The S&P 500's AI-driven rally to record highs is a mid-cycle acceleration, not a bubble, with years of upside remaining, Federated Hermes' Steve Chiavarone said.
The S&P 500's AI-driven rally to record highs is a mid-cycle acceleration, not a bubble, with years of upside remaining, Federated Hermes' Steve Chiavarone said.

The S&P 500's record-breaking rally is a mid-cycle acceleration, not a bubble, with years of upside remaining, Federated Hermes' Steve Chiavarone said.
"Bubbles laugh at current valuations," Chiavarone, global equity co-chief investment officer at Federated Hermes, said in a June 1 note. "Long-term bull markets typically last 20 years, and we're only at the midpoint. The market is accelerating, not peaking."
The S&P 500 has gained 11 percent this year to fresh all-time highs, with nearly 80 percent of that advance coming from just 10 companies — all in technology and seven of them semiconductor stocks. The Philadelphia Stock Exchange Semiconductor Index surged 69 percent in the past two months, on pace for its best quarter ever. The 10-year US Treasury yield held near 4.35 percent, while the dollar index edged lower to 104.2, providing a supportive macro backdrop. The Shiller cyclically adjusted price-to-earnings ratio stood at 42.04 as of May 22, more than double the 155-year average of 17.4 and within 5 percent of the dot-com bubble peak of 44.19.
The stakes are enormous because AI-related stocks now account for roughly 40 percent of total US equity market value, according to Geiger Capital data. If Chiavarone is wrong and the AI trade is a bubble, the unwind could erase trillions in market capitalization. If he is right, the current concentration in mega-cap tech could broaden into other sectors as the cycle matures.
Shiller P/E Near Dot-Com Records
The CAPE ratio's level has historically preceded significant drawdowns. Since 1871, there have been six instances where the Shiller P/E exceeded 30 during a continuous bull market, and the previous five all ended with the Dow, S&P 500, or Nasdaq losing at least 20 percent of their value. The one caveat: the CAPE ratio is not a timing tool. During the dot-com era, stocks remained expensive for up to four years before the bubble burst.
Former Lehman Brothers trader Larry McDonald has drawn direct parallels between the current AI enthusiasm and the dot-com peak, warning that the SpaceX IPO — seeking a valuation of at least $1.8 trillion — resembles buying Amazon in the spring of 2000. "The stock market is drunk on a narrative," McDonald said. "Artificial intelligence is the new everything."
AI CapEx Boom Faces Optimization Test
The four biggest buyers of computing equipment — Amazon, Meta Platforms, Alphabet, and Microsoft — expect to spend as much as $725 billion on capital expenditures in 2026, with most directed at AI data centers. They plan to spend significantly more in 2027. Yet history suggests businesses persistently overestimate how quickly they can optimize new technologies to generate profits, a pattern that played out during the internet buildout of the late 1990s.
Memory chipmakers have been the biggest beneficiaries. Micron Technology's shares have more than tripled this year, while SK Hynix surged 260 percent and Samsung Electronics gained 165 percent — all three now boast market capitalizations above $1 trillion. Micron's earnings are projected to jump to $66.8 billion in 2026 from $8.5 billion in 2025, according to Bloomberg Intelligence data.
For investors, the bull-versus-bubble debate carries concrete portfolio implications. If Chiavarone's mid-cycle thesis proves correct, the next leg of the rally could see leadership rotate from AI infrastructure into sectors that have lagged, including financials, industrials, and small-cap value. If the bears are right, the concentration risk in passive portfolios — where AI tech already accounts for nearly half of the S&P 500's market value — could trigger a severe drawdown when the narrative shifts.
This article is for informational purposes only and does not constitute investment advice.