摩根大通公司西塔拉·桑达尔在达尼·伯杰撰写的文章中表示，鉴于人工智能行业正经历两个周期同时推进所带来的“健康”波动，她敦促投资者分散配置人工智能领域的投资。
By Dani Burger JPMorgan Chase & Co.’s Sitara Sundar urged investors to diversify their bets on artificial intelligence because the industry is caught in the “healthy” volatility of two cycles occurring at once.
桑达尔周三接受彭博电视采访时表示，由金融和基础设施驱动的周期目前已进入“中局”，超大规模云服务商正转向通过资本市场发行证券融资，而非依赖自身现金流。
The finance-and-infrastructure cycle is now in its “mid-innings,” Sundar said Wednesday in a Bloomberg Television interview, as hyperscalers move to capital markets issuance rather than relying on their own cash flow.
与此同时，人工智能与经济各领域的融合仍处于早期阶段，生产率提升才刚刚开始惠及企业。桑达尔是摩根大通私人银行另类投资策略主管。她说：“我认为，我们在科技领域看到的波动是健康的。这是正常的，也是我们今年初可能已经预料到的。”今年由人工智能推动的股市上涨掩盖了科技行业内部巨大的分化。
截至周二，费城半导体指数上涨了78%，而iShares扩展科技—软件板块ETF同期下跌0.4%。
At the same time, AI’s integration into the economy is still in the early stages, with productivity gains just starting to reach businesses, she said. “The volatility that we’re seeing within the tech space, I think, is healthy. It’s normal, and it’s something that we probably anticipated coming into this year,” said Sundar, the head of alternative investment strategies at JPMorgan’s private bank. The AI-driven surge in stocks this year masks huge variations across tech, with the Philadelphia Semiconductor Index surging 78% through Tuesday and the iShares Expanded Tech-Software Sector ETF down 0.4% in the same period.
桑达尔主张重点投资少数超大规模云服务商，在半导体行业精心筛选标的，并关注能够在三年内受益于人工智能生产力提升的创业投资及私募股权持有企业。
Sundar favors leaning into select hyperscalers, being choosy within semiconductors, and looking to venture capital and private equity-owned businesses that stand to benefit from AI-driven productivity over a three-year horizon.
桑达尔还运用这种双轨方法评估企业人工智能即将进一步扩张等议题，例如谁能在这场扩张中胜出；Meta Platforms Inc.本周推出了一个新平台。
Sundar applied her dual-track approach to assessing issues such as who wins in the upcoming expansion of enterprise AI, where Meta Platforms Inc. created a new platform this week.
她说，在面向消费者的广泛人工智能应用场景中，超大规模云服务商可能仍会占据优势，而专业运营商则很可能在金融或法律等细分领域取得成功。她补充道：“我认为，两种相互矛盾的判断可以同时成立。”
Hyperscalers probably will still prevail when competing on broad, consumer-oriented use cases for AI, she said, while specialty operators likely will succeed in niches such as finance or law. “I think two truths can exist at the same time,” she added.
对于政府债券收益率升至多年高位所带来的挑战，她同样持审慎态度。
She was equally measured on the challenge from government bond yields at multiyear highs.
桑达尔表示，人工智能基础设施建设的中性利率与更广泛经济体的中性利率“根本不同”，因为人工智能算力需求仍未受到影响。
The neutral rate for the AI buildout is “fundamentally different” from the neutral rate for the broader economy, Sundar said, because demand for AI compute remains unaffected.