Everything About AI
NextFin News - The most important number in the U.S. economy this week is not the one the Bureau of Labor Statistics is about to publish. Jeffrey Rosenberg, a systematic multi-strategy fund portfolio manager at BlackRock, says investors are paying less attention to the employment report than to overseas developments, futures positioning, and a deeper question: whether the AI-driven growth outlook can sustain the valuations built on top of it. The comment, made Thursday in a television interview, lands a day after the worst U.S. stock-market session since April 2025, when the Nasdaq Composite fell 5.4%, the S&P 500 dropped 4.4%, and roughly $2.5 trillion of equity market value evaporated. The message is blunt: the jobs report is data; AI is the regime.
The Jobs Report Everyone Is Ignoring
The October employment report, due Friday at 8:30 a.m. ET, is widely expected to show a meager gain of 59,000 nonfarm payrolls, with the unemployment rate holding at 4.2%. September's print was just 29,000 jobs, one of the weakest of the cycle, and the private-sector ADP report earlier in the week showed hiring of 86,000 positions in October. On paper, this is exactly the kind of soft labor data that should dominate the macro conversation and pin hopes on a Federal Reserve rate cut.
Instead, the market's attention has shifted. Rosenberg's point is that the marginal price action is being set by the health of the AI trade, not by the labor tape. That reading is consistent with what actually happened Wednesday: the Philadelphia Semiconductor Index sank 12.25%, Nvidia fell 9.3%, and AMD dropped about 15%, while the benchmark 10-year Treasury yield fell to its lowest level since Sept. 9 — a classic flight-to-quality move, not a jobs-driven one.
Why the combination matters: a soft jobs report and an AI selloff should, in a normal cycle, point the same direction — toward lower discount rates and relief for long-duration growth stocks. Instead, the AI growth engine is being questioned at the same moment the case for Fed easing is strengthening. That divergence is what Rosenberg is flagging, and it is the tension this piece unpacks.
The Overseas Link: Why This Is Bigger Than One U.S. Data Point
Rosenberg's reference to "overseas developments" is not incidental. The AI trade is global, and the reversal has been global with it. Asian technology shares slumped after the U.S. tech rout, with semiconductor names across Taiwan, South Korea, and Japan caught in the same downdraft that began on Wall Street. The Nasdaq 100 alone gave back roughly $2.3 trillion from its peak as the AI trade went into reverse, and the contagion ran through the supply chain rather than stopping at U.S. borders.
This is the structural reason a jobs report struggles to command attention. A single monthly payroll print is a domestic, backward-looking statistic. The AI build-out is a cross-border capital cycle that ties U.S. chip designers to Taiwanese foundries, Korean memory makers, Japanese equipment suppliers, and data-center developers from Virginia to Southeast Asia. When the marginal question is whether that entire chain is overbuilt, a 59,000-job U.S. estimate is not just small — it is the wrong variable.
The futures market tells the same story. Equity index futures were trading the AI growth narrative, not the employment narrative, into the close. That is what a systematic manager means when he says the focus has shifted: the risk factor that explains the largest share of daily variance is no longer macro data; it is the perceived durability of AI-driven earnings.
What "Focused on AI" Means in the Positioning
When a systematic multi-strategy manager says markets are focused on AI rather than jobs, he is describing a positioning hierarchy. The AI trade has become the dominant factor in equity risk: the Nasdaq's 5.4% one-day drop erased $2.5 trillion in market value, and the chip index's 12.25% plunge shows where the leverage sits. Jobs data, by contrast, is now largely a Fed-input variable — it matters insofar as it shifts the implied path of policy, and that path is already heavily priced.
The evidence that AI is the active variable, not a passive backdrop, is in the dispersion. On the selloff day, the companies that make the AI chips fell hardest — Nvidia down 9.3%, AMD down roughly 15%, the semiconductor index down more than 12% — while the software platforms that consume the chips fell far less, in the 2% to 4% range. That is not a broad macro de-risking. That is a repricing of the capital-expenditure assumptions inside the AI build-out.
"This is a true panic," said Jay Hatfield, chief executive at Infrastructure Capital Advisors. "This is 100% about the chip stocks and the potential slowdown in the AI trade."
