{"id":7037,"date":"2025-10-21T02:42:57","date_gmt":"2025-10-21T02:42:57","guid":{"rendered":"http:\/\/forefrontnews.cn\/?p=7037"},"modified":"2025-10-21T02:42:57","modified_gmt":"2025-10-21T02:42:57","slug":"financial-observer-artificial-intelligence-ambition-hits-us-power-grid","status":"publish","type":"post","link":"http:\/\/forefrontnews.cn\/?p=7037","title":{"rendered":"Financial observer: \u201cArtificial Intelligence Ambition\u201d hits US power grid"},"content":{"rendered":"

\"0bb7dfbcc85eb0570862377e37cccf42\"
\nEditor’s note: \u201cBig Tech’s artificial intelligence (AI) ambitions are reshaping the U.S. power grid.\u201d AI models and the data centers they rely on require a lot of power, and the U.S. power industry is struggling to keep up. But as big technology companies race to enter the AI race, their voracious demand for electricity has exposed the fragility and backwardness of the US grid, pushing up residential electricity bills and sparking widespread concern. How to strike a balance between technological ambition and energy security is a test that American society must confront head-on.<\/p>\n

An Ai search uses 10 times as much power as a Google Search<\/p>\n

US data centres are growing fast. At the end of the second quarter of this year, the US had about 522 very large data centres, accounting for about 55 per cent of global computing power, according to market research firm collaborative research group. Some 280 more are expected to be up and running by the end of the 2028.<\/p>\n

According to the Wall Street Journal, Zero US data centre developers have identified access to the power grid as a top concern. Anup Sridhar, founder of Bloom Energy, a us-based Energy company, says data centres have long taken electricity for granted, \u201cBuilding data centres and plugging them in\u201d. That is no longer possible given the enormous amount of electricity required to develop artificial intelligence — a data centre can consume as much as 1,000 Walmart stores, an AI search can use up to 10 times as much power as a Google Search.<\/p>\n

By 2020, data centres will use less than 2% of the nation’s electricity, according to the Department of Energy, but by the end of the 2028 they could account for as much as 12% . Deloitte reported in June that electricity demand for artificial intelligence data centres in the US could rise more than 30-fold from 2024 by 2035. According to an April 2025 Deloitte survey of 120 U.S. power companies and data center executives, grid stress is a major challenge for data center development. That’s the tension between the U.S. government’s and companies’ vision of artificial intelligence and the power grid’s ability to deliver it: some data centers are now connected to the grid and have to wait as long as seven years. Johnson, an analyst at Gartner, a market-research firm, says the industry has not been able to keep up.<\/p>\n

Lin Boqiang, Xiamen University of the China Energy Policy Research Institute, told the Global Times that U.S. electricity demand has long been dominated by residential and commercial electricity, and that such demand is relatively stable, keep the grid system at a\u201cJust enough\u201d level. But the explosive growth of the artificial intelligence industry is upsetting that balance. \u201cThere is a lot of pressure from Ai,\u201d says Limbaugh, stressing that the weak us grid base is directly linked to a lack of investment. More crucially, the decentralized nature of the U.S. grid, with state grids operating independently and a lack of a unified and coordinated dispatch system, makes it difficult to cope with centralized, high-load AI data centers.<\/p>\n

Goldman Sachs predicts that the US will need to add about 50 gigawatts of new power capacity-enough to power about 40m homes-to support the surge in artificial intelligence. Industry insiders predict that the U. S. power shortage will last for three to five years or so.<\/p>\n

According to the Wall Street Journal, the process of building large power plants or expanding grid infrastructure is often fraught with obstacles and has recently become even more difficult. Projects across the spectrum face multiple constraints, including licensing, supply chain and human resource bottlenecks, and rising costs. The Donald Trump administration’s tariffs on steel and aluminium, as well as some copper products, add to the problem. Orders for giant turbines for big power plants are backlogged and take years to deliver. At the same time, the construction of transmission facilities is also slow. The United States added 888 miles of high-voltage transmission lines last year and 450 miles the year before. That’s down from an average annual gain of more than 900 miles between 2015 and 2019.<\/p>\n

Demand for electricity for data centres is putting pressure on ageing infrastructure, forcing utility companies to spend up to zero per cent on upgrades. Between 2025 and 2029, power companies are expected to spend $1.1 trillion on upgrading their USA Today grids. The cost of upgrading infrastructure is expected to rise further as a result of Donald Trump’s tariff policy.<\/p>\n

