Everything About AI
NextFin News - The worst two-day US equity selloff since April 2025 has produced a counterintuitive verdict from Wall Street's most influential strategists: buy it. Morgan Stanley's Mike Wilson frames the S&P 500's drop as a temporary washout that should bottom between 7,100 and 7,200 before rallying toward 8,000 by year-end, while Bank of America's Savita Subramanian - who lifted her year-end target to 7,400 - says the market is "overdue for a pullback" and should be used. The rout erased roughly $2.8 trillion from the S&P 500 and sent the VIX fear gauge to its biggest one-day leap since the Liberation Day tariff shock, yet the dominant read on the Street is that the bull market's foundations - earnings growth, AI-driven capital spending, and a divided government that reins in fiscal excess - remain intact.
The Rout in Numbers
Thursday's session was a clean, violent repricing. The S&P 500 dropped 4.36% to 7,353.73, the Dow Jones Industrial Average shed 1,833.06 points, or 3.44%, to 51,523.06, and the tech-heavy Nasdaq Composite slid 5.52% to 24,315.00, entering correction territory - a decline of more than 10% from its record. The Cboe Volatility Index briefly touched 28.09, up as much as 48% on the day, its largest single-session jump since April 4, 2025. All three major averages posted their worst day since April 2025, when President Donald Trump's global tariffs rocked markets.
The decline was broad, not narrow. The Russell 2000 small-cap index fell 4.45%, the technology sector dropped 8%, and the Philadelphia Semiconductor Index sank 12% - its worst day since the Covid-19 breakout of March 2020. AMD and Micron each fell more than 17%, Nvidia gave back 9.26%, and the iShares Semiconductor ETF posted a 12.15% loss. Market breadth closed at 243 advancers to 1,314 decliners on the S&P 500, a ratio that typically marks a capitulation event rather than a measured rotation.
By Friday's close the selling had moderated - the S&P 500 edged up 0.3% and trimmed its weekly loss, the Nasdaq Composite rallied 1.1%, and the Dow dipped 0.5% - but the damage was done to the AI trade that had powered the market higher. The neocloud names that had been the most leveraged expression of the AI boom came under particular pressure: CoreWeave, down 44.59% in a month, and Nebius, off 39.15%, now trade as the market's Exhibit A for the question nobody could ignore after Thursday - how much of this AI investment is actually going to pay for itself.
The selloff was not confined to the United States. Japanese and South Korean stocks fell sharply on Friday, and the Nasdaq 100 index dropped 6.3% on Thursday, erasing $2.3 trillion in market value. The breadth of the move - across regions, across the AI supply chain, and into the highest-beta expressions of the theme - is what makes the strategists' buy-the-dip call worth taking seriously. A narrow decline invites a narrow explanation. A broad one forces a judgment about the regime.
Why Strategists See a Buy Signal, Not a Bear Market
The bullish case rests on a simple distinction: this is a cyclical clearing event, not a structural break. Wilson told clients he sees "a few 'clearing events' ahead for the stock market that have the potential to take it lower temporarily, creating a great opportunity for investors to pile back in." His framework identifies three near-term headwinds - the midterm elections, stress over artificial-intelligence spending, and liquidity constraints - each of which resolves on a calendar rather than a crisis. The first has now cleared. Democrats won control of both chambers of Congress, taking the House with 248 seats to 150 for Republicans as of Friday afternoon, with 37 races still uncalled, and flipping the Senate. Historically, that outcome has been good for equities: US stocks have advanced during the six- and twelve-month periods following every midterm election since 1950.
"We are entering a seasonally weak period and in our view are overdue for a pullback," Subramanian wrote in a client note, framing the decline as a timing opportunity rather than a reason to exit.
The earnings backdrop supports the buy-the-dip thesis. Bank of America expects S&P 500 earnings to grow 33% in 2026 before slowing to 12% in 2027. JPMorgan, which raised its year-end target to 8,000 from 7,800, now forecasts 2026 earnings per share of $365 - up 35% - and $420 in 2027, while holding its forward valuation multiple at roughly 20 times. In other words, the Street is underwriting an 8,000 S&P 500 on profit growth, not multiple expansion.
"Despite AI bubble and valuation concerns, we see current elevated multiples correctly anticipating above-trend earnings growth, an AI capex boom, rising shareholder payouts, and easier fiscal policy," Lakos-Bujas wrote.
The claim is specific and testable: the index gets to 8,000 because earnings get to $365, not because investors decide to pay more for each dollar of earnings. There is a wider consensus behind the number. Oppenheimer's John Stoltzfus holds an 8,100 year-end target, and Yardeni Research's Ed Yardeni - before cutting his view on stagflation concerns - had argued for an earnings-driven cycle that would carry the index toward 10,000 by 2029. The bulls are not a lonely crowd making a contrarian call. They are the consensus, and the rout is a test of whether the consensus is right.
The macro setup is more complicated than the strategists' notes suggest, and that is where the second-order question lives. Markets are not pricing an intact easing path. As of early November, rate-probability trackers put the chance of a 25-basis-point increase at the Federal Reserve's December 9 meeting at roughly 73%, with about 56 basis points of net tightening expected over the following twelve months. That is the opposite of the rate-cut tailwind that powered the rally into the summer. The strategists' answer is that earnings can outrun rates - that a 35% jump in profits offsets a higher discount rate. It is a defensible arithmetic, but it is not the same market that rallied on the way up.
The Second-Order Question: Who Actually Pays for the AI Boom?