Retail investors amplified the move, unloading $265 million of equities on the selloff day — their largest single-day exit since 2020. When the most rate-sensitive, highest-multiple corner of the market leads a decline of this size, the interpretation is not "the economy slowed." It is "the growth model is being re-underwritten."
The Mechanism: Capex, Credit, and the Second-Order Channel
The first-order story is simple: if AI capital expenditure slows, chip demand falls, earnings estimates fall, and multiples compress. The second-order channel is what Rosenberg's framing points toward, and it runs through credit.
Hyperscalers are forecast to spend more than $600 billion on capital expenditure in 2026, roughly a 36% increase over 2025, with about three-quarters of that — some $450 billion — tied directly to AI infrastructure. Data-center capital expenditure alone rose 57% in 2025 and is projected to surpass $1 trillion in 2026. That spending has to be financed. Goldman Sachs estimates that roughly one-third of hyperscaler capex will be debt-financed in 2026, implying around $250 billion of direct bond supply from the group this year, after $194 billion had already been issued by early November, compared with $108 billion across all of 2025.
Here is the transmission mechanism: equity investors question the AI growth outlook, so chip and data-center names reprice lower; the cost of capital for the build-out rises because credit investors demand wider spreads to fund the same projects; the economics of marginal AI projects deteriorate; capex plans get trimmed; and the earnings estimates that justified the equity multiples come down further. That loop is self-reinforcing, which is why a "growth outlook" concern can matter more than a single labor print.
The bond market is already hinting at the stress. Four of the five hyperscalers have underperformed the broader market on a spread basis this year, and Salesforce's A-rated notes recently priced above the U.S. BBB index — a sign that investors are starting to price AI disruption risk into credit, not just equity. When a technology company's A-rated debt trades wider than the average BBB credit, the market is no longer pricing default risk; it is pricing a business-model question.
The Fed Is Not the Story — But It Is the Backdrop
There is a reason the jobs report still matters, and it is the Fed. Market participants have been pricing a high probability of a December rate cut as labor data softened, with some estimates placing the implied odds of a quarter-point reduction near 87% after the week's data. A cut would lower the discount rate applied to every long-duration asset, and the AI complex is the longest-duration asset class in the market.
But this is precisely why Rosenberg's point cuts the other way. If the market is simultaneously pricing a Fed cut and selling the AI trade, it is making a distinction that matters: it is treating the rate cut as reactive rather than preventive. A preventive cut — insurance against a manageable slowdown — lifts growth stocks. A reactive cut — a signal that the economy is weakening faster than expected — drags earnings estimates down faster than the discount rate falls. The selloff in chips alongside the rally in Treasurys suggests investors are leaning toward the latter interpretation for the AI complex specifically, even as they price the cut.
In other words, the Fed can cut rates and the AI trade can still fall. That is the uncomfortable reality the "jobs don't matter" framing is really describing: monetary policy sets the tide, but sector-level earnings credibility sets which boats float.
Cyclical Shakeout or Structural Crack? The Call
This is the decision that determines the conclusion. Is the AI selloff a cyclical shakeout — a mean-reverting de-rating after a parabolic run — or the first crack in a structural regime?
The evidence points to a cyclical leg riding on top of a structural trend, and the two must be separated. Structurally, AI is not a speculative theme without cash flows: the hyperscalers funding the build-out are among the most cash-generative companies in market history, and many data-center financings carry amortization features, construction guarantees, and hyperscaler lease backstops. The secular direction of compute demand has not been repealed by a bad week.
Cyclically, however, the pace of the build-out is absolutely mean-reverting. Capex that grows 57% in a year cannot grow 57% the next; the marginal return on each new GPU cluster declines as the easiest use cases are saturated; and the equity market's pricing assumed a straight-line extrapolation that a single 5.4% day in the Nasdaq has now called into question. The cyclical leg is the valuation and the financing tempo; the structural leg is the direction of travel. Confusing the two is how investors turn a 20% correction into a permanent bear market in their heads — or, on the other side, talk themselves into holding a bubble because "AI is real." Both are wrong. AI is real, and this correction is also real.
History offers a useful comparison. The April 2025 selloff — driven by tariff fears — was a policy shock that reversed when the policy path clarified. This selloff is different: it is a valuation shock rooted in the sustainability of a multi-year capital cycle. Policy shocks mean-revert quickly. Valuation shocks mean-revert only when earnings catch up to price, and that takes quarters, not days.
