미국 AI 데이터센터 증축 절반 지연·취소

Copy link Facebook X Whatsapp Reddit Pinterest Flipboard Email Share this article 10 Join the conversation Follow us Add us as a preferred source on Google Newsletter Stay On the Cutting Edge: Get the Tom's Hardware Newsletter Get Tom's Hardware's best news and in-depth reviews, straight to your inbox. Contact me with news and offers from other Future brands Receive email from us on behalf of our trusted partners or sponsors By submitting your information you agree to the Terms & Conditions and Privacy Policy and are aged 16 or over. You are now subscribed Your newsletter sign-up was successful An account already exists for this email address, please log in. Subscribe to our newsletter The trade-war between the U.S. and China has forced server makers out of the People's Republic, greatly reducing reliance of American companies on producers from Tianxia. However, China remains the world's largest producer of electrical equipment that is required to build power infrastructure inside and outside of AI data centers. To that end, shortages of power delivery equipment, including devices from China and other countries, are slowing project timelines, Bloomberg reports. Despite the unprecedented level of investment in AI infrastructure — Alphabet, Amazon , Meta, and Microsoft are expected to spend more than $650 billion in 2026 to expand AI capacity — close to half of the planned U.S. data center builds this year are projected to be delayed or canceled, according to Bloomberg . One major reason behind these setbacks is the availability of key electrical components — such as transformers, switchgear, and batteries — that are used both at data center sites and outside of them, as AI companies must expand grid infrastructure to supply enough power to their data centers. Meanwhile, grid infrastructure is also stressed by electric vehicles and electrified heating systems. Approximately 12 gigawatts (12 GW) of data center capacity is expected to come online in the U.S. in 2026, according to data by market intelligence firm Sightline Climate cited by Bloomberg . Yet only about one-third of that capacity is currently under active construction because of various constraints. Article continues below Electrical infrastructure represents less than 10% of total data center cost, but it is as vital as compute hardware. A delay in any single element of the power chain can halt the entire project, which makes transformers, switchgear, and similar devices critical items despite their relatively small share of CapEx. Due to high demand, lead times for high-power transformers have expanded dramatically in the U.S.: delivery typically took 24 to 30 months before 2020, but waiting periods can stretch to as long as five years today, according to Sightline Climate cited by Bloomberg . For AI data centers, this is a catastrophe as their deployment cycles are under 18 months. To address shortages, companies are turning to global markets. As a result, Canada, Mexico, and South Korea became the biggest suppliers of high-power transformers for AI data centers to AI data centers. At the same time, imports of high-power transformers from China surged from fewer than 1,500 units in 2022 to more than 8,000 units in 2025 through October, according to Wood Mackenzie data cited by Bloomberg . The volatility of exports from China does not end with transformers, as the PRC accounts for over 40% of U.S. battery imports, while its share in certain transformer and switchgear categories remains near 30%, according to Bloomberg . Stay On the Cutting Edge: Get the Tom's Hardware Newsletter Get Tom's Hardware's best news and in-depth reviews, straight to your inbox. Contact me with news and offers from other Future brands Receive email from us on behalf of our trusted partners or sponsors Without resolving constraints in transformers, switchgear, and batteries, even trillions of dollars in AI investment may not translate into actual AI capacity, as deployments will depend on power infrastructure availability, not capital or compute hardware constraints. Despite a decade of reshoring initiatives, U.S. manufacturing capacity for electrical equipment remains insufficient, which means that AI companies continue to rely on imports even amid tariffs and national security concerns. Meanwhile, tensions between China and the U.S. threaten to further disrupt supply chains, which will raise costs and could delay deployments of advanced AI data centers. Follow Tom's Hardware on Google News , or add us as a preferred source , to get our latest news, analysis, & reviews in your feeds. TOPICS See all comments (10) Anton Shilov Contributing Writer Anton Shilov is a contributing writer at Tom’s Hardware. Over the past couple of decades, he has covered everything from CPUs and GPUs to supercomputers and from modern process technologies and latest fab tools to high-tech industry trends. 10 Comments Comment from the forums Globalisation... You can't have your cake and eat it, too. Reply Power is 50% of the issue, the other are the people, no one wants a Data Center in their backyard. I feel like I live in the center of this, 8 planned Datacenters in the last year, 2 have already been cancelled because of protestors and lack of public support, 1 is being built, and 5 more who knows? Reply Rand0m_Guy said: Power is 50% of the issue, the other are the people, no one wants a Data Center in their backyard. I feel like I live in the center of this, 8 planned Datacenters in the last year, 2 have already been cancelled because of protestors and lack of public support, 1 is being built, and 5 more who knows? Guess the AI datacenter satellites are going to be a winner. Reply Even if they get power they will have nothing to put in them. The equipment manufacture like memory seem to feel there is no need to upgrade capacity. If all these data centers went online they still could sell all the memory they can make. So even though the AI companies are making huge bets it seems the more normal business feel the AI companies are wrong and will not actually need these new data centers. They do not want to risk expanding capacity and then the AI bubble pop. Reply Faiakes said: Globalisation... You can't have your cake and eat it, too. I mean it isnt just that. We didnt keep strong cores in multiple countries for most of this stuff and so its either Taiwan or China Reply Also doesn't hurt that China has been implementing a massive power grid upgrade for the last ten years or so. They've built the capacity to build and maintain that, I don't think they planned out having to provide switch gear and mains equipment for hundreds of additional data centers globally. Reply You know it's so over when parts for critical infrastructure are outsourced, and then your government forgets that it's no longer self-sufficient at maintaining itself. Reply Notton said: You know it's so over when parts for critical infrastructure are outsourced, and then your government forgets that it's no longer self-sufficient at maintaining itself. Exactly. In my opinion this is a huge reason why governments need to be balanced, and have checks and balances. Our government has been gutted and deregulated, and corporations have not been held in check. As a result we get short term greed driving all of the decisions. Corporations get bigger, they buy more control, politicians give them whatever they want - more deregulation, support for outsourcing jobs, changing the laws to allow bigger mergers, etc. For example, The Telecommunications Act of 1996 not only allowed our media to be controlled by 5 conglomerates, it allowed two of those behemoths to merge (Disney acquiring Fox/NewsCorp). A merger like that should never have been allowed to happen. But Clinton’s signing of the Telecom Act made it legal. And in exchange the politicians get bribes, kickbacks, continued support and of course big money jobs in the private sector down the road.