and don't have committed tenants; the size, scale, and energy and water consumption of individual projects and the concentration of these projects in "datacentre corridors" where projects are concentrated; and supply chain issues, including getting the chips to go in them. "The global supply chain just cannot support the level of projects out there, on the timeline that is projected. The scale is such that it's going to slow things down," he says. And, as the Prince William legal brief shows, the perennial issue of opposition from local community and environmental groups is another to consider. Uptime says we are entering an era of mega-gigawatt datacentres. It identified six projects last year, each aiming for at least 5GW of power - five in the US and one in the United Arab Emirates. To put that in perspective, Ireland's peak energy demand is 6GW. The energy demands are vast. Taking planned projects announced last year alone, and assuming they run at 25% of planned power capacity, they would consume 1.3% of the world's projected electricity usage for 2025, according to Uptime. It's a near-doubling of current datacentre demand. About 80% of the new power demand is coming from US projects. Uptime is not optimistic about these power needs being met. "Surging datacentre power demands, particularly in N[orth] America, cannot be supported by power grids already operating under heavy strain," said Uptime in a January report. In California, datacentres are standing empty for years because the local grid cannot supply them with power. In Amsterdam, an Australian datacentre developer recently sued the Dutch grid after its request for a connection was turned down - an eerie development that signals the potential for growing conflict between datacentre projects and the houses, hospitals and businesses that also need that electricity. In a heating world with mounting geopolitical instability, the choices involved in building massive AI datacentres - as opposed to directing those resources elsewhere - will become even more stark. In the UK, the Guardian's investigations have shown that the government's sweeping ambitions to make Britain an AI superpower appear to be underpinned by minimal attention to what tradeoffs - and resources - that might require. In announcing a series of multi-billion-dollar projects to mainline AI "into the veins" of Great Britain, the government did not even bother to audit the promised sums. In choosing sites for the UK's largest AI developments, it appeared to pay little heed as to whether or not they had electricity. Some observers are more upbeat. JLL, a US property consultancy, expects that around 1,200 datacentres will be built globally between now and 2030 - with demand overwhelmingly driven by AI. Andrew Batson, global head of datacentre research for JLL, says he is confident the capacity will be built, adding that lease signings and groundbreakings for the first half of 2026 are slightly ahead of his estimates. Citing factors such as improvements in battery storage and onsite power generation - ie not leaning so much on the local grid - he says energy constraints can be overcome too. "I am confident that the industry will work through the energy challenges," he says. "Energy constraints will not go away, but the industry has been developing and implementing solutions for a number of years and that legacy of innovation will continue." According to an Uptime report published in January, the seven largest planned datacentres in the world are proposing a combined 45GW of onsite power, with gas as the primary energy source. The UK's peak energy demand is 45GW. The Prince William Gateway submission goes on to acknowledge that datacentres are a "fundamental part of the technology infrastructure that supports the modern economy." But local resistance and universal problems, like energy provision, are hampering this global revolution.

'The challenge is no longer only how much power is needed, but whether it can be delivered reliably': Report finds AI data centers are draining more power than the grid can provide

* Electricity demand is now growing faster than energy suppliers can keep up with * Volatile AI workloads cause unpredictable peaks and troughs in demand * AI could actually help predict, despite also being the cause With three in four (77%) electricity execs now believing that data center energy demand will grow faster than utilities can keep up with, two-thirds (68%) expect electricity shortages to become more commonplace as demand for AI soars. New data from a Capgemini report reveals just how unpredictable AI energy demands can be, with 77% admitting they struggle to accurately forecast demand amid volatile AI workloads. Not only is this leading to more constrained energy supply, but also more extreme and less predictable demand spikes. Data center energy demand is a whole new ball game All of this comes as local opposition continues to mount against data centers, with residents increasingly concerned about power outages and rising energy costs. Just last week, a county in Virginia told data centers to revert to backup generators to free up grid capacity for local residents, with an ongoing heatwave causing a spike in electricity demand for air conditioning units. Even data center companies are struggling to anticipate how much they could consume, with 67% of electricity execs reporting speculative applications for future capacity. Around a fifth (19%) of these don't even materialize, creating what Capgemini calls 'phantom demand,' forcing utilities to either overinvest unnecessarily or underinvest and create capacity shortages. "The challenge is no longer only how much power is needed, but whether it can be delivered reliably, where and when it is required," Capgemini Global Head of Energy and Utilities Claire Gauthier wrote, citing AI's potential in helping to predict demand despite also being the cause of fluctuating and high demand. However, at the moment fewer than half (45%) currently use AI for grid optimization. Looking ahead, most (87%) data center operators expect electricity consumption to rise over the next three to five years by an average of 30%. Follow TechRadar on Google News and add us as a preferred source to get our expert news, reviews, and opinion in your feeds.