The Strongest Counter-Thesis
The counter-argument is straightforward, and it is not weak: a genuinely soft labor market forces the Fed's hand regardless of what is happening in semiconductors. If October's payrolls come in near the expected 59,000, or worse, and unemployment ticks above 4.2%, the market will price a higher probability of a December rate cut. Lower rates mechanically lift long-duration assets, and the AI complex — the most rate-sensitive corner of the market — would be the first beneficiary. In that world, the "AI over jobs" framing is a luxury of a strong labor market, and the jobs report reclaims its throne.
There is also a credit-side counter. The AI build-out is not financed by speculative startups alone. A large share sits on hyperscaler balance sheets with operating cash flow that still covers capital expenditure, and project-level debt carries structural protections. Independent credit strategists note that many AI-related deals benefit from amortization and lease backstops — meaning the financing channel may be more resilient than the equity selloff implies, and the second-order loop described above may sputter rather than compound.
Both counters are credible. But they attack the timing of the repricing, not its direction. Even if the Fed cuts and credit holds, the question Rosenberg raises — how much AI growth is already in the price — remains the unresolved variable. A rate cut changes the discount rate; it does not answer whether $600 billion of annual capex earns its keep.
The Signal That Would Prove This Wrong
The "AI over jobs" thesis is wrong if the labor data reasserts itself as the dominant marginal driver. Specifically: if October nonfarm payrolls print at or above 150,000 with unemployment at or below 4.0% and average hourly earnings re-accelerate, the soft-landing narrative strengthens, the Fed-cut trade unwinds, and the market's attention snaps back to the macro tape. Conversely, if the 10-year Treasury yield — which fell to its lowest level since Sept. 9 during the selloff — breaks back above its post-selloff highs on growth repricing rather than inflation, the flight-to-quality reading is falsified, and the AI trade was never really about growth fears at all.
There is a third falsifying signal, and it is the cleanest: if the hyperscalers report uninterrupted capex guidance through the next earnings season while credit spreads tighten back toward their pre-selloff levels, the "overbuilt" thesis fails on its own terms. The market would have priced a slowdown that never arrived.
What Comes Next, by Time Horizon
The practical implication is a split by time horizon. In the short term, sentiment and positioning dominate: the Nasdaq is already in correction territory, more than 10% below its recent high, and the liquidation flow from retail investors shows the move has a momentum component that tends to overshoot in both directions. Expect volatility to remain elevated regardless of Friday's payroll print.
Over the medium term, fundamentals take over, and the key variable is not the jobs report but the capex cycle. The watch items are the hyperscalers' capital-expenditure guidance through the next earnings season, the spread behavior of AI-related credit, and the Anthropic initial public offering — the company filed its registration statement on Oct. 26 and applied to list on the Nasdaq under the ticker ANT — where investor discussions have already pushed the expected valuation down from the $1.5 trillion-to-$2 trillion range. That repricing is a cleaner read on AI demand than any payroll number, because it is the price of new capital, not a lagging survey of old jobs.
Over the long term, the structural call stands: AI compute demand is a multi-year trend, but the financing and valuation of that trend are cyclical and will revert. The beneficiaries of a sustained AI slowdown would be the rate-sensitive defensives and quality credit that benefit from lower growth and lower yields; the exposed are the chip suppliers, data-center developers, and leveraged software names whose models assume uninterrupted capex growth.
The base case: the jobs report prints soft, the Fed-cut trade persists, and the market chops between relief rallies and AI-driven selling until capex guidance provides a clearer anchor. The upside case: hyperscalers reaffirm full-year capex plans, credit spreads stabilize, and the selloff is marked as a 2026 buying opportunity. The downside case: capex guidance rolls over, the Anthropic pricing resets materially lower, and the second-order credit channel activates — in which case a 5.4% day in the Nasdaq is the first chapter, not the last.
The jobs report tells you where the economy was last month; the AI trade tells you what the market believes about the next five years. This week, investors are voting on the future and ignoring the rearview mirror — which is exactly when both deserve the most scrutiny.
Explore more exclusive insights at nextfin.ai.
