中期选举后,数据中心之争或将升级

11月3日的中期选举结果让民主党全面掌控国会,为可能是这十年间最为分裂的一届美国政府埋下伏笔——白宫仍由共和党把持。
周二的结果表明,围绕数据中心——这些容纳驱动 AI 的计算机的大型设施——的不满情绪正成为一股政治力量。随着民主党如今掌控国会,这可能导致对数据中心选址以及当地社区话语权的新限制。
在多个州,数据中心都成了投票议题。NBC 的一项出口民调显示出选民对 AI 的担忧程度:超过半数受访者认为 AI 弊大于利,而近 80% 的人反对在自己所在社区建设数据中心。与此同时,近 60% 的人表示,数据中心是他们投票时的关键或重要因素。种族正义组织 Color of Change 的总裁兼 CEO Nadine Smith 称周二的选举是一场关于“这套经济体系为谁而建”的公投。
“国家安全很重要,”Smith 说,她指的是数据中心支持者的一种常见论点,即这一基础设施对美国在战争中以及在与中国的竞争中保持领先至关重要。“南孟菲斯那些呼吸着为 AI 超算供电的燃气轮机废气的家庭,也同样重要。”(她指的是 xAI 位于田纳西州孟菲斯的数据中心,该中心使用燃气轮机供电。)
多个州的选民对 AI 建设扩张投下反对票,尽管在这个中期选举周期,支持 AI 的超级政治行动委员会投入数百万美元,希望能影响他们的选择。反数据中心的风潮在俄亥俄州等地尤为明显——OpenAI 曾在那里物色了一个 5000 亿美元的数据中心项目,由 NVIDIA 提供逾 1000 亿美元融资支持。事实上,Meta 曾计划建设一座 70 万平方英尺设施的俄亥俄州县,刚刚通过了数据中心禁令,全州各地的社区也以压倒性多数投票反对数据中心的建设措施。
摩根士丹利的分析师预测,当选州长的民主党人 Amy Acton 也将成为俄亥俄州数据中心建设的一大风险。在其他地方,得克萨斯、缅因、密歇根、内华达、佐治亚和南卡罗来纳等州的候选人都在表明反对数据中心快速建设的立场后赢得席位。“很明显,金钱买不来结果,”科技与政策研究员 Anshi Bhatt 说,并补充称,在现代政治中,塑造一个“地方反派”是积累竞选势头的一种方式。
接下来可能的举措包括出台更多建设禁令,或要求数据中心自行承担电网升级费用,而不是把这些成本转嫁给附近社区。事实上,高盛预测,由于民主党取得进展,佐治亚和得克萨斯等州的数据中心建设将出现延迟。
民主党控制国会也可能提高今年早些时候参议员伯尼·桑德斯和众议员亚历山德里娅·奥卡西奥-科尔特斯提出的那类法案的通过几率,不过任何此类法案都可能面临否决。该法案提议在全国范围内暂停新建数据中心,直到国会增加新的防护机制,并确保数据中心建设不会推高电价、损害环境,或在未获社区批准的情况下推进。
还有类似新墨西哥州参议员 Martin Heinrich 提出的《GRID Savings Act》(电网节约法案)等法案,这些法案将阻止在大型 AI 数据中心建设过程中,把电网基础设施成本转嫁给社区。
环境也是一个主要担忧,尤其是在 xAI 位于孟菲斯的数据中心,以及遍布弗吉尼亚州的数据中心——弗吉尼亚被称为数据中心的世界之都——这些地方使用燃气轮机供电。非营利研究机构 Data & Society 的项目总监 Maia Woluchem 认为,能源监管、土地权利及其他环境政策的改变可能影响数据中心的建设,尤其是在国会民主党人进一步调查特朗普政府对某些能源保护措施的撤销之际——据《纽约时报》报道,他们可能会这么做。
“不管联邦层面采取什么行动,考虑到自下而上不断积聚的势头,我可以想象我们会在州和地方层面看到更多动向,”Woluchem 继续说道,并以当选州长 Gina Hinojosa 为例,后者支持在全州范围内暂停数据中心的建设。Hinojosa 和州长 Greg Abbott 都呼吁实施某些限制,但 Hinojosa 的竞选活动大部分以反对数据中心为核心,这在得克萨斯州是一个引人注目的选择——过去十年,该州的电网已被证明相当脆弱。
但困难的部分还在后面。例如,Smith 现在预计会出现一场争斗,甚至会有更多行政命令,以绕过国会目前很可能阻止的事情。“即便在民主党内部,全面冻结也是有争议的,”她说,指的是部分民主党人担心暂停建设可能拖慢 AI 进展并损害国家安全。
她还预计,AI 游说人士将开始给议员打电话。“民主党人是打着让大型科技公司付出代价的旗号竞选的,”Smith 说。“我们的工作就是确保在镜头离开之后,他们还记得这一点。”
这种反弹的根源比数据中心建筑本身更深,而这些建筑并不新鲜。一位私募股权高管最近告诉我,人们未必是讨厌数据中心,而是讨厌 AI,数据中心已经成为人们对这项技术的恐惧与焦虑的替身。“看起来,调查结果至少指向人们对科技权力的真实不满,”Woluchem 说,并指出这同样是一个两党共同关注的问题。
过去一个月里,人们告诉我,他们担心数据中心对环境的影响,也担心自己在数据中心是否兴建这件事上没有发言权。
例如,Amazon 在加利福尼亚州 Gilroy 建了一座数据中心,许多居民并不知情,其做法是满足了当地一项已有 45 年历史的区划法规。含糊或过时的区划法规,再加上要求地方官员签署保密协议(NDA)、不得讨论数据中心项目的规定,只会加剧人们对这轮建设潮的愤怒。数据中心还获得了数十亿美元的税收减免,尽管许多科技公司大肆宣传其好处,比如就业和社区项目,但对很多人来说,这远远不够。
人们希望知道自己对社区事务仍有发言权。许多人还担心,AI 建设潮主要会让少数富人获益,而他们的社区却要承受代价,生活成本也会上升。
“如果 AI 基础设施确实是国家优先事项,那么建设它的企业就有能力为自己的电力买单、达到清洁空气标准、披露他们使用了什么,并在邻居们的同意下进行建设,”Smith 说。“家庭永远不应该通过水电费来补贴史上最富有的企业。”