Is AI the energy technology the world has been waiting for?

AI could become the technology that finally makes the grid work the way it always should have. The most consequential energy technology of the coming decade may not look like a power plant at all. It may look like an 'AI factory' or what NVIDIA's CEO, Jensen Huang, calls the new generation of computing campuses that specialize in producing tokens of artificial intelligence. Calling AI a revolutionary energy technology sounds strange. The dominant narrative about AI and energy is the opposite: AI has a voracious appetite for energy and represents a looming crisis that could break the grid and raise everyone's power bills. Data centres already consume roughly 6% of all electricity in the United States and the United Kingdom. The International Energy Agency projects that global AI data centre electricity demand could more than quadruple by 2030. No industrial energy load in modern history has grown like this. But the framing of 'AI as load' is wrong. AI is not just an unusually large new load. It is the most software-defined, the most controllable and the most spatially mobile workload ever to consume electricity on an industrial scale. Orchestrated carefully, power-flexible AI factories -- data centres engineered to modulate their own electricity use in real-time -- become a new energy technology and could become one of the most consequential in history. They could reduce power bills, protect grid reliability and unlock massive power capacity to turbocharge the AI revolution. Consider what we usually mean by an 'energy technology.' Some energy technologies create new supply - solar panels, nuclear plants and the fusion reactors that are on the horizon. Others reshape how energy moves and is stored -- high-voltage transmission, lithium-ion batteries. A third, quieter category transforms how energy is consumed: heat pumps replaced furnaces, variable-speed motors replaced fixed-speed ones, LED light bulbs displaced incandescents. AI belongs in that third category, and it could have even more potential than its predecessors. Through intelligent and flexible energy management, AI factories become precise and controllable assets on the power grid. As the grid approaches peak energy demand on a hot summer day, AI factories can dynamically slow down AI jobs that are inherently flexible, whether research workloads, model fine-tuning or batchable inference jobs that can be paused and rescheduled. Even the workloads that must run in real time -- a chatbot query, an autonomous agent's action -- can be routed at the speed of light to a region where power is plentiful. No other large industrial load has this combination of flexibility across both time and geography. A steel mill cannot relocate from Phoenix when Texas peaks. A semiconductor fab cannot slow or pause for two hours and resume seamlessly. AI can, while meeting the performance requirements of its users. To be sure, at first glance, the economics make this sound impossible. A one-gigawatt AI campus spends roughly $5 billion a year servicing the cost of its GPUs and another $2.6 billion on the building and network around them. Its annual electricity bill -- about $590 million -- is more than ten times smaller. Why would any operator ever throttle billions of dollars of accelerators to chase a discount on a comparatively cheap input? But this objection misunderstands the scale of the opportunity for AI to serve as a revolutionary energy management technology. AI factories can unlock billions of dollars of annual value by generating tokens of artificial intelligence, far outweighing the rare cases when flexibly throttling power can ease the grid's peak strain. Power systems are built to peak, which means they sit underused most hours of every year. Flexible AI factories monetize that latent capacity. Independent analyses by Duke University put the unlock at roughly 100 gigawatts on the existing U.S. grid alone -- enough to absorb several years of AI growth without a single new transmission line. And, by avoiding expensive new grid upgrades while better utilizing existing grid infrastructure, flexible AI factories can actually reduce power bills for local communities - curbing the powerful political backlash currently brewing against AI infrastructure that could raise local power bills. Renewable generation is cheap and abundant but uneven and the grid needs demand that can absorb surpluses and step back during shortfalls. The flexibility tools we have today - utility-scale batteries and a thin set of legacy industrial demand-response programmes - are either expensive, slow or both. Flexible AI factories can flex faster and at greater scale than either. There is a still deeper implication. The world's AI factories are increasingly interconnected by fibre, software and shared standards. They are starting to behave less like isolated industrial sites and more like nodes on a single, planet-spanning network -- a kind of complementary grid sitting on top of the public power grid. The electric grid moves electrons. The AI grid moves computation. When those two grids are designed to talk to each other, they become more useful than either could be alone: power can be routed to where computation is and computation can be routed to where power is. AI flexibility is not just a theory. At Emerald AI, we have demonstrated grid-responsive AI infrastructure at five commercial data centres worldwide over the past year, including on the latest NVIDIA Blackwell Ultra systems. We have shown live, on real workloads, that AI factories can cut power on command in seconds and sustain reductions for hours without losing performance on the most crucial workloads. Silicon Valley Power (SVP), the municipal utility of Santa Clara, has launched a first-of-its-kind programme in which Emerald administers flexibility for major data centres in the heart of Silicon Valley, allowing