AVGO 320B Wipeout|AI Chip Bar Too High?

Priced for Perfection

AVGO beat on every number that Wall Street asked for, and the stock fell 14% anyway — erasing $320 billion in a single session.

Broadcom reported revenue of $22.19 billion, up 48% year over year, beating estimates. AI semiconductor revenue hit $10.8 billion, up 143% from a year ago. The CEO described demand as "simply insatiable." The forward guidance came in above consensus.

None of it was enough.

The sell-off was not about the quarter. It was about the positioning structure that had built up around the stock — AVGO had run from $300 to nearly $480, and the options market showed extreme upside call skew in the weeks before earnings. That is the signature of a stock priced for a number that does not exist yet.

Broadcom's AI chip forecast for next quarter was $16 billion. Wall Street had quietly penciled in $57 billion for the full fiscal year 2027; the company reiterated its existing target of "in excess of $100 billion" without raising it. That reiteration, in a stock that had already priced the raise, produced the reaction.

The sell-off rippled outward immediately. Micron fell 7%, Arm Holdings 5%, AMD 4%, Intel 2%. The contagion was not random — these are the names that had ridden the same positioning logic: infinite compute demand justifies infinite multiple expansion. What Broadcom's earnings session revealed is that infinite demand does not automatically mean infinite price appreciation when expectations are already embedded.

Jefferies flagged a structural concern the surface read misses: Broadcom's largest hyperscaler customer is reportedly shifting some inferencing workloads to MediaTek. If that shift accelerates, it creates a competitive pressure at the custom ASIC layer that the $100 billion target does not account for. The market priced that risk into the stock before management could address it.

The margin picture added a second layer. Gross margin fell 230 basis points year over year to 77.1%, and the company guided for further compression to approximately 74% next quarter. Custom AI chips — the ASICs driving the revenue explosion — carry structurally lower margins than legacy infrastructure. As AI revenue becomes a larger share of the mix, the earnings quality of that growth degrades. That is the mechanism the earnings call confirmed in plain language.

What remains open is whether the $320 billion erasure represents a permanent repricing of AI chip multiples or a clearing of a positioning overhang that leaves the underlying demand intact. The Broadcom CEO's demand language, the forward bookings that tripled shipments at $30 billion, and the 39-event historical pattern — where the stock recovers to new highs within three months in nearly 90% of cases — each argue for the latter. But the MediaTek shift and the margin trajectory argue for something that institutional holders cannot simply wait through.

The Capital That Left

The money that exited chip names on Thursday did not disappear — it moved directly into the one sector that had been priced for a different story entirely.

UnitedHealth rose 5%, Humana 6%, Cigna 4%. The Health Care Select Sector SPDR fund gained 3% on the session — on a day when the Nasdaq was down and the Dow hit a record high. That divergence is the session's most important structural signal.

Bank of America upgraded UnitedHealth to Buy from Neutral and raised its target to $450, citing softening medical cost trends and a Q2 setup that now looks favorable. Morgan Stanley lifted its target to $453 and maintained Overweight, estimating that AI-driven efficiency gains could deliver 45% average earnings-per-share upside for managed care organizations as automation scales. Truist went to $440. Three independent upgrade actions in a single session, all anchored on the same variable: medical loss ratio improvement.

UnitedHealth's Q1 medical cost ratio improved 90 basis points to 84%. Humana's insurance benefit ratio ran at 89%, slightly favorable to guidance, while individual Medicare Advantage membership grew 22% year to date. The cost trend that damaged the sector through 2025 — elevated utilization driving loss ratios above plan — is visibly decelerating.

The positioning logic is the inverse of what drove chip names. Health insurers spent most of 2025 pricing in continued cost deterioration, CEO turnover, and regulatory scrutiny. UnitedHealth entered Thursday near the bottom of Dow performance for the year. The analyst upgrade cluster signals that institutional holders who had been underweight the sector are now reestablishing positions against a lower cost base — not against a peak.

The difference matters for what it reveals about the day's broader capital movement. Institutions rotating out of chip names at maximum expectation were not moving to cash. They moved to the domestic sector where the bar had been set at its lowest point and where the first credible evidence of improvement was arriving. That is a defined positioning trade, not a macro call.

What the managed care move does not resolve is whether the UnitedHealth recovery can sustain when Q2 results arrive. The medical cost ratio improvement in Q1 preceded any AI efficiency contribution — those gains are still forward estimates, not reported results. Morgan Stanley's 45% EPS upside from AI efficiency is a 2027 and 2028 projection. The stock is already repricing toward it. If Q2 cost trends reverse, the multiple compression would be as sudden as the chip sector's was on Thursday.

The Verification Window

Two domestic repricing events on the same day, opposite sign, driven by the same underlying variable: whether earnings delivery matches the expectations already embedded in price.

Broadcom's session established that a stock growing AI revenue 143% year over year can still produce a $320 billion wipeout when the positioning structure is more aggressive than the guidance it receives. The sell-off was not a fundamental deterioration call — 51 of 54 analysts still rate the stock Buy — but a collision between positioning overhang and a reiterated rather than raised target.

The UnitedHealth session established the mirror condition: a sector priced for continued deterioration can absorb multiple simultaneous upgrades and a 5% gain in a single session when the positioning structure entering the event is light rather than crowded.

The question these two outcomes together raise is not which sector is correct. It is whether the positioning gap between them — AI chip names at maximum expectation, managed care near minimum — is now closing. If it is, the Dow hitting a record while the Nasdaq lags is the beginning of a rotation, not a one-day anomaly.

The near-term verification point for the chip side is Broadcom's Q3 guidance print and whether the MediaTek shift appears in bookings data. The $16 billion AI chip revenue forecast for next quarter needs to arrive without further margin erosion for the stock to stabilize at current levels.

The verification point for managed care is UnitedHealth's Q2 earnings, where the first real test of whether utilization softness persists will surface. The Bank of America upgrade target of $450 implies roughly 14% upside from current levels — achievable only if the medical cost ratio holds at or below 84%.

Both verification windows open within the next 60 days. The positioning structure entering those prints — chips still crowded even after Thursday's clearing, health insurers still underweight relative to historical allocation norms — means the stakes of each print are asymmetric. A chip beat may not produce the same upside the positioning had once enabled. A health care miss, however, would land against a recently rebuilt position structure, and the exit would be faster than the entry.