GOOD MORNING, Friday turned the AI pullback into something more serious. The Philadelphia Semiconductor Index closed 20.2% below its June record, officially entering bear-market territory as the selloff spread into Meta, Alphabet and the broader market. Yet the earnings backdrop underneath it was still unusually strong: about 90% of S&P 500 companies that had reported were beating expectations, with aggregate profit growth tracking around 26%. At the same time, only three commodity vessels crossed the Strait of Hormuz Thursday. AI valuations cracked before AI earnings did — while energy risk kept the macro floor unstable. MARKETS | TLDR
AI SELL-OFFThe AI Trade Broke Before the Earnings DidThe Philadelphia Semiconductor Index ended Friday 20.2% below its June 22 record close, crossing the conventional threshold for a bear market. The index had already fallen more than 18% in July and logged its worst weekly decline in over a year. The damage also spread beyond chips: Meta fell 2.7%, Alphabet dropped 3.2% and every Magnificent Seven stock except Apple finished lower. The striking part is that the fundamental backdrop had not collapsed with prices. Of the 49 S&P 500 companies that had reported second-quarter results, roughly 90% were beating expectations, according to LSEG. Analysts were tracking aggregate earnings growth of about 26% year over year, up sharply from the 19.2% expected at the start of April. This was not an earnings recession. It was a reset in what investors were willing to pay for the market's most crowded trade. That distinction matters heading into the next phase of earnings season. AI demand remains strong, but valuations had embedded years of extraordinary growth and accelerating spending. Once expectations get that high, even good numbers can stop being enough. The next question is not whether AI continues to grow. It is whether hyperscaler spending, chip revenue and cash generation can keep beating a bar the market has already pushed extremely high. AI CAPEXInvestors Are Starting to Rotate Away From the BuildersSome active managers are starting to cut semiconductor exposure and move toward hyperscalers, software companies and industries expected to benefit from AI adoption. The reason is not an expected collapse in infrastructure spending. UBS still sees hyperscaler capex rising 76% this year to roughly $673 billion. The concern is what happens when that growth rate slows to 25% in 2027 and just 6% in 2028. That matters because chip suppliers benefited from acceleration, not simply from a high absolute level of spending. If Microsoft, Amazon, Alphabet and Meta keep investing hundreds of billions of dollars but stop increasing those budgets at the same pace, the economics can improve for the buyers while revenue expectations for suppliers become harder to meet. There is a financing constraint too. Hyperscalers are increasingly turning to external capital after funding the first AI buildout mostly from internal cash. Bond-market cover ratios have fallen sharply this year, and the BIS has warned that disappointment in AI returns could trigger a sudden pullback in financing. The AI boom can remain real and enormous while the market still changes its mind about which layer deserves the premium. HEADLINES
UPCOMING
DEEP INSIGTHSAmong AI Crowd, Some Investors Position for Slower Hyperscaler Spending GrowthRead this for the portfolio logic behind the semiconductor selloff. UBS sees hyperscaler capex growth slowing from 76% this year to 25% next year and 6% in 2028. That does not imply the AI buildout is ending. It implies the winners can rotate from companies selling the infrastructure toward the hyperscalers, software companies and end users that benefit when spending growth becomes more disciplined. Strait of Hormuz Transits Drop as U.S. and Iran EscalateThis is the physical-market counterpart to the geopolitical headlines. Only three commodity vessels crossed the strait Thursday versus roughly 125 per day before the conflict, and no VLCC or LNG tanker made the passage for a second day. That is the kind of disruption that can turn a geopolitical risk premium into a real inflation input if it lasts. |