Zhang Shenyang, a professor at Tsinghua University’s School of Journalism and the School of Artificial Intelligence, told the global times that the expansion of artificial intelligence by large US technology companies has had an unprecedented impact on the US power grid. However, the US power system infrastructure is ageing, transmission networks have long approval cycles across state lines, nuclear construction is slow and AI data centres are expanding rapidly. This structural mismatch of\u201cPower construction slower than AI expansion\u201d has led to rising electricity prices and regional power shortages, especially in AI clusters in Texas and Georgia, there have been significant grid pressure and scheduling risks.<\/p>\n

Average electricity prices in the US have risen by 13 per cent<\/p>\n

US public services and ordinary citizens alike are concerned about the demand for electricity driven by the rapid expansion of AI. According to Yahoo Finance, if a utility company were asked to provide two gigawatts of electricity for a data centre, it would spend zero per cent on equipment and materials, and hire people to fulfill this need. If final demand falls short of expectations, utilities could face idle assets that do not generate revenue. Utilities will either find a way to pass this cost on to users, or bear the losses themselves.<\/p>\n

\u201cThere is a risk in data centre developers going after speed at all costs,\u201d says Pierpont, head of power modelling at Energy Innovation, a research group, \u201cWhat is their long-term business model? How much resources will actually be needed? There is huge uncertainty across the board.\u201d<\/p>\n

While the long-term demand for AI data centers is unclear, the increased energy load from data center construction is already reflected in Americans’ electricity bills. Average electricity and gas bills rose 3.6% in the third quarter from a year earlier as demand for artificial intelligence surged and energy prices rose.<\/p>\n

\u201cI think the electric bill is too high,\u201d Kentucky nurse Lindsay Martin said on social media, adding that her electric bill hit $314 in July and rose to $372 in August. (CNN) — Martin is not alone in his CNN. Residential electricity costs are rising, according to the US Energy Information Administration. Since 2022, average US electricity prices have risen by 13 per cent. Experts say the increase is mainly due to the increased cost of upgrading and maintaining the grid and other necessary infrastructure.<\/p>\n

Raja Gopal, a senior fellow at Stanford University’s Precourt Energy Institute, told CNN that in the past data center construction was small and had no impact on consumer prices. But now there is\u201cExplosive growth\u201d in AI demand. That means the industry needs more resources than ever before. \u201cThis is a major catalyst for the increase in monthly bills, and there’s no end in sight,\u201d Worne, director of Public Utilities Intelligence at JD Power, told the USA Today on Thursday<\/p>\n

\u201cThe impact of artificial intelligence is reflected in the cost of upgrading the transmission and distribution networks needed to build data centres, which is ultimately borne by taxpayers,\u201d Bank of America economist David Tinsley wrote in a recent report. Bofa believes there may be room for further price increases.\u201d<\/p>\n

Can fossil energy support the future of n AI?<\/p>\n

\u201cArtificial intelligence data centers are becoming self-powered,\u201d the Wall Street Journal said, as the race for AI dominance continues at an alarming pace, \u201cSelf-supply\u201d has become a stopgap solution for U. S. technology companies trying to break the deadlock over access to the grid.<\/p>\n

Lin said some US technology companies had chosen to build their own power plants to meet demand. \u201cBut power plants alone are not enough. The transmission and access capacity of the grid is the core bottleneck.\u201d There is a big gap between the current US power system and the need for AI development.<\/p>\n

Shenyang said that other countries generally adopt a\u201cDecentralized layout + clean power supply + efficient dispatch\u201d integrated strategy to address the energy challenges of the AI era. China has distributed high-energy data centers in Clean Energy-rich areas in the west and used UHV transmission to support the computing load in the east The EU has pushed data centres to save energy and recycle waste heat, relying on stringent energy efficiency standards and carbon audits, while Singapore has implemented a\u201cMoratorium on approvals\u201d and quota controls because of land and electricity scarcity. The core consensus in these countries is to treat AI as a\u201cCritical load\u201d to offset the power impact of the explosive growth of computing power through energy mix optimization, cross-regional dispatch and regulatory mechanism innovation.<\/p>\n