SVP to offer upsized capacity to customers it could not previously serve, while protecting rate affordability for residents. And, later in 2026, NVIDIA, Digital Realty and Emerald AI will bring online the world's first commercial-scale, power-flexible AI factory -- a 96MW facility in Virginia capable of modulating its power use in response to signals from the grid. Google has run a carbon-intelligent computing platform for several years, shifting non-urgent workloads in time and across regions to match cleaner grid hours. The Electric Power Research Institute's DCFlex initiative, launched in 2024 with more than twenty utilities and hyperscalers, is running multi-site demonstrations of data-centre flexibility on real grids. The opportunity is most acute outside the United States. Ireland's grid operator has restricted new data-centre connections in the Dublin region. Singapore lifted its moratorium only after introducing strict efficiency standards. India is forecasting data-centre capacity to roughly triple by 2030 while simultaneously electrifying industry and transport. The countries that figure out how to make AI infrastructure flexible by default will have a structural advantage in attracting compute investment without sacrificing reliability or affordability for the rest of the grid. The Forum's Centre for Energy and Materials is already convening utilities, regulators and hyperscalers around precisely this question; that work needs to accelerate. Taken seriously, this reframes the AI energy story. Yes, AI will require enormous quantities of electricity. But because AI is software-defined, those same workloads can be made to align with the grid, rather than fight it. Flexibility does not shrink AI's appetite for power; it reshapes it. What strains a grid -- driving up bills and prompting the connection moratoria now seen from Dublin to Singapore -- is not the total electricity a region consumes over a year, but the load it draws at the system's tightest and most expensive hours. By pulling back when the grid is scarce and leaning in when power is abundant, flexible AI factories can add their enormous new demand without piling onto that peak: easing, rather than worsening, the risk of blackouts and rate increases and sparing communities the cost of generation and wires that would otherwise sit idle most of the year. In the United States, where data centres are on course to account for nearly half of all growth in electricity demand this decade, that difference is the difference between a grid that buckles and one that absorbs the boom. More power generation, like nuclear reactors, is essential, but the largest, most controllable and most rapidly scaling new participant on the world's electricity grids today is not a power plant. It is computation itself. If we design AI infrastructure to be flexible by default -- and if regulators and utilities reward that flexibility with faster interconnection and access to grid-services markets -- AI will be remembered not as the crisis that overwhelmed the grid, but as the technology that finally made the grid work the way it always should have. A new energy technology has arrived. It is made of GPUs, fibre and code. And it is just getting started.

AI data center servers predicted to glug more power than 'conventional servers' by 2027

Data centres require a lot of energy to run. It's why it's so frustrating to hear AI's major players attempt to play a game of misdirection by making claims along the lines of a single LLM prompt requiring but a fourteenth of a cup of tea. Worse still, as big tech continues to build out its AI infrastructure, data centres' power demands are only set to increase. Worldwide data centre electricity consumption is expected to rise from 447 terawatt-hours in 2025, to 565 terawatt-hours in 2026. That will mark a 26% year-over-year increase, according to Gartner's latest forecast. Data centres' power demand globally is expected to rise 27% this year, peaking at a predicted total of 132 gigawatts. That's up from 2025's 104 GW total, with analysts expecting to see power demand continue to rise -- according to Gartner, data centre demand may cross the 290 GW mark by 2030. Gartner's Director Analyst, Linglan Wang, explains, "Surging demand for compute-intensive AI workloads is driving unprecedented data center power growth, while AI capacity is now constrained by power availability, making data center power security the new battle ground for scaling and protecting margins in the global AI race." Gartner estimates that AI-optimized server adoption will account for 31% of data centre power consumption this year, having grown by 84.2%. By 2027, the power demands of AI servers is expected to grow by 47.8%, surpassing that of 'conventional servers.' Comparatively, the power demands of the 'conventional server' segment grew by about 1.2% in 2026, and is on a trajectory to grow by 2.4% in 2027. Given that recent analysis of Google's latest environmental report suggests the company's energy consumption is on a trend of exponential growth, an upward trend across the industry is unsurprising. It's also perhaps worth noting that not all of Google's energy comes from renewable sources, and that the total electricity consumption for this one company last year was 43 TWh. That's almost 10% of Gartner's global numbers. Add to that a reported 80% of the world's data centres have been built in less than optimal climates, having to compensate with a whole lot of temperature control systems, and environmentally, things start to look all the bleaker for the climate conscious. In fact, Gartner lists the power consumption of the 'Cooling and other Infrastructure' data centre segment as having grown by 22.6% in 2026. Not only that, but increased demand from data centres will likely tax local power infrastructure -- not to mention any of the other reasons why data centres make especially poor neighbours.

