那些担心人工智能（AI）热潮正在演变成泡沫的投资者可能是对的——只不过这个过程比预期要早了几年。Federated Hermes公司的首席投资官史蒂夫·基亚瓦罗内（Steve Chiavarone）指出，相关基础设施的建设仍需要数年时间才能完成，只有在那时，潜在的负面后果才会变得更加紧迫。
Investors worried that the AI boom is turning into a bubble may be right — just several years too early, according to Federated Hermes’s Steve Chiavarone, who says the infrastructure buildout still has years to run before the risk of a hangover becomes more pressing.
基亚瓦罗内表示，当前市场对于哪些公司能够以何种利率获得资金有着明确的判断力；这与泡沫时期的情况截然不同——在泡沫时期，“所有公司都能以最低的利率获得资金”。他补充说，这些担忧虽然合理，但还为时过早。
Markets are being discerning about which companies receive capital at what rates — the opposite of bubble behavior, where “everybody gets financed at the cheapest rate,” said Chiavarone, the asset management firm’s chief investment officer for global equities. The concerns are legitimate, just premature, he added.
“我们仍需要三到四年的时间来完善这些基础设施。只有到那时，我们才能开始考虑可能出现的负面后果，”他在接受彭博电视（Bloomberg Television）采访时说道。“但在公司们还在继续筹集资金以加速资本支出的时候，根本无法考虑这些问题。”他的言论与亿万富翁肯·格里芬（Ken Griffin）在周二纽约会议上发出的警告相呼应：格里芬认为投资者“几乎可以肯定”正处于人工智能泡沫之中，但他同时也强调这项技术本身是真实存在的。
“You still need three, four years to build this infrastructure. Then on the other side, you can start to think about a hangover,” he said in a Bloomberg Television interview. “But you can’t think about that while companies are still raising capital to accelerate capex further.”His comments follow billionaire Ken Griffin’s warning at a New York conference Tuesday that investors are “almost certainly” in an AI bubble, even as he stressed the technology itself is real.
基亚瓦罗内指出，目前大部分人工智能相关的资本支出都是通过企业的自由现金流来支付的，债务的作用相对有限；而那些负债过高的公司则正面临压力。他认为，像Anthropic和OpenAI这样的公司存在被商品化（即其技术被其他公司模仿或替代）的风险，而其他大型科技公司则正在利用人工智能来巩固自己的竞争优势，并从中获得投资回报。
Chiavarone said most capital expenditure spending is funded through free cash flow, with debt playing a limited role, while over-levered firms are coming under stress. He sees commoditization risk for model companies such as Anthropic and OpenAI, in contrast with other large technology firms that are “using AI to build bigger moats around tremendous businesses” and seeing returns on investment.
他将当前的投资周期比作美国的公路建设计划以及二战后欧洲的重建过程，指出其中普遍存在的模式——建设期间创造大量就业机会，随后经济活动会陷入低迷，接着旧产业会被取代、新的产业才会出现——很可能会再次上演，只不过这次的情况可能会更加剧烈。在谈到美国政治时，他对中期选举对人工智能产业发展构成的威胁持轻视态度，称这些选举“是美国历史上最毫无意义的选举”，因为无论选举结果如何，都无法产生一个能够抵御阻挠议事程序（filibuster）的参议院多数派，也无法产生一个能够有效否决政府决策的国会。
He compared the current investment cycle to the US highway program and the rebuilding of Europe after World War II, saying the pattern that follows — job creation during construction, a subsequent lull, then displacement of work before new industries emerge — is likely to repeat, “maybe a little more intensely.”On US politics, he was dismissive of the midterm elections as a risk to the AI trade, calling them “the most meaningless election in the history of the United States” because no configuration of results would produce a filibuster-proof Senate majority or veto-proof Congress capable of changing national AI policy.
他认为，到2029年（届时新总统将上任并有可能改变相关政策），大部分数据中心基础设施已经建成或接近完工，因此任何政治上的喧嚣或争议实际上都难以对人工智能产业的发展产生实质性影响。
He argued that by 2029, when a new president could alter the policy landscape, much of the data center infrastructure will already be in the ground or nearing completion, limiting the practical impact of any political bluster.