Here is the uncomfortable part of the bull case that most buy-the-dip notes gloss over. The rout was not random - it was concentrated in the AI supply chain, and it was triggered by a specific fear: that the capital being poured into data centers is outrunning the revenue coming back out. CoreWeave's second-quarter numbers crystallize the tension. The company posted revenue of $2.575 billion, up 112% year over year, and a backlog of $104.2 billion. Demand is real. But it also reported interest expense of $640 million against adjusted operating income of just $128 million - lenders collected five times what the business earned from operations. Over the year, interest expense rose 140% while operating profit fell 36%.
This is the transmission mechanism the market is now pricing, and it runs in three steps. First order: AI capex is expensive, and the companies building the infrastructure are financing it with debt at rising rates. Second order: when funding costs rise faster than operating margins, equity holders get squeezed - which is exactly what happened to CoreWeave's stock, down 44.59% in a month, and Nebius, off 39.15%. Third order: if the neoclouds must raise more capital - through debt at higher rates or equity that dilutes existing holders - the hyperscalers' promised returns on AI infrastructure compress, and the entire earnings-upgrade story that underpins the 8,000 target gets re-underwritten at a lower multiple.
The coming weeks will test this directly. CoreWeave, Nebius, IREN and the rest of the neocloud cohort are due to report third-quarter results in November. Investors will be checking three things: interest coverage, top-client concentration, and debt due within twelve months against cash on hand. A clean set of prints would validate the cyclical read. A string of warnings would turn a 9% correction into something more structural. The market is not asking whether AI demand exists - the backlogs prove it does. It is asking whether the economics of serving that demand can support the valuations attached to it.
There is a parallel stress point in the rate market. The ten-year Treasury yield stood at 4.84% as of Wednesday, and the two-year yield at 4.46% is pricing the Fed's near-term tightening path. Even without a fresh surge, those levels do two things at once: they raise the discount rate applied to future AI earnings, and they raise the coupon the neoclouds must pay on their next debt issuance. The same variable attacks both sides of the strategists' thesis - the multiple and the financing cost. That is why the bond market's move matters more than the equity market's.
The Strongest Case Against the Bulls
The counter-thesis is not that earnings will collapse tomorrow. It is that the market has already priced perfection, and that the AI trade is a structural margin problem, not a cyclical liquidity one. Subramanian, whose 7,400 year-end target is still among the Street's lowest, has warned that the companies making up the bulk of the S&P 500 are "the AI plays, the kind of the buy-the-dream companies," and that they could be headed for "a little bit of an air pocket." Her concern is multiple compression: investors paying 20 times forward earnings for growth that is about to slow from 33% to 12%.
That argument has teeth, and the arithmetic shows why. A forward multiple of 20 times on $365 of earnings produces an index level near 7,300 - not 8,000. To reach Wilson's and Lakos-Bujas's 8,000, either earnings must come in above the $365 consensus, or the multiple must hold at 20 times even as growth decelerates. Both require the AI revenue story to keep delivering. If it doesn't, the "buy the dip" call becomes a value trap: cheap relative to a peak that is not coming back.
There is also the rate problem. The strategists' base case assumes the Fed's path does not derail the expansion. But with markets pricing a 73% chance of a December hike and roughly 56 basis points of tightening over the next year, the discount rate is moving against the very earnings multiples the bull case depends on. A higher risk-free rate does not just lower the present value of future AI profits - it raises the hurdle those profits must clear to justify today's prices. This is why the counter-thesis is not a fringe view: it is what a 4.84% ten-year Treasury yield is telling you.
The falsifying signal is specific and observable. If CoreWeave and Nebius report third-quarter interest coverage below one times - meaning operating income still does not cover interest - and guide to another year of negative free cash flow, the cyclical-clearing thesis is wrong. At that point the AI capex cycle is not a temporary headwind; it is a structural drag on equity returns, and 7,100 would be a floor to break, not a floor to buy. A secondary signal: if the ten-year yield pushes back above 5% while AI revenue guidance disappoints, the multiple-compression case wins.
What to Watch and What It Means
The near-term path is volatile but directionally constructive - if the base case holds. A dip into the 7,100-7,200 zone followed by a grind toward 8,000 requires three things: earnings growth near the 33% consensus, the Fed's tightening path contained rather than accelerated, and neocloud financing stress contained. The upside case - a faster recovery toward 8,000 and beyond - needs clean Q3 neocloud prints and evidence that hyperscaler AI revenue is accelerating faster than capex. The downside case - a break below 7,100 toward the 6,400-6,500 zone that bears like Yardeni have flagged - would follow from two consecutive months of weak AI revenue guidance, a renewed spike in the ten-year yield above 5%, or a deterioration in the labor market that turns the Fed's tightening into something more aggressive than markets expect.
For investors, the asymmetry is clear but not one-sided. Quality balance sheets with real AI revenue - the hyperscalers and the chip designers with pricing power - are the beneficiaries of a correction driven by financing stress at the margin. The exposed are the highly leveraged neoclouds and the "buy-the-dream" names trading on promise rather than cash flow. The strategists are not saying everything is cheap; they are saying the index-level panic is disproportionate to the damage.
As of Friday's close, the S&P 500 stood 5.9% below its 52-week high of 781.62, and the Nasdaq 100 sat 14% below its peak - already inside correction territory. The strategists' call is that this gap closes not because sentiment recovers, but because earnings reports over the next six weeks confirm that the AI revenue stream is real. If it is, 7,100 will look like a gift. If it isn't, 7,100 will look like the first step down.
The verdict on Thursday was not wrong - it was a rational repricing of risk at the margin. The strategists' answer is that the margin is small, and the rest of the market is still earning its way higher. A correction that erases $2.8 trillion in two days feels structural while it is happening. It is the earnings print, not the headline, that decides which it was.
Explore more exclusive insights at nextfin.ai.