Why AI's energy future depends on power from space

Global electricity demand from data centres will more than double by 2030, according to the International Energy Agency (IEA). New analysis from United Nations University suggests that if data centre growth continues on its current trajectory, power demand could approach three times the combined annual electricity consumption of Pakistan, Bangladesh and Nigeria this decade. AI is forcing an urgent question onto the global agenda: Where will all the power come from? Tech executives are blunt about the current energy supply bottleneck. "Power is my problem today," Microsoft CEO Satya Nadella said on a recent podcast. The scramble for power is already reshaping infrastructure decisions. Market intel firm Cleanview has identified 84 GW of proposed data centre projects that plan to deploy onsite gas-fired generation as developers rush to develop power sources. When companies are willing to deploy gas turbines just to get access to electricity faster, it becomes clear how valuable new sources of power could be. But data centres' need for massive amounts of reliable, always-on power is colliding with slow grid expansion, transmission bottlenecks and the realities of building new infrastructure. Those same pressures will also come for the rest of the global economy in the years ahead. The scale of innovation has always depended on the scale of energy. Space now presents a critical new domain for energy access. For decades, space infrastructure has quietly supported life on Earth through communications, navigation, weather forecasting, remote sensing and national security. Now space is expanding from being an information layer to become an energy layer as well. With mounting pressure on terrestrial infrastructure, new technologies can take advantage of continuous solar energy, one of space's most abundant resources. Solar panels have become indispensable in space, powering satellites long before they became commonplace on Earth. Now, advances in launch, optics and manufacturing could make it practical for space to power Earth For most of the commercial space era, the industry's defining challenge was access to orbit. SpaceX has dramatically lowered the cost of reaching orbit, and now companies are shifting to creating meaningful value for Earth once they're there. AI and energy demand are an urgent driver. Companies like Starcloud and SpaceX are exploring whether AI workloads should be co-located with the abundant energy in orbit. The appeal is obvious. Moving compute closer to a virtually continuous energy source could reduce dependence on increasingly constrained power infrastructure and siting environments. The tradeoff is that energy is only one part of the equation. Latency, thermal management, maintenance and hardware replacement are central problems to solve in the architecture. Cooling is a good example. Modern AI systems generate enormous amounts of heat and operating those systems in orbit introduces a different set of challenges than operating them on Earth. Rather than relocating demand, space solar energy expands the supply of available energy on Earth by delivering power from space directly into terrestrial electricity systems. That creates opportunities to build on infrastructure that already exists. For example, Overview Energy is designing the technology to use utility-scale solar projects as receiving infrastructure. It collects energy in orbit and transmits it using safe, invisible, near-infrared light optimized for photovoltaic panels, allowing solar assets to generate electricity at any hour. Energy from orbit then flows into infrastructure that is already connected to the grid rather than requiring entirely new sites. That matters in a world increasingly obsessed with speed to power. The remaining questions are primarily industrial: how quickly these systems can be manufactured, deployed and scaled at the right price point. Data centres are receiving most of the attention right now, but economies will continue to need more reliable electricity for other uses. Advanced manufacturing, industrial electrification, desalination, hydrogen production and future digital infrastructure will all require abundant electricity. And access to power shapes what gets built and where. The industry has spent the last decade building a huge amount of renewable infrastructure. The largest hyperscalers alone have contracted roughly 30 GW of solar capacity. Today, those assets generate electricity only when sunlight is available. That's one of the most compelling parts of delivering energy from space straight to solar projects. Instead of building an entirely new class of energy asset, you increase the utilization of one that already exists. A solar project that generates power for more hours of the day produces more energy, creates more value for the grid and improves the economics of infrastructure that has already been built. Every major wave of innovation eventually encounters physical constraints. Energy is the constraint of our era. Meeting that challenge will require expanding access to reliable power in ways that were difficult to imagine even a decade ago. For decades, space infrastructure has helped move information around the world. In the future it will move energy.

