America got the anti-data-center wave it voted for. Here come the economic consequences

A stunning upset in the Texas governor race and sweeping Democratic victories across the country threaten to curtail the biggest-ever spending boom in US history.

That could disrupt a key growth engine for America’s economy .

Come January, states controlled by Democrats who support expanding or extending AI construction moratoriums will account for 45% of America’s planned data center capacity, according to Goldman Sachs. A quarter of America’s planned AI infrastructure buildout is set to take place in Texas alone.

That doesn’t mean half of America’s planned data centers are on hold. Demand for AI is off the charts. Donald Trump is still president, and he’s already waving around his veto hat. And Democratic governors couldn’t stop every data center from being constructed even if they wanted to.

But Tuesday’s election nevertheless represents a paradigm shift that could alter the pace and path of America’s nonstop AI spending spree – an investment boom that has, according to ING economists, accounted for as much as a quarter of total US economic growth this year.

Those economic benefits are already real, supporting thousands of jobs. And AI has the potential to revolutionize the way we work.

But an AI data center slowdown could come with important silver linings. It could rein in unfettered growth that has contributed to rebounding inflation and led to widespread public antipathy toward the technology and the data centers that house it. It could also curtail some questionable business models that have raised serious skepticism on Wall Street.

Tapping the brakes could ultimately put AI – and the US economy – on a more sustainable path.

Jay Janner/The Austin American-Statesman/Getty Images

AI infrastructure spending is expected to approach $1 trillion a year, every year, for the foreseeable future, according to Columbia Business School economist Stijn Van Nieuwerburgh. As a percentage of America’s gross domestic product, that would outpace every other investment boom in modern US history – including the construction of the railroads.

The outcome of Tuesday’s elections could slow that spending, with vast consequences for the thousands of jobs and massive amounts of economic development that AI supports.

Democrats gained gubernatorial seats in Nevada, Georgia, Ohio and Texas, where candidates campaigned on tightening data center regulations. Proposals from the governors-elect include ending tax breaks, tightening climate and electricity restrictions and allowing local governments to impose their own moratoriums on data center construction.

Eli Hartman/Bloomberg/Getty Images

But Texas is the big kahuna. Wells Fargo estimates that for every 10% reduction in planned Texas data center capacity, Big Tech companies will reduce their AI infrastructure spending by a stunning 1.6% (a lot of money when you’re measuring spending in the trillions of dollars).

State and local politicians have a lot of jurisdiction and sway over that spending: enough to change whether some projects get financed, said Mona Dajani, co-chair of Cooley’s infrastructure, energy and real estate practice.

Hundreds of data center developments that are still awaiting permits, utility service or tax incentives could be particularly exposed, Dajani noted. But even active construction projects could be threatened. Texas law grandfathers in some projects’ permits, protecting them from future changes to local permitting requirements – but that doesn’t resolve state permitting changes or electrical service guarantees that Governor-elect Gina Hinojosa could vacate.

The wait for approvals could add significant costs for AI companies, Dajani said: Interest, equipment commitments and utility expenses will all continue to accrue, even while a project is on ice. At some point, existing projects could stop being financially viable.

If that happens, AI companies – who have already strained to build out data centers fast enough to meet their anticipated demand – may not be able to meet their explosive spending targets.

“This is a rare case where state and local leaders have influence of geopolitical-level import,” said Harry Krejsa, director of studies at the Carnegie Mellon Institute for Strategy & Technology.

Data centers aren’t exclusively a Democratic issue: Americans have displayed bipartisan consensus for putting guardrails around data center buildouts.

Texas Republican Governor Greg Abbott had already imposed a data center moratorium that Hinojosa campaigned on extending until lawmakers establish stronger AI regulations.

But Trump had railed against any attempts to stymie AI’s growth and spending, citing economic and national security – particularly in the high-stakes race against China. Many Republican politicians who may have otherwise supported AI regulations, including frameworks for data center construction, remained silent.

Still, there’s evidence that the AI buildout could be on an unsustainable path.

The United States had 5,427 data centers at the end of last year, according to Stanford University’s AI Index Report , and AI companies have announced plans for 3,969 new US data centers, according to Aterio , a data center research company. But only about 800 are under construction, and AI experts have raised doubts about the viability of many of the announced projects.

If Tuesday’s election results trigger an investment slowdown, that could trim some of the industry’s fat.

