美国利率的一个巨大变数是人工智能投资热潮的轨迹及其对股市估值的影响。
A big wild card for US interest rates is the trajectory of the artificial-intelligence investment boom and its impact on stock market valuation.
这股热潮目前正从两个方面推高利率。首先，投资支出正在提振经济增长并对该国的融资能力造成压力。其次，热潮伴随着股市的繁荣，这正在提升家庭财富并支撑消费支出。
The boom is currently pushing interest rates up in two ways. First, the investment spending is boosting economic growth and straining the nation’s financing capacity. Second, the boom has been accompanied by a buoyant stock market, which is lifting household wealth and bolstering consumer spending.
对短期和长期利率的影响很难预测，因为我们无法确信这场投资热潮的规模和持续时间及其对股市估值的影响。但有几点似乎很可能发生。首先，对美国经济增长的直接影响明年可能会减弱，因为2027年投资支出的增长——对增长而言重要的是变化量，而不是总量——可能小于2026年的增幅。这在一定程度上反映了建设新数据中心的产能瓶颈、其他业务现金流较少的人工智能服务商面临的更严苛融资条件，以及政治阻力。选民对人工智能建设对电价的影响感到不满，并担忧人工智能将如何影响他们的生计。
The effect on short- and long-term interest rates is difficult to predict because we can’t be confident about the magnitude and duration of the investment boom and its impact on stock market valuation. But a few things seem likely. First, the direct impact on US economic growth is likely to diminish next year because the increase in investment spending in 2027 — it’s the change, not the level, that matters for growth — is likely to be smaller than the rise in 2026. This reflects in part capacity constraints in building new data centers, tougher financing conditions for those AI providers with little cash flow from other operations, and political opposition. Voters are unhappy about the impact of the AI buildout on electricity prices and are concerned about how AI will affect their livelihoods.
其次，相比之下，人工智能投资对国家储蓄池的压力所导致的对实际利率的影响将继续增加。这是因为对储蓄供应的压力是由投资规模驱动的，而不是由其变化驱动的。
Second, in contrast, the impact on real interest rates caused by the pressure of AI investment on the nation’s saving pool will continue to increase. That is because the pressure on the supply of saving is driven by the level of investment, not its change.
因此，人工智能热潮对利率的上行压力在2027年可能会变得更加微妙——来自经济增长的压力减少，但来自储蓄压力的影响增加。利率必须上升，以压低消费者对耐用品、住房以及与人工智能无关的企业投资的支出，从而腾出产能来适应更高水平的人工智能支出。
Thus, the upward pressure the AI boom puts on interest rates is likely to be more nuanced in 2027 — less from economic growth but more from the pressure on savings. Interest rates have to rise to push down consumer spending on durable goods, housing and non-AI related business investment to generate the capacity to accommodate a higher level of AI spending.
长期来看，随着人工智能投资热潮达到顶峰随后回落，利率的上行压力将会逆转。当这股热潮不可避免地消退时——无论是由于基础设施建设完成，还是由于人工智能提供商无法获得所需的外部资金——情况将发生逆转：人工智能投资将成为经济增长的拖累，并将占用较少的储蓄池。与此同时，随着热潮带来的利润消退，股市可能会收缩。这两者的结合将导致经济放缓，美联储将扭转方向并下调短期利率，长期利率也将随之下降。
In the longer term, the upward pressure on rates will reverse as the AI investment boom peaks and then declines. When the boom inevitably unwinds — either because the buildout is complete or because AI providers can’t obtain the outside funding they need — things will run in reverse: AI investment will be a drag on growth and will take down less of the saving pool. At the same time, the equity market will likely retrench as the boom-generated profits subside. The combination will slow the economy, the Fed will reverse course and cut short-term rates, and longer-term rates will decline.
对投资者（以及美联储）而言，问题在于这种逆转的时机和强度仍然高度不确定。我的观点是，它极有可能会比预期发生得更早。这是因为我怀疑人工智能提供商能否产生足够的收入来获得可接受的投资回报。此外，在我们到达那个时间点之前，那些没有强劲现金流的人工智能提供商将失去进入资本市场的机会。
The problem for investors (and the Fed) is that the timing and intensity of this reversal remain highly uncertain. My view is that it will most likely occur sooner than anticipated. That’s because I’m skeptical that the AI providers will be able to generate sufficient revenue to achieve an acceptable return on investment. Moreover, before we get to that point, those AI providers without strong cash flows will lose access to the capital markets.
那么，这对美国利率意味着什么？在我看来，当美联储通过大幅降息来应对泡沫破裂时，其走势将首先是上升——其幅度甚至可能大大超出预期——而不是下降。有趣的是，这并不是金融市场参与者所预期的。这一点可以从有担保隔夜融资利率（SOFR）期货价格中看出，这些期货价格一直延伸到2030年代初（不同于联邦基金市场），并且与截至2027年6月左右再加息约75个基点、此后保持相对平稳的预期相符。
So what does this imply for US interest rates? In my view, the pattern will first be up — perhaps significantly more than anticipated — rather than down, when the Fed will respond to the bust by cutting interest rates sharply. Interestingly, this is not what financial market participants expect. This can be seen in SOFR futures prices, which extend into the early 2030s (unlike the federal funds market) and are consistent with about 75 basis points of additional hikes by June 2027 and then remain relatively flat thereafter.