The world's biggest battery maker on AI's energy demand

Engaging the whole system and co-designing from the outset are increasingly regarded as key to long-term success. For many leaders, businesspeople and policy-makers worldwide, the future of AI, including how it will be powered, is one of uncertainty, high costs, and some degree of anxiety. Pushing back against this prevailing sentiment is Robin Zeng, Founder, Chairman and Chief Executive Officer, Contemporary Amperex Technology (CATL), who provided a very different vision while speaking at the Annual Meeting of the New Champions in Dalian, China. In "No Power, No AI", Zeng revealed that in China the demand on the grid from the country's burgeoning number of data centres is small, relative to the country's grid growth. "The energy system is so mature... the AI data centre consumes not much electricity compared to China's growth. So, the grid is no problem." Instead, he argued that the real question for China's business and political leaders isn't one of capacity, it's how the power is sourced. Regulation in China dictates that all new data centres must employ 80% renewable energy, a situation that is accelerating research into grid stability and battery technology, with energy storage a key issue. According to Zeng, there are three phases to consider when assessing the maturity of energy storage solutions: technology capability - can the supplier provide a reliable and constant supply of energy to support a big data centre (1 gigawatt of power year-round); is it cost-effective (cost must be equal or lower than traditional energy to be competitive); and long-term reliability and performance. Currently, nearly one in five large-scale energy storage power stations worldwide are underperforming, underscoring why continuing innovation in this area is vital. Concurrently, supply chain reliability is another pressing issue. CATL is developing sodium-ion batteries to help reduce China's dependence on lithium (defined as a critical mineral). Zeng revealed that his company has already produced a sodium-ion battery that can be deployed on a large scale, and in three to five years, he expects these batteries will be able to reach 100 gigawatt-hours every year. In doing so, they will be able to fully support a modern data centre. "We can produce a large-scale sodium battery for energy storage, so we get rid of the lithium dependence." China is already using AI to optimize the efficiency (particularly energy usage) of its data centres. AI systems purchase electricity when prices are low, while also maintaining operational stability at facilities. Speaking about CATL's AI usage, Zeng revealed that the company is already making savings of approximately 30% on its electricity bills: "We're already using [it] for some data centres. We can have the AI auto-bidders buy... low-cost electricity from grid supply to our manufacturing plants, and also keep the manufacturing plants very stable." Zeng's vision for the not-so-distant future is one of even greater integration. Vehicle-to-grid technology already allows EVs to supply electricity back to the grid, which for a country where, according to Zeng, there are already more than 40 million EV cars, is a very attractive proposition. At battery swapping stations, there are batteries with large capacities that can store energy, particularly overnight, helping to balance renewable energy and supply. Zeng revealed a future where EVs are a lot more than just a transport option, but instead become "tokens" with their valuable battery and computing resources used in energy and digital systems when not in use by the owner. "You didn't use your battery, didn't use your chips, didn't use your computing power - so you can use that as a token, if you do the technology right." This becomes a societal win-win, with EV car ownership supporting wider grid and digital stability and power, while also benefitting the car owner. Zeng is certainly not alone in envisioning a future where AI and the data centres that power them return power to the grid. Vanessa Chan, Inaugural Vice-Dean Innovation and Entrepreneurship, University of Pennsylvania, raised the issue of "new architectures", which focus on innovation in the way in which systems are approached, communities engaged, regulation and taxation addressed, and technology applied. Like Zeng, Chan argued that we shouldn't regard data centres as a drain on the grid, but instead a dynamic system, which can "flex". "We think too much about the wires and all that, but flexibility itself becomes an actual asset. An AI centre could be a flexible place to bring energy back into the grid." For Chan, however, a starting point is the re-evaluation of not just energy supply but also demand. How can large users, like data centres, consume less? Chan and her colleagues are researching ways to make data centres more efficient, for example, by replacing general AI models responding to tasks with specialized ones that are more energy efficient. Chip configurations are also being studied, including researching monolithic 3D configurations, which are four times faster than the current 2D ones, while battery technology - particularly storage and distribution - is being doggedly pursued. In the US, policy-makers are increasingly demanding that tech giants such as Google develop local energy infrastructure and support grid modernization, as well as "start funding batteries, heat pumps and electric vehicle charging". In Virginia, state authorities have been creating "complicated and sophisticated" large load tariffs, while in Georgia, large customers help fund new clean energy resources, receiving energy value credits in return. "We need to think about how distributed energy resources are not just market participants, but are actual planning assets that are helping with the grid and... some of the load challenges." There's also been a shift towards "grayscaling," where data centres take on stranded assets like retired coal plants and transform them into digital hubs designed to serve the local community. "If there's ways to repurpose assets that are stranded, that's another win for a community." This shift to engaging far wider ecosystems requires much broader thinking. Reflecting this, co-designing is becoming more common. Instead of designing everything in isolation, power, storage and computing need to be co-designed from the outset, a situation that participants agreed would boost energy affordability, ideally in a way that's beneficial to the wider community.

