172K Jobs Hit AI Stocks|Rate Hike Odds Surge

Jobs vs. AI Trade

The Nasdaq fell 4% on Friday — its worst single-day drop in six years — and the surface explanation misses the actual mechanics entirely. Consensus expected 80,000 May payroll additions. The Bureau of Labor Statistics printed 172,000, more than double the forecast, with unemployment unchanged at 4.3%. The surface read was simple: strong jobs data means no Fed cuts. But that framing understates why chip stocks specifically took the most damage.

The positioning context matters here. Heading into Friday, AI-infrastructure equities — AVGO, MRVL, MU, NVDA — had already posted triple-digit gains since January. The investor class most exposed was not retail. Institutional long/short funds had built concentrated positions in these names specifically because falling rates were expected to expand the valuation multiples on high-CapEx growth stocks. When 10-year Treasury yields spiked to 4.69% on the jobs print — their fastest single-day move since the October selloff — the rate assumption inside those positions broke simultaneously. Leveraged institutional longs unwound first, and passive ETF outflows in semiconductor funds amplified the move. The S&P 500 fell 2.6%; the Nasdaq Composite fell 4%. The gap is the rate-sensitivity differential between the two indices — not two separate stories.

New Fed Chair Kevin Warsh had already signaled higher-for-longer in May. Friday's print makes a 2026 rate hike materially plausible for the first time since the Iran war began. Prediction markets moved the odds of at least one 2026 hike from 14% to 31% within hours of the release. The mortgage market reacted immediately — rates jumped the same afternoon. What matters for the AI trade is not whether a hike happens, but whether the institutional consensus on the rate path has permanently shifted. That consensus is now contested, and contested rate expectations are what forced the position unwind. The rate signal alone explains the broad selloff. It does not explain why the chip sector fell twice as hard as the rest of Nasdaq.

Broadcom's Guidance Gap

The chip sector's excess damage relative to the broader market traces to a second trigger that had already arrived 24 hours before the jobs number. Broadcom reported Q2 FY2026 earnings Thursday evening with AI semiconductor revenue of $10.8 billion — up 143% year-over-year. The stock fell 13%. The contradiction here is not a paradox; it is a position pressure signal.

AVGO had risen 38% year-to-date before the print. Options positioning data showed extreme upside call skew — FOMO-driven retail and levered institutional accumulation of upside contracts at strike prices well above spot. For that positioning to pay, Broadcom needed to raise its FY2027 AI revenue target above $100 billion. Management reaffirmed $100 billion without upward revision. The market had priced in a raise; it did not get one. Institutional holders who had bought the run-up on the expectation of guidance acceleration began reducing exposure Thursday evening. Friday's jobs print then arrived into a market already in de-risking mode, and the selling cascaded into the broader AI-chip complex. Marvell fell 6%, below the $300 level where it had tripled year-to-date. Micron fell 5% and broke below $1,000. Neither company had bad news Friday — both were caught in the contagion from the Broadcom repositioning combined with the rate repricing.

The Broadcom move is specifically interpretable as a valuation regime signal rather than a fundamental deterioration. AI demand is not contracting — $10.8 billion in a single quarter, with $30 billion in pending customer orders, confirms the opposite. What changed is the multiple the market is willing to assign when rate assumptions shift. A chip stock priced at 80x forward earnings on a declining-rate assumption reprices quickly when yields spike 15 basis points in a single session. The chip selloff was not about AI spending collapsing; it was about rate-multiplied valuation compression hitting names that had run fastest. The question that compression opens is who funds the AI infrastructure build-out if the stocks that have been financing it through appreciation are no longer appreciating.

Who Pays for AI?

The answer to that question arrived Wednesday and is the most structurally consequential development of the week. Alphabet priced an $84.75 billion equity offering — the largest secondary stock offering in financial market history, larger than the Saudi Aramco IPO. Berkshire Hathaway anchored $10 billion of the raise. The stated purpose: expansion of AI compute infrastructure to meet demand that exceeds available supply. The market's immediate response was to sell Alphabet stock 4%.

That reaction is the frame disturbance. Alphabet is a cash-generative business with $126 billion on its balance sheet. It did not need this equity issuance to survive. It issued equity because the scale of AI infrastructure spending — Alphabet has guided $180-190 billion in capex for 2026 alone — has exceeded what even the most profitable company in the world can comfortably self-fund through operations. Berkshire's participation is the signal inside the signal: the new management at Berkshire, under Greg Abel, made a $10 billion commitment to an AI infrastructure bet. That is a generational portfolio reallocation from a firm historically averse to technology. Meta is reportedly preparing a similar equity raise of tens of billions of dollars. The pattern here is not isolated. Hyperscalers are collectively signaling that AI compute spending is now large enough to require external capital markets to absorb it.

For holders of AVGO, MRVL, and MU, the dilution question now runs in both directions. Hyperscaler customers issuing equity to fund chip purchases means the chip demand is real and sustained — Broadcom's $100 billion FY2027 target exists because Google, Meta, Anthropic, and OpenAI have already committed purchase orders. But hyperscaler equity issuance also means their stock prices face dilution pressure simultaneously with the chip stocks they buy from. The AI infrastructure trade was built on the assumption that compounding returns flowed upward through the supply chain — from data center spending into chip revenue into chip stock appreciation. Friday's combined signal — rate repricing plus hyperscaler dilution — is the first session where that compounding assumption was explicitly tested against a competing pressure. The verification point for whether this was a one-session repricing or the beginning of a regime shift is the June Fed meeting and Warsh's next public commentary. If the Broadcom order book for Q3 confirms that AI chip demand held through the repositioning — that figure prints in late August — the selloff reads as valuation compression, not demand destruction. If institutional net selling in semiconductor ETFs persists through the coming two weeks, the rate assumption has permanently shifted and the positioning structure has not found a new floor.