ROI-华尔街对AI的赌注如今已太大，难以实现：McGeever 持续推动股票上涨的AI驱动因素面临着惊人的高门槛。而随着看似坚不可摧的科技大厦开始出现裂纹，越来越多的迹象表明，AI风险如今远不止是“情绪不佳”这么简单。尽管上周标普500指数和纳斯达克指数创下历史新高，但一系列与AI相关的负面新闻——从澳大利亚一项首次公开募股（IPO）被搁置，到有报道称ChatGPT的开发者OpenAI的收入未达早前预期——表明市场可能很快就会遭遇一些严重障碍。
ROI-Wall Street’s AI bet is now too big to hit: McGeever The hurdle for continued AI-driven equity gains is strikingly high. And with cracks starting to appear in the seemingly impenetrable tech edifice, there are mounting signs AI risk is now far more than just a case of "bad vibes." While the S&P 500 and Nasdaq climbed to fresh highs last week, a string of negative headlines related to AI – from a shelved initial public offering (IPO) in Australia to reports that ChatGPT maker OpenAI's revenue isfalling short of earlier estimates – suggests markets could soon run into some serious obstacles.
未来几周，来自主要AI企业的财报和预测或许能缓解这些担忧。但门槛很高——非常高。根据LSEG收益与股票研究主管Tajinder Dhillon的说法，分析师预计标普500指数成分股公司第三季度的盈利将同比增长超过30%，其中三分之二将来自科技和AI。一些分析师表示，标普500指数公司利润总增长的预期增加中，多达80%将来自这一单一领域。投资者保持谨慎是正确的。
Earnings releases and forecasts from the big AI players in the coming weeks may ease those concerns. But the bar is high – very high. Analysts expect S&P 500 earnings to rise more than 30% year-over-year for the third quarter, with two-thirds of that coming from tech and AI, according to Tajinder Dhillon, LSEG's head of earnings and equity research. Some analysts say up to 80% of the anticipated increase in total profit growth from S&P 500 companies will come from this one area. Investors are right to be cautious.
周五，澳大利亚的Firmus公司——一家由英伟达支持的数据中心运营商——搁置了其50亿美元的首次公开募股（IPO）。这本来将是该国历史上第二大的首次公开募股，也是今年全球第四大的IPO。这显然是一个负面转折，但即便是来自AI企业的积极消息也可能不够好。周四，全球最大内存芯片制造商——韩国的三星电子表示，它预计将宣布第三季度盈利约为800亿美元，这将成为任何科技公司有史以来最高的季度利润。
On Friday, Australia's Firmus, a data center operator backed by Nvidia, shelved its $5 billion IPO. It would have been the country's second-largest initial share sale ever, and the fourth largest globally this year. That was obviously a negative twist, but even positive news from AI firms may not be good enough. On Thursday, South Korea's Samsung, the world's largest memory chipmaker, said it expects to announce third-quarter earnings of around $80 billion, which would be a record quarterly profit for any tech company.
同一天，全球最大代工芯片制造商台积电报告称，其第三季度收入创纪录达到467亿美元，超出预期，较去年同期增长50%。然而，两家公司的股价均出现下跌。当然，孤立的股价波动和公司特定公告并不能反映一个价值数万亿美元行业的发展趋势。
On the same day, Taiwan's TSMC, the world's largest contract chipmaker, reported record third-quarter revenue of $46.7 billion, beating forecasts and rising 50% from a year earlier. Yet shares in both companies fell. Of course, isolated share price moves and company-specific announcements aren't indicative of the trajectory of a multi-trillion-dollar industry.
但它们是连续不断的不佳新闻滴滴的一部分。我们也不应忘记，上周还传出OpenAI截至9月的年化收入约为500亿美元，较之前的预测下降了200亿美元。这一切表明，在这一财报季，更广泛的人工智能前景可能面临严峻审视。
But they are part of a steady drip of less-than-stellar headlines. Let's not forget, it also emerged last week that OpenAI's annualized revenue for September was around $50 billion, down $20 billion from previous indications. This all suggests the bigger AI picture could face some seriously harsh scrutiny this earnings season.