AI's energy appetite is reshaping the electric grid

The five largest technology companies on earth spent more than $400 billion on capital expenditure in 2025, most of it toward AI. Chips, data centers, servers, and software pipelines absorbed capital at a pace the technology industry had never seen. The constraint holding the whole machine back has almost nothing to do with any of those things. According to the International Energy Agency, global data center electricity consumption is on track to roughly double by 2030, reaching levels equivalent to Japan's entire annual power demand today. In the United States, data centers are projected to account for nearly half of all electricity demand growth through the end of the decade. That shift is creating a secondary investment cycle in energy infrastructure that is reshaping which companies and regions matter most to AI's next phase. Electricity is becoming as strategic as semiconductors. Training and deploying frontier AI models burns through electricity at a scale that was barely imaginable a decade ago. The largest data center campuses consume as much power as small cities, and the demand is accelerating faster than grid infrastructure can accommodate. Utilities are scrambling. Permitting timelines are stretching. The IEA estimates that roughly 20% of planned data center projects globally are already at risk of delays caused by grid constraints. Transformer and cable delivery lead times have doubled in the past three years. Building new transmission lines typically takes four to eight years in advanced economies. Leo Fan, founder of Cysic, believes the bottleneck is now physical rather than technical. "Yes. The constraint is no longer just chips or capital. It is deliverable power, which includes generation, transmission, cooling, interconnection and more. AI growth will increasingly depend on who can secure reliable electricity," Fan said in an interview with TheStreet. The shift is already visible in the data. Goldman Sachs Research projects US data center power demand will more than double to 66 gigawatts by 2027 from 31 gigawatts today. PJM, the grid operator covering a large stretch of the northeastern United States, projects that data centers will account for 30 of the next 32 gigawatts of load growth by 2030. That is not an AI story. It is an infrastructure story. The connection problem that generation alone won't solve The instinct when power demand rises is to build more generation. The people working closest to grid infrastructure say that instinct is missing the actual problem. Samuel Videau, chief technology officer at Genius, put it plainly. "Everyone talks about generation. That's not the bottleneck. The bottleneck is connection. When a hyperscaler tries to buy a [digital] miner just to skip the interconnection queue, that tells you everything. Transmission and permitting are the real constraints. Connected megawatts trade at a premium to everything else in the sector," Videau said in an interview with TheStreet. The CoreWeave story makes his point concrete. CoreWeave's Core Scientific deal was structured primarily to secure 1.3 gigawatts of grid-connected power capacity. The $9 billion price tag was not for computing equipment. It was for an existing grid connection that would have taken years to permit and build from scratch. Goldman Sachs Research estimates the grid itself may require approximately $720 billion in spending through 2030 to meet rising data center demand. The transmission bottleneck is not a side issue. It is the rate-limiting step. A decade of tight power markets and the investor opportunity For investors trying to position around AI growth, the electricity constraint is gradually reframing which companies matter. Michael Heinrich, CEO of 0G Labs, argues the scale of what is coming is still not fully priced into how the market is thinking about AI infrastructure. "Power is quietly becoming the hard ceiling on AI. The four largest hyperscalers spent over 500 billion dollars on capex in 2025, and AI is on track to double US data center electricity use by 2030. You cannot permit, finance, and build gigawatts of new generation as fast as model demand is growing, so the grid, not the GPU, is the bottleneck," Heinrich told TheStreet. He argues the policy response has been too narrowly focused on new centralized generation. The more valuable near-term opportunities are in the physical infrastructure that shortens the time between planning and operational power capacity. Where new capital is beginning to flow in energy infrastructure: * On-site natural gas generation: Around one-fifth of US data center projects under development are now building their own gas-fired power to bypass grid connection delays, according to IEA analysis. This is creating new supply