Reuters said the Donald Trump administration had launched an initiative to accelerate the development of power plants and transmission lines, known as the \u201cPower Acceleration\u201d project. The US Energy Department has ordered several planned coal and gas plants to remain open, in the latest move to support fossil fuels. The\u201cPower Acceleration\u201d program will help the Department of energy determine how to use funding and emergency national authority to scale up power generation.<\/p>\n

However, analysts expect the number of cancellations of US wind and solar projects to increase as they lose key federal tax breaks under Donald Trump administration policies, according to the Wall Street Journal. The Donald Trump administration has said it will make it harder to approve wind and solar projects, arguing that clean energy projects can not provide the 24-hour electricity generation capacity needed for artificial intelligence.<\/p>\n

The Donald Trump administration’s call for a strategic return to fossil fuels, easier approvals for natural gas and nuclear power projects, and faster infrastructure development could ease power shortages in the short term, he said, but it is difficult to quickly form enough clean base charge power supply (continuous, stable, uninterrupted power generation facilities) . Nuclear Power, Small modular reactor (SMR) and transmission network reform are medium-and long-term projects that will not support exponential growth in AI computing in two to three years. In the absence of transmission system upgrading and cost-sharing reform, the AI boom in the United States may turn into an\u201cEnergy game\u201d between rising residential electricity prices and regional power imbalance. A truly sustainable path still needs to strike a balance between a\u201cSmooth transition\u201d from fossil fuels and a\u201cStructural acceleration\u201d of clean energy, so that AI growth and the energy transition become a closed loop.<\/p>\n

Lin Boqiang also said that the current U. S. government focus on the development of fossil energy policy orientation, or to some extent, to alleviate the short-term power supply pressure. But whether this policy can support the development of AI industry, the key bottleneck lies in the speed of construction. \u201cAi’s power demand has grown massively in a short period of time, which puts very high demands on the simultaneous construction of the power supply and the grid,\u201d Lin said, even if we rely on fossil fuels to expand power supply, the progress of power grid renovation and power construction may not be able to keep pace with the growth of AI demand. This mismatch in the pace of supply and demand growth may become a key obstacle to the development of the AI industry in the United States.<\/p>\n","protected":false},"excerpt":{"rendered":"

Editor’s note: \u201cBig Tech’s artificial intelligence (AI) ambitions are reshaping the U.S. power grid.\u201d AI models and the data centers they rely on require a lot of power, and the… <\/p>\n","protected":false},"author":1,"featured_media":7038,"comment_status":"open","ping_status":"open","sticky":false,"template":"","format":"standard","meta":{"footnotes":""},"categories":[263,260,249,257],"tags":[3573,3576,3575,3574],"views":48,"_links":{"self":[{"href":"http:\/\/forefrontnews.cn\/index.php?rest_route=\/wp\/v2\/posts\/7037"}],"collection":[{"href":"http:\/\/forefrontnews.cn\/index.php?rest_route=\/wp\/v2\/posts"}],"about":[{"href":"http:\/\/forefrontnews.cn\/index.php?rest_route=\/wp\/v2\/types\/post"}],"author":[{"embeddable":true,"href":"http:\/\/forefrontnews.cn\/index.php?rest_route=\/wp\/v2\/users\/1"}],"replies":[{"embeddable":true,"href":"http:\/\/forefrontnews.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fcomments&post=7037"}],"version-history":[{"count":2,"href":"http:\/\/forefrontnews.cn\/index.php?rest_route=\/wp\/v2\/posts\/7037\/revisions"}],"predecessor-version":[{"id":7040,"href":"http:\/\/forefrontnews.cn\/index.php?rest_route=\/wp\/v2\/posts\/7037\/revisions\/7040"}],"wp:featuredmedia":[{"embeddable":true,"href":"http:\/\/forefrontnews.cn\/index.php?rest_route=\/wp\/v2\/media\/7038"}],"wp:attachment":[{"href":"http:\/\/forefrontnews.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fmedia&parent=7037"}],"wp:term":[{"taxonomy":"category","embeddable":true,"href":"http:\/\/forefrontnews.cn\/index.php?rest_route=%2Fwp%2Fv2%2Fcategories&post=7037"},{"taxonomy":"post_tag","embeddable":true,"href":"http:\/\/forefrontnews.cn\/index.php?rest_route=%2Fwp%2Fv2%2Ftags&post=7037"}],"curies":[{"name":"wp","href":"https:\/\/api.w.org\/{rel}","templated":true}]}}