尽管这可能仅仅代表了那些认为热潮将更具持续性的人与持相反观点的人之间的分歧，但无论原因如何，我认为这不是一条可能的发展路径。如果短期利率上升75个基点或更多，随之而来的很可能是一个急剧的下行周期，因为这与人工智能热潮对经济的可能影响以及对美联储反应的决定作用更加契合。
Although this could just represent disagreement between those who think the boom will prove more persistent and those who don’t, regardless of the reason, I think this is an unlikely path. If short-term rates rise by 75 basis points or more, a sharp down cycle seems likely to follow because that would be more consistent with how the AI boom is likely to affect the economy and determine the Fed’s response.
那么，下周的联邦公开市场委员会会议将如何？美联储是否滞后于形势？我认为美联储应该更快采取行动，原因有二。首先，人工智能的热潮正在推动经济足够快地发展，从而使劳动力市场维持在美联储充分就业的职责目标上。其次，这种情况发生时通胀率依然处于高位，美联储几乎可以肯定是连续第六年错过2%的通胀目标。
So what about next week’s Federal Open Market Committee meeting? Is the Fed behind the curve or not? I think the Fed should act more quickly for two reasons. First, the AI boom is pushing the economy along fast enough to keep the labor market at the Fed’s mandate of full employment. Second, this is happening when inflation is still elevated, with the Fed almost certain to miss its 2% inflation target for the sixth consecutive year.
此外，政策失误的风险仍然严重倾向于通胀这一侧，因为通胀率持续未能达标可能会导致通胀预期失锚。如果这种情况发生，美联储的信誉将受到损害，这将使这家央行更难将通胀压回其目标水平。
In addition, the risks of a policy mistake remain very much tilted to the inflation side because further persistent misses could cause inflation expectations to become unanchored. If that were to happen, the Fed’s credibility would be damaged, and that would make it more difficult for the central bank to push inflation back down to its target.
联邦公开市场委员会在下周不会提高其联邦基金利率目标。两位副主席——纽约联邦储备银行行长约翰·威廉姆斯（联邦公开市场委员会成员）和菲利普·杰斐逊（联邦储备委员会成员）的言论并非偶然，这些言论似乎旨在缓和市场预期。所谓的“三人领导小组”由主席和两位副主席组成，自前主席本·伯南克时期起便成为主要的决策集团和一致的沟通渠道，目前看来依然在发挥作用。
The FOMC is not going to raise its federal funds rate target next week. Comments by the two vice chairs — New York Fed President John Williams (the FOMC) and Phillip Jefferson (the Board of Governors) — which seemed designed to temper market expectations, were no accident. The so-called troika, which consists of the chair and the two vice chairs and became the dominant decision-making group and a consistent means of communication beginning with former Chair Ben Bernanke, still seems to be operative.
我得出这一结论是因为，主席凯文·沃什上周在曼谷举行的国际货币基金组织和世界银行年度会议上的发言，实质上赞同了威廉姆斯和杰斐逊的观点。
I reach this conclusion because Chair Kevin Warsh essentially endorsed Williams and Jefferson’s sentiments last week in remarks at the annual meetings of the International Monetary Fund and World Bank in Bangkok.
尽管沃什对提供前瞻性指引表现得较为谨慎，但他确实指出，自9月联邦公开市场委员会（FOMC）会议以来，全球经济大体上符合美联储的预期。由于期货市场定价显示下周收紧货币政策的概率仅为15%左右，如果沃什通过不提供前瞻性指引来违背他一直努力保持不受干扰的金融市场信号，那将令人震惊。此外，此举也将引人注目，因为它基本上标志着“三驾马车”成员一致传达信息的做法走到了终点。
While Warsh demurred from providing forward guidance, he did note that the global economy had broadly conformed to the Fed’s expectations since the September FOMC meeting. With futures markets pricing in only about a 15% probability of tightening next week, it would be shocking if Warsh contradicted the signal from financial markets that he has worked to keep untainted by not providing forward guidance. Moreover, a move would be noteworthy because it would essentially mark the end of consistent messaging by the troika members.
在2009年至2018年期间，当我与前主席伯南克、珍妮特·耶伦和杰罗姆·鲍威尔共同组成“三驾马车”时，我确保了自己的言论与他们的偏好以及委员会在即将召开的会议上实际采取的行动保持一致。如果市场定价出现偏差，我可能会在主席的许可下尝试拨乱反正。例如，在2015年8月，我曾说过：“目前来看，在9月会议上开始政策正常化进程的决定似乎没那么令人信服……”这番话并非偶然，而是旨在重置市场预期。
When I was part of the troika with former chairs Bernanke, Janet Yellen and Jerome Powell, from 2009 to 2018, I made sure that my remarks lined up with their preferences and what the committee would actually do at the coming meeting. And if market pricing was wrong, I might — with the blessing of the chair — try to set things right. For example, in August 2015, it was no accident that I said: “At the moment, the decision to begin the normalization process at the September meeting seems less compelling… .”That comment was designed to reset market expectations.
虽然美联储下周的会议可能会有些乏味，但请系好安全带，因为在未来几年里，情况将不再如此。
While next week’s Fed meeting is likely to be a bit of a snooze, buckle your seat belt because over the next few years, that will not be true.