chain demand for turbine manufacturers and fuel suppliers. * Small modular reactors: Microsoft, Google, and Amazon have all signed offtake agreements with nuclear developers. The IEA projects the first SMRs come online around 2030, partly to serve data center demand for reliable, always-on power. * Grid equipment: Transformer and cable shortages are now a standalone bottleneck. Delivery lead times for critical grid components have doubled over the past three years, and equipment manufacturers are running full order books years out. * Battery storage inside data centers: The IEA projects 20-25 gigawatts of battery storage could be installed inside data centers globally by 2030, potentially making them stabilizing assets to the broader grid rather than purely consumers of it. * Demand response markets: Grid operators are placing increasing value on industrial loads that can power down quickly during grid stress. As AI data centers require near-constant uptime, other large electricity consumers with flexible loads are becoming more valuable, not less, as grid pressure mounts. Fan, the founder of Cysic, sees the investment cycle lasting years, not quarters. He expects utilities to benefit from significant capital spending opportunities, but warns that grids will face higher congestion, more price volatility, and reliability pressure if investment lags behind demand. Getting the timing right matters as much as getting the direction right. "Expect a decade of tight power markets, rising baseload value, and a scramble for anything that shortens time to energized capacity. The opportunity is in the picks and shovels of power delivery," Heinrich added. How energy-intensive computing is already reorganizing around power The pressure on grid capacity is not only changing where new AI data centers get built. It is also reshaping operators across the broader computing industry who have long competed for large blocks of cheap electricity. AI hyperscalers are now paying premium prices for reliable, grid-connected capacity, changing the economics for every other large electricity consumer in the market. Many of the physical sites best positioned for AI workloads currently run other types of compute operations. Those operators are weighing the economics differently as the gap between what hyperscalers will pay and what traditional computing workloads generate continues to widen. Some are converting existing sites. Others are moving toward stranded energy sources, curtailed renewable generation, and behind-the-meter power projects that AI data centers cannot easily reach. Bitcoin mining is worth understanding in this context. At its peak, the global Bitcoin network consumed more electricity than many mid-sized countries, an estimated 120 to 150 terawatt-hours annually. Mining operators built large-scale power infrastructure specifically designed for high-density, continuous compute loads. They negotiated long-term contracts with utilities, developed expertise in managing enormous electricity demand, and in many cases secured grid connections that took years to establish. That physical infrastructure, built for one form of intensive computation, is now being evaluated by operators running a very different one. The economics are shifting fast. AI workloads pay considerably more per megawatt than proof-of-work mining at current prices, so grid-connected mining sites with existing utility contracts are attractive acquisition targets. Some operators are converting capacity directly. Others are holding their positions and leasing to AI tenants. The result is a quiet reorganization of who controls the grid-connected compute capacity that AI companies need most urgently. "The map is splitting. Grid-connected US sites convert to AI, and hash rate chases stranded power in Paraguay, Ethiopia, the Gulf," Videau added. The distributed, behind-the-meter opportunities that geographic dispersal creates are the same ones Heinrich identifies as a natural home for decentralized AI infrastructure. Spreading compute load across a wider range of power sources, rather than concentrating everything in a handful of large campuses, could reduce pressure on the existing grid while opening capacity in places the major hyperscalers have not yet reached. For investors, the energy and technology sectors are converging on the same conclusion. AI capacity is increasingly an infrastructure story as much as a technology one. The companies positioned to deliver reliable power, build grid connections, manufacture critical equipment, and upgrade transmission networks may define as much of AI's next chapter as the companies building the models themselves. The Arena Media Brands, LLC THESTREET is a registered trademark of TheStreet, Inc. This story was originally published July 6, 2026 at 9:33 AM.

