The S&P 500 is coming off its best quarter since 2020 and sits within a whisker of its record, but Bank of America says the next decisive move is down. In a Tuesday note, the bank kept a 7,100 year-end target on the index, roughly 5% below recent levels, and warned that speculation has hit extremes. That call lands as mega-cap AI winners pull markets higher on good days — Broadcom jumped on a long-term supply deal with Apple — while semiconductors wobble and flows turn cautious. Traders are now weighing whether the next leg is a healthy pause or a valuation snapback that erases much of this year’s gains.
A relief bid in AI bellwethers pushed the S&P 500 up 0.7% on Monday, with the Nasdaq up 1.1% as Broadcom’s new pact with Apple revived chip demand hopes. But the rebound masked a shaky foundation. Hedge funds have been net sellers of tech hardware for four straight weeks, and the Philadelphia Semiconductor Index fell 4.2% in the week through July 3. The S&P 500, after printing a fresh peak in June, has chopped lower by about 2% since, while Korea’s Kospi — dominated by SK Hynix and Samsung — swung from record highs to one of its biggest single-day plunges in history. That is the kind of whipsaw that tends to show up late in cycles, not early.
BofA’s equity team reads the tape the same way. Their bear-market signposts, they say, are lighting up: high-multiple stocks gapping higher, free cash flow lagging net income, and rapid multiple expansion in a handful of winners. When that cocktail has appeared before, a valuation snapback followed. The target at 7,100 is not a crash call, but it is a bet that the path between here and year-end runs through lower prices, tighter financial conditions, and a reset in the parts of the market that have carried the tape.
What could force that reset? Policy. BofA expects the Federal Reserve to hike rates three times this year to break sticky inflation after more than five years above target. Stocks don’t usually die at the first hike; historical peaks often come six to 12 months afterward. But this time is different, BofA argues, because the S&P 500 enters the next tightening step at valuations surpassed only by 1999–2000. When multiples are already stretched, higher real yields become a direct hit to equity duration — especially for growth names whose cash flows sit far into the future.
That dynamic pairs uncomfortably with the profit math inside the AI trade. BofA highlights a widening gap between reported earnings and free cash flow, driven in part by what it calls hyperscaler capex burn. The AI build-out is real, and it is expensive. When earnings per share rise while cash generation falls, the market is paying up for operating leverage that has not yet converted into cash returns. A hawkish Fed compressing equity multiples while free cash flow is under pressure is how snapbacks start.
None of this is to say AI demand has cracked. Far from it. BofA’s semiconductor strategists, looking through the recent selloff, called the chip pullback a healthy reset ahead of the next rally and project a staggering $1.5 trillion of global cloud and AI infrastructure capex by 2027. Monday’s pop in Broadcom on its Apple deal underscored that order books at the best-positioned vendors remain firm. In other words, the long-term secular drivers are intact.
But the tape trades the short term. Micron’s surge — up 242% year to date and 700% from a year ago, even after a selloff — is the poster child for how fast expectations have moved. Meanwhile, hedge funds are dumping hardware, the SOX index is negative for the week, and realized volatility is climbing. That split-screen is classic late-stage momentum: fundamentals that look great on a three-year view, price action that looks crowded on a three-week view, and a setup where one earnings miss from a leader can flip the story.
The leadership question matters. On July 2, the Dow set a record at 52,900 while the Nasdaq fell and the S&P finished flat, a sign of money rotating into old-economy winners as tech paused. If that broadening continues, it can cushion the index from tech air pockets. If not, concentration risk grows. JPMorgan, in a mostly bullish update, lifted its S&P 500 target to 7,800 but paired that with a blunt warning: crowded momentum, especially in low-quality and speculative growth, carries a high probability of a flash crash. It also cited a tougher bar into second-quarter earnings, rising equity supply, and the prospect of tighter policy that squeezes multiples.
Put simply, even optimists are bracing for non-linear paths. That’s not a technicality. When everyone owns the same trade — AI enablers, hyperscalers, and their suppliers — liquidity vanishes quickly when the narrative wobbles. Price gaps widen, dealer hedging exacerbates moves, and mechanical selling can overwhelm fundamentals in a session. BofA’s talk of extreme speculation is aimed at that feedback loop, not the 2027 revenue line in an AI server plant.
The catalyst set is obvious. A hotter CPI or a hawkish Fed communication flips rate expectations back toward hikes and lifts real yields. A megacap misses on free cash flow or guides capex higher again, and valuation support wobbles. Internationally, another violent swing like Korea’s could spill into U.S. semis through ETFs and risk-parity rebalancing. None of those require a recession to ding stocks. They require only a market priced for perfection in a handful of sectors that has begun to trade as if perfection is the base case.
There are also micro tells. Dealers report rising skew in AI complex options, consistent with demand for downside protection. Semiconductor breadth has narrowed even inside the winners, with performance hinging on a tighter group of names. And the correlation between long-duration tech and real yields remains high. If the Fed does hike into sticky inflation — as BofA projects — that sensitivity moves to the front of the queue.
Bulls have a coherent rebuttal. Ed Yardeni just raised his S&P 500 target to 8,250, arguing this is nothing like the dot-com melt-up. Back then, forward P/E led the charge on FOMO. Today, he says, the engine is FEMO — fabulous earnings momentum — with corporate profits robust enough to carry loftier multiples. JPMorgan’s higher target also leans on stronger earnings estimates. If AI demand is compounding and the leaders keep beating on revenue and margins, the tape can absorb a lot of macro noise.
Both views can be right on different time frames. BofA’s own cross-currents show it: equity strategists warn of a valuation snapback while its chip team calls the selloff a reset before another leg higher. That split is the story. The secular AI thesis is alive. The near-term setup is fragile. When free cash flow lags net income, positioning is crowded, and policy risk rises, markets clear the imbalance with price. How deep that has to go depends on whether leadership broadens and whether earnings momentum meets the new bar.
The next tests arrive fast. Second-quarter results from the AI complex will reveal whether capex is peaking or reaccelerating and whether cash generation is catching up. Fed minutes and inflation prints will reset the policy path. Flows into and out of semis — after a four-week hedge fund sell streak — will show if the reset is done. If breadth improves and megacaps keep delivering cash, a drawdown can stay shallow. If not, BofA’s snapback call moves from thesis to tape, and this year’s gains get marked to a more honest price.