最高门槛 人工智能革命所需的投入，以及因此而要求的盈利能力，数额惊人。
THE HIGHEST BAR The sums being spent on the AI revolution, and therefore those required to make it profitable, are eye-watering.
根据哥伦比亚大学教授Stijn Van Nieuwerburgh的估算，从去年到2032年融资人工智能建设的总成本约为10.3万亿美元，相当于美国国内生产总值年均的3.63%。假设保守的10%回报率——低于许多分析师认为人工智能投资所瞄准的15%-20%投资回报率——Van Nieuwerburgh估计，到2032年，为“收回投资、更换折旧资本以及补偿投资者资本机会成本”所需的年收入约为3.55万亿美元。
The total cost of financing the AI buildout from last year to 2032 is estimated at $10.3 trillion, or an average of 3.63% of US gross domestic product per year, according to Columbia University professor Stijn Van Nieuwerburgh. Assuming a conservative 10% return — below the 15-20% return on investment many analysts say AI investment is targeting — Van Nieuwerburgh estimates the annual revenue required "to recover the investment, replace depreciating capital, and compensate investors for the opportunity cost of capital" is around $3.55 trillion by 2032.
这相当于美国国内生产总值的9%以上，大约相当于美国人每年在机动车辆、交通运输、家庭家具、服装和能源上的总支出。考虑到OpenAI和Anthropic目前的年合并收入约为1000亿美元，Van Nieuwerburgh表示，为了达到所需的3.55万亿美元总收入，年收入增速需在2032年前保持每年超过80%的增长。“如果信息技术设备的经济寿命更短，收入需求将显著上升，”他表示。
That's more than 9% of GDP, roughly what Americans spend on motor vehicles, transport, household furnishings, clothes, and energy combined. When you consider that annual combined revenue of OpenAI and Anthropic is currently around $100 billion, total revenue would need to grow more than 80% a year through 2032 to get to the required $3.55 trillion total, Van Nieuwerburgh says. "If the economic life of IT equipment is shorter, the revenue requirement rises substantially," he said.
有史以来最大的资金错配？人工智能和科技公司的利润迄今为止基本上符合华尔街飙升的预期。它们能否继续保持这一趋势？
BIGGEST MISALLOCATION EVER? AI and tech firms' profits have, so far, mostly met Wall Street’s soaring expectations. Can they continue to do so?
摩根大通私人银行的迈克尔·森巴莱斯特表示，他将人工智能相关股票缩小到41只，这些股票自2022年11月ChatGPT推出以来，贡献了标普500指数收益的75%、盈利增长的80%以及资本支出增长的90%。如此少数公司承担如此沉重的负担，实属罕见。这41家公司的首席执行官和首席财务官在未来几周提供的展望，可能表明市场修正是否即将到来——以及可能的规模有多大。
JPMorgan Private Bank's Michael Cembalest says AI-related stocks, which he narrows down to 41 names, have accounted for 75% of S&P 500 returns, 80% of earnings growth and 90% of capital spending growth since ChatGPT launched in November 2022. That’s an extraordinarily heavy burden for so few companies. The outlooks provided by the CEOs and CFOs of these 41 firms in the coming weeks could indicate whether a market correction could be on the horizon – and just how large it could be.
人工智能显然是美国历史上最大的投资热潮。它是否也可能最终成为有史以来最大的资本错配？
AI is far and away the biggest investment boom in US history. Could it also end up being the biggest misallocation of capital?
喜欢这篇专栏吗？请查看路透社公开利润（ROI），这是您获取全球金融评论的必备新来源。在领英和X上关注ROI。And
Enjoying this column? Check out Reuters Open Interest (ROI), your essential new source for global financial commentary. Follow ROI on LinkedIn, and X. And