Energy for AI - The Korea Times

Korea is in the midst of a stock market frenzy. Stock prices are influenced by a company's future value, and Korea's market is heavily weighted toward semiconductor firms whose share prices are shaped by global artificial intelligence (AI) investment. This is why investors in the Korean market are fixated on the future of AI. Their focus is on how much AI will reshape past industries and cultures, and whether it can generate returns commensurate with expectations. Recently, a new element has entered this mix: energy. Concerns have emerged that the energy essential for AI training and inference may not be supplied as smoothly as anticipated, potentially delaying AI adoption in certain regions. According to data released last August by global consulting firm McKinsey, worldwide data center capacity is projected to reach a cumulative 220 gigawatts by 2030, six times the 2020 level. This growth is driven primarily by the expansion of AI data centers. Relatedly, earlier this month, research firm Gartner projected this year's data center power consumption to rise 26 percent from last year to 565 terawatt-hours, and estimated next year's figure at 702 terawatt-hours, illustrating the explosive growth in power demand. Compared with Korea's 2025 power consumption of 625 terawatt-hours, the pace of this increase becomes tangible. By 2030, data center power consumption is estimated to exceed 1,200 terawatt-hours, surpassing not only Japan but even Russia based on 2025 figures. The energy industry inherently requires long lead times to build large-scale supply, while AI demands massive power quickly. This mismatch has made securing energy one of the top priorities for AI development companies. This phenomenon is already materializing in the market. Last March, U.S. tech outlet TechCrunch warned that up to half of currently announced data center construction could be delayed due to power supply issues, and observed that Big Tech companies are building their own power plants or signing separate power purchase agreements to overcome such crises. In addition, AI analytics firm SynMax reported that, as of last April, 40 percent of data centers targeted for completion in the United States this year risk being delayed, while 60 percent of those slated for completion by next year have yet to break ground. Data centers have various options for securing electricity, but no readily available alternative is truly optimal. Drawing from the grid is difficult. So many new users have joined that grid connection waiting times have doubled, and key expansion equipment such as transformers is in short supply. Turning to relatively fast and stable gas-fired generation is also problematic because generator orders are backlogged, requiring waits of more than five years. Small modular reactors, which supply carbon-free power, are widely viewed as unlikely to see meaningful deployment before at least 2030, given the need to prove their commercial viability. Meanwhile, fuel cells, which have recently emerged as an alternative, have their own drawbacks: For now they must run on fossil fuels, and both system costs and clean hydrogen prices remain burdensome. Finally, the most environmentally friendly options -- solar and wind -- are intermittent, generating power only when the sun shines or the wind blows, which does not align with AI's need for round-the-clock supply. Even with energy storage systems, achieving a full 24-hour supply is realistically difficult, so support from other energy sources is inevitably required. An additional and important point when choosing among these options is the burden associated with fossil fuel use. AI development companies have already established and disclosed carbon-neutral or carbon-reduction targets. Because these are voluntary, softening them is possible, but persuading key stakeholders -- such as future-generation consumers and long-term investors -- of the reasons for change is far from easy. Their past carbon emissions are minimal compared with other industries, and they would want to avoid being unfairly cast as principal culprits of today's climate crisis -- caused by past emissions -- simply because their future emissions are expected to be large. Accordingly, if the energy essential to the AI business is supplied by fossil fuels on a large rather than small scale, this could become a significant burden on a company's sustainability. Taken together, there is no single best alternative to satisfy the explosively growing large-scale power demand in the short term; the only path is to choose second-best alternatives and supplement their weaknesses. For example, one approach is to meet primary power demand through renewable energy and energy storage, while covering only auxiliary power from low-efficiency gas generation (such as single-cycle turbines) or the grid. Another is to meet primary demand with fuel cells, gradually blending clean hydrogen alongside natural gas. The U.S. Energy Information Administration's observation that 93 percent of planned new generation capacity in 2026 will be solar, wind and energy storage, together with the forecast of energy market analytics firm Rystad Energy that roughly 10 gigawatts of fuel cells will be installed at U.S. data centers over the next five years, suggests that this approach is already being adopted. Clearly, energy has come to play a critical role in the future of AI, one of the dominant themes of the stock market. Supplying energy in a stable and quick manner to meet explosive short-term demand growth is important, but so too is the recognition that, for AI to remain sustainable over the long term, the energy powering it must itself be sustainable. Kim Sung-woo, head of Environment & Energy Research Institute at Kim & Chang, is a board member of the Korea Institute of Energy Technology Evaluation and Planning.