当您通过我们文章中的链接购买时,我们可能会赚取少量佣金。这不会影响我们的编辑独立性。
高级记者,风险投资
Dominic-Madori Davis 是 TechCrunch 的高级风险投资与初创企业记者。她常驻纽约市。
您可以通过发送邮件至 [email protected],或通过 Signal 加密消息 +1 646 831-7565 联系 Dominic,以核实其外联信息。
正在加载下一篇文章
译文由英文原文机器翻译,术语按行业惯例统一,可在文末展开原文核对。
查看英文原文(After the midterms, a bigger clash over data centers could be coming)
After the midterms, a bigger clash over data centers could be coming | TechCrunch
After the midterms, a bigger clash over data centers could be coming
5:02 PM PST · November 4, 2026
The November 3 midterm results gave Democrats full control of Congress, setting the stage for what could be one of the most divided U.S. governments this decade, with Republicans in the White House.
Tuesday’s results suggest that anger over data centers, the large facilities that house the computers powering AI, is becoming a political force. With Democrats now in control of Congress, that could lead to new limits on where data centers can be built and how much say local communities have.
Data centers became a ballot-box issue in various states. An NBC exit poll showed how concerned voters are about AI: More than half of respondents believe that AI will do more harm than good, while almost 80% are against the building of data centers in their neighborhoods. Almost 60%, meanwhile, said data centers were a crucial or major factor in their vote. Nadine Smith, president and CEO of the racial justice organization Color of Change, called Tuesday’s election a referendum on “who this economy is built for.”
“National security matters,” Smith said, referring to a common argument from data center supporters that the infrastructure is essential for the U.S. to stay ahead in war and in its rivalry with China. “And so do the families in South Memphis breathing exhaust from the gas turbines powering an AI supercomputer.” (She was referring to xAI’s data center in Memphis, Tennessee, which uses gas turbines for power.)
Voters in several states pushed back against the AI buildout, even as pro-AI super PACs spent millions hoping to sway them this midterm. The anti-data center backlash was especially felt in states like Ohio, where OpenAI has eyed a $500 billion data center, backed by more than $100 billion in financing from Nvidia. In fact, the Ohio county where Meta had planned a 700,000-square-foot facility just passed a ban on data centers, and local communities across the state overwhelmingly voted against data center buildout measures.