Example A: CoreWeave, a neocloud business that rents out computing capacity, recently projected such extensive losses that its bond yields surged to 15% – the danger zone. That was particularly troubling, because its expected return on investment was only around 9%.

In other words, CoreWeave loses money on every single dollar it spends on expanding its data center capacity.

Deep-pocketed hyperscalers – Amazon, Google, Meta and Microsoft – don’t have that problem. But their stocks have been slammed in recent weeks as investors grow impatient with what they view as an unsustainable amount of spending that may not return profits for years.

“AI is catching a break here,” said Joe Brusuelas, chief economist at RSM US.

Brusuelas expects growth to slow from last quarter’s 3.4% pace to something closer to 2.4% if there is an AI slowdown. That’s still solid long-term growth. And America’s economy is massive and diverse. Investment that otherwise would have gone to data centers can go elsewhere.

“It’s slowing, but it doesn’t mean we’re going to have a recession,” Brusuelas said.

Omar Ornelas/El Paso Times/USA Today Network/Reuters

It also presents an opportunity for AI companies to get smart about their future AI data center rollouts – and get put on a more sustainable course.

The helter-skelter approach to data center construction has resulted in significant amounts of money getting flushed down the toilet in unworkable projects – collateral damage from massive competition to be first and biggest. Data centers already face $200 billion in project delays, according to Mike Mathews, global digital infrastructure practice leader at risk adviser Marsh.

A more rational, politically sensitive approach could create longer-term benefits for the communities that host the data centers and for the AI companies themselves.

There’s precedent for that. In the previous decade, the Googles of the world built out data centers for the cloud computing revolution, promoting the fact that they’d be built sustainably with renewable energy – even before it was cost-effective to do that. A similar, community-based approach could provide AI companies with the compute they need and give them the political and social license to operate their data centers for the long haul.

“Like the saying goes: If you want to go fast, you go it alone; if you want to go far you go together,” said Krejsa. “If you squander political license to operate at home, that’ll be a bigger impediment to remaining competitive with China.”

德克萨斯州州长选举的惊人逆转以及民主党在全国范围内的全面胜利，有可能遏制美国历史上规模最大的支出热潮。

这可能会扰乱美国经济的关键增长引擎。

据高盛预测，到明年1月，由支持扩大或延长人工智能建设禁令的民主党控制的州将占美国计划数据中心容量的45%。美国计划的人工智能基础设施建设中，四分之一将集中在德克萨斯州。

但这并不意味着美国一半的数据中心建设计划都被搁置了。人工智能的需求正处于空前高涨的状态。唐纳德·特朗普仍然是总统，而且他已经开始动用否决权。即便民主党州长们想阻止所有数据中心的建设，他们也无法做到。

但周二的选举仍然代表着一种范式转变，可能会改变美国在人工智能领域持续不断的支出热潮的速度和路径——据荷兰国际集团(ING)经济学家称，这场投资热潮已经占到今年美国经济增长总量的四分之一。

这些经济效益已经显现，创造了数千个就业岗位。人工智能还有可能彻底改变我们的工作方式。

但人工智能数据中心增速放缓也可能带来重要的积极影响。它或许能够抑制无节制的增长，这种增长此前已导致通胀反弹，并引发公众对这项技术及其数据中心的普遍反感。此外，它还有助于遏制一些备受质疑的商业模式，这些模式已在华尔街引发了严重的质疑。

踩下刹车最终可能会使人工智能——以及美国经济——走上一条更可持续的道路。

Jay Janner/《奥斯汀美国政治家报》/Getty Images

哥伦比亚商学院经济学家斯泰恩·范·尼乌韦尔伯格表示，在可预见的未来，人工智能基础设施支出预计将每年接近1万亿美元。如果按美国国内生产总值（GDP）的比例计算，这将超过美国现代史上任何其他投资热潮——包括铁路建设时期。

周二的选举结果可能会减缓这部分支出，对人工智能所支持的数千个就业岗位和大量经济发展产生巨大影响。

民主党在内华达州、佐治亚州、俄亥俄州和德克萨斯州赢得了州长席位，这些候选人在竞选期间都承诺加强对数据中心的监管。当选州长的提案包括取消税收优惠、收紧气候和电力限制，以及允许地方政府自行暂停数据中心建设。

Eli Hartman/Bloomberg/Getty Images

但德克萨斯州才是真正的大敌。富国银行估计，德克萨斯州计划中的数据中心容量每减少 10%，大型科技公司就会大幅削减其人工智能基础设施支出 1.6%（考虑到支出以万亿美元计，这可是一笔巨款）。

