Since the end of July a handful of artificial-intelligence stocks has been propping up the S&P 500 near all-time highs while the vast majority of the index's constituents were either in a bear market or something close to it. Over the next six months there's going to be "a monstrous bull market under the surface" of the S&P 500, and it will be the "unmagnificent 493" that drives it.

Major indexes like the S&P 500 and Nasdaq won't reflect the strength of the underlying market, Thomas Hayes is telling investors, because of the heavy weightings in the indexes of the AI plays, but those companies that are adopting AI and boosting productivity as a consequence will "absolutely gallop."

Hayes is the founder and a managing member of hedge fund Great Hill Capital, and he was conveying his contrarian views on David Lin's podcast, which first aired Tuesday. His opinions on the AI sector, and the recent correction in its infrastructure build-out plays, were emphatic: A commodity trade within the AI trade "is done."

By commodity trade, Hayes is mainly referring to the memory stocks that have been labeled the picks-and-shovels play of the AI infrastructure boom. In commodity markets, Hayes opines, "high prices cure high prices," meaning that the huge margins that were on offer for the chip makers will evaporate because of the intensifying competition for cheaper suppliers. "The story of semiconductors and memory is the same every four years," Hayes says, explaining: "You have a shortage, they increase the capacity, competition comes on board just in time for demand to go down."

Thomas Hayes of the hedge-fund firm Great Hill Capital.

History shows that those companies funding capital spending were "the biggest losers," while those utilizing the new technology - in this instance artificial intelligence - to boost productivity and expand margins will be the ultimate winners. The costs of developing frontier models, he believes, are "out of control," and the math involved in generating sufficient free cash flow for OpenAI and Anthropic justify their valuations is too demanding, as far as Hayes is concerned.

Hayes's fund is implementing AI and it's increasing productivity, but he is running open-weight models internally. This means switching to open-source models like Mistral AI, Alibaba's (BABA) Qwen or Meta's (META) Llama.

So, if semiconductors are too frothy and the AI developers are mispricing risk, where does Hayes see opportunities? He likes Comstock Resources $(CRK)$, a low-cost natural-gas producer strategically located near Gulf of Mexico export terminals and some power-hungry data centers under construction in eastern Texas.

Hayes is also extremely bullish on Alibaba because "it owns equity in pretty much every AI startup in Asia," and he highlights its Moonshot holding, which could go public soon with a $50 billion valuation. Additionally, Hayes cites stocks like cell-tower play Crown Castle $(CCI)$ and GXO $(GXO)$, which is involved in logistics and warehouses.

In general, Hayes believes the huge-stock market rotation he predicts away from the so-called Magnificent Seven MAGS and AI-centric names, toward what can be called the "S&P 493" RSP, is already underway. Predominantly defensive sectors like healthcare XLV, utilities XLU and staples XLP are all getting a bid, he observes, and rate-sensitive companies like residential builders XHB are starting to perform.

Hayes explains that this as a result of investors now finding value in bonds. A guaranteed yield of 5.30% on U.S. 10-year debt BX:TMUBMUSD10Y seems attractive to him, compared with a 5% nonguaranteed yield in equities.