Analysts at Morgan Stanley predict that Democratic governor-elect Amy Acton will also pose a major risk to Ohio’s data center buildout. Elsewhere, candidates in Texas, Maine, Michigan, Nevada, Georgia, and South Carolina all won seats after taking a stance against the rapid buildout of data centers. “It is clear that money does not buy outcomes,” said Anshi Bhatt, a tech and policy researcher, adding that in modern politics, creating a “local villain” is one way to build campaign momentum.
Likely next steps include more construction moratoriums or requiring data centers to pay for their own power grid upgrades instead of passing those costs on to nearby communities. Goldman Sachs, in fact, predicts data center construction delays in states like Georgia and Texas because of Democratic gains.
Democratic control of Congress could also improve the odds for bills like the one Sen. Bernie Sanders and Rep. Alexandria Ocasio-Cortez introduced earlier this year, though any such bill would also face a possible veto. It proposes a nationwide moratorium on new data centers until Congress adds new guardrails and ensures that the buildout does not raise electricity costs, harm the environment, or proceed without community approval.
There are also bills like the GRID Savings Act, proposed by New Mexico Sen. Martin Heinrich, which would prevent grid infrastructure costs from being passed on to communities during large AI data center buildouts.
The environment is also a major concern, especially at xAI’s data center in Memphis and at the data centers throughout Virginia, known as the world capital of data centers, where gas turbines are used for power. Maia Woluchem, program director at Data & Society, a nonprofit research group, believes that changes to energy regulations, land rights, and other environmental policies could affect the data center buildout, especially as congressional Democrats look further into the Trump administration’s rollback of certain energy protections, as the New York Times reported they might.
“Regardless of federal action, I could imagine that we will see additional movement at the state and local level, given the momentum that’s been building from the ground up,” Woluchem continued, citing governor-elect Gina Hinojosa, who supports a statewide moratorium on data centers, as an example. Both Hinojosa and Gov. Greg Abbott called for some limits, but Hinojosa centered much of her campaign on opposing data centers, a notable choice in Texas, whose electrical grid has proved vulnerable over the past decade.
But the hard part is still to come. Smith, for example, is now expecting a fight, or even more executive orders, to bypass what Congress is now likely to block. “Even among Democrats, a blanket freeze is contested,” she said, referring to some Democrats’ worry that a moratorium could slow AI progress and hurt national security.
She also expects AI lobbyists to start calling lawmakers. “Democrats ran on making Big Tech pay,” Smith said. “Our job is to make sure they remember it once the cameras leave.”
The backlash runs deeper than the buildings themselves, which aren’t new. One private equity executive told me recently that people don’t necessarily hate data centers, but rather AI, and that data centers have become a stand-in for people’s fears and anxieties about the technology. “It does look like the results are pointing to a real frustration with tech power at minimum,” Woluchem said, noting that this is also a bipartisan issue.
For the past month, people have told me they are worried about data centers’ effect on the environment and about their lack of say in whether they’re built.
Amazon, for example, built a data center in Gilroy, California, without many residents knowing, by satisfying a 45-year-old zoning law. Vague or outdated zoning laws, along with requirements that local officials sign nondisclosure agreements (NDAs) barring them from discussing data center projects, have only fueled people’s outrage about the buildout. Data centers have also received billions in tax breaks, and although many tech companies have touted the benefits, such as jobs and community programs, for many people, that simply wasn’t enough.
People want to know they still have a say in what happens in their communities. Many also fear that AI buildouts will mainly enrich a few wealthy people while their communities suffer and living costs rise.
“If AI infrastructure is truly a national priority, then the companies building it can afford to pay for their own power, meet clean air standards, disclose what they use, and build with the consent of the people who live next door,” Smith said. “Families should never be subsidizing the richest corporations in history through their utility bills.”
When you purchase through links in our articles, we may earn a small commission. This doesn’t affect our editorial independence.

Senior Reporter, Venture
Dominic-Madori Davis is a senior venture capital and startup reporter at TechCrunch. She is based in New York City.
You can contact or verify outreach from Dominic by emailing [email protected] or via encrypted message at +1 646 831-7565 on Signal.
Loading the next article
Error loading the next article
原文为英文,正文由自动抓取并保留原文。
原文发布于 TechCrunch AI(After the midterms, a bigger clash over data centers could be coming),标题、摘要与正文由机器翻译,术语按行业惯例统一,未经人工逐句校订。