州和地方政客对这些支出有很大的管辖权和影响力：足以改变一些项目是否获得融资，Cooley 基础设施、能源和房地产业务联席主席 Mona Dajani 表示。

达贾尼指出，数百个仍在等待许可证、公用设施服务或税收优惠的数据中心开发项目可能尤其容易受到影响。但即使是正在建设中的项目也可能受到威胁。德克萨斯州法律对一些项目的许可证实行“祖父条款”，使其免受未来地方许可要求变更的影响——但这并不能解决州级许可变更或州长当选人吉娜·希诺霍萨可能取消的电力服务保障问题。

Dajani表示，等待审批可能会给人工智能公司带来巨大的成本：即使项目搁置，利息、设备投入和公用事业费用也会继续累积。在某个时候，现有项目可能会失去经济可行性。

如果这种情况发生，那些已经竭尽全力快速建设数据中心以满足预期需求的AI公司，可能无法实现其爆炸式的支出目标。

“这是一个罕见的案例，州和地方领导人具有地缘政治层面的影响力，”卡内基梅隆大学战略与技术研究所研究主任哈里·克雷萨说。

数据中心并非民主党独有的问题：美国两党已就数据中心建设设置防护措施达成共识。

德克萨斯州共和党州长格雷格·阿博特此前已经实施了数据中心建设禁令，而希诺霍萨竞选时承诺将延长该禁令，直到立法者制定更严格的人工智能法规为止。

但特朗普强烈反对任何阻碍人工智能发展和投资的企图，理由是经济和国家安全——尤其是在与中国的激烈竞争中。许多原本可能支持人工智能监管（包括数据中心建设框架）的共和党政客却保持沉默。

然而，有证据表明，人工智能的发展可能走上了一条不可持续的道路。

根据斯坦福大学人工智能指数报告，截至去年年底，美国共有5427个数据中心；数据中心研究公司Aterio的数据显示，人工智能公司已宣布计划在美国新建3969个数据中心。但目前仅有约800个正在建设中，人工智能专家对许多已宣布项目的可行性表示怀疑。

如果周二的选举结果引发投资放缓，这可能会削减该行业的一些冗余部分。

例A：CoreWeave是一家提供计算能力租赁服务的新型云计算公司，该公司近期预计亏损巨大，导致其债券收益率飙升至15%——进入危险区间。这尤其令人担忧，因为其预期投资回报率仅为9%左右。

换句话说，CoreWeave 在扩展数据中心容量上花费的每一美元都在赔钱。

财力雄厚的超大规模数据中心巨头——亚马逊、谷歌、Meta和微软——并不存在这个问题。但近几周来，由于投资者对他们认为不可持续的巨额支出（可能多年都无法带来利润）越来越不耐烦，这些公司的股价遭受重创。

RSM US首席经济学家乔·布鲁苏拉斯表示：“人工智能在这里迎来了一个转机。”

布鲁苏拉斯预计，如果人工智能发展放缓，经济增长速度将从上季度的3.4%降至接近2.4%。但这仍然是稳健的长期增长。而且，美国经济规模庞大且多元化。原本会流向数据中心的投资可以转向其他领域。

“经济增速正在放缓，但这并不意味着我们将面临经济衰退，”布鲁苏拉斯说。

奥马尔·奥内拉斯/埃尔帕索时报/今日美国网络/路透社

这也为人工智能公司提供了一个机会，让他们能够更明智地规划未来的人工智能数据中心部署，并走上更可持续的发展道路。

数据中心建设的混乱局面导致大量资金白白浪费在无法实施的项目中——这是竞相抢占先机、争当规模的激烈竞争所带来的附带损害。据风险咨询公司Marsh的全球数字基础设施业务负责人Mike Mathews称，数据中心项目目前已面临高达2000亿美元的延误损失。

更理性、更具政治敏感性的做法可以为数据中心所在的社区和人工智能公司本身带来更长远的利益。

这并非没有先例。过去十年，谷歌等公司为了推动云计算革命，建设了大量数据中心，并大力宣传这些数据中心将采用可再生能源以可持续的方式建造——即便当时这样做并不划算。类似的、基于社区的模式可以为人工智能公司提供所需的计算能力，并赋予它们长期运营数据中心所需的政治和社会许可。

克雷萨说：“俗话说：独行快，众行远。如果浪费了在国内开展业务的政治许可，那将更不利于你与中国保持竞争力。”