Two events are arriving at the same moment, and the market is treating them as one. The first is a repricing of the semiconductor complex that erased more than a trillion dollars of value in a single Friday session. The second is the largest concentration of equity supply in the history of public markets, three listings totaling north of $200 billion in float, all filed within three weeks of each other. The first looks like a valuation accident triggered by one cautious guide. It is not. It is the leading edge of the second.
The Semiconductor Repricing Was Not About Broadcom
The proximate cause was clean enough. Broadcom beat on revenue and earnings for fiscal Q2, then guided Q3 AI chip sales to $16 billion against a $17.2 billion consensus and declined to raise its full-year AI forecast. The stock fell 14% on Thursday. On Friday the contagion went systemic: the SOX dropped roughly 10%, Marvell fell about 16%, Micron about 13%, AMD and Intel around 11%, Nvidia near 6%. The Nasdaq closed down 4.18%, its worst day since the tariff panic of April 2025.
A single in-line guide does not vaporize a trillion dollars unless the group was priced for something it could no longer deliver. The chip complex spent the last two years carrying multiples that assumed monopoly economics and uninterrupted hyperscaler capex. Broadcom’s refusal to raise was not a miss. It was a refusal to validate the terminal-value story the cohort had been borrowing against. Once one name declined to extend the narrative, the entire group derated together, because they had all been underwritten on the same thesis. This is what a cohort derating looks like: the names do not fall on their own news, they fall on each other’s.
The $200 Billion Drain
The repricing is not happening in a vacuum. SpaceX filed its S-1 on May 20, targeting a valuation near $1.75 trillion and a raise of up to $75 billion, the largest offering ever attempted. OpenAI filed confidentially on May 22, seeking $852 billion to $1 trillion and a September listing, with at least $60 billion in proceeds. Anthropic is structuring a late-2026 listing at a reported valuation above $900 billion. Together these three pull more than $200 billion of free float out of the public market in roughly two quarters.
Index funds do not conjure new capital to absorb new supply. When a name enters a benchmark, every passive vehicle tracking that benchmark must buy it, and to buy it they must sell something already held. The marginal seller funding the largest listings in history will be the largest existing positions in the same indices. Those positions are the mega-cap technology names, and the most over-owned, most index-concentrated sub-group inside them is semiconductors. The drain does not threaten the chip complex incidentally. It targets it structurally.
The Mechanics Decide the Timing
The rebalancing rules are the part most investors are not watching, and they are the part that matters. The Nasdaq-100 admits new constituents under a Fast Entry rule and does not adjust for free float, which means SpaceX would become a material index position the moment it qualifies, forcing immediate one-sided buying that must be funded by immediate one-sided selling elsewhere. S&P Dow Jones has been consulting on cutting its seasoning period to six months and waiving profitability requirements for megacaps, changes that could compress these forced rebalancing events into the back half of 2026 rather than spreading them across years.
The relevant stress is not the share of total market cap displaced. It is the share of each existing constituent’s daily trading volume hit with forced, one-directional selling on a single rebalance session. Concentrated index names with thin float relative to their weight absorb that pressure worst. The semiconductor leaders are precisely those names.
The New Gravity Centers
The deeper shift is intra-sector, and it is permanent in a way the Friday selloff is not. The incoming cohort does not add weight to the technology sector evenly. It redirects weight within technology, away from the infrastructure and hardware layer that has dominated index concentration, toward the platform and application layer. The new constituents are AI platforms, satellite-connectivity operators, and compute-application businesses, not chip fabricators. MSCI’s own modeling of the listings shows the largest weight declines falling on the existing hardware leaders as the new entrants claim their share.
The center of gravity for a global equity allocation is moving from the picks-and-shovels layer to the platform layer. For the chip names this is not a sentiment problem that resolves on the next earnings beat. It is a slow, mechanical reallocation of the benchmark itself. The market that priced Nvidia and Broadcom as the indispensable core of the AI trade is being rebuilt around a different core, and the rebuild is being executed by passive flows that do not care about valuation.
The Macro Overlay Removes the Cushion
None of this would bite as hard with rates falling. They are not. May payrolls came in at 172,000 against expectations, the unemployment rate held at 4.3%, and the ten-year yield pushed to 4.54% as rate-cut odds collapsed and rate-hike odds appeared. Strong labor data is now read as a tightening signal, which means the discount rate working against long-duration growth equity is rising at the exact moment the supply of growth equity is exploding.
Higher-for-longer does two things to this setup simultaneously. It compresses the multiples on the unprofitable platform names being brought to market, and it raises the cost of holding the speculative tail, the quantum and neocloud names that ran on liquidity rather than earnings. The macro does not cause the rotation. It removes the cushion that would have let the rotation happen gently.
The Most Probable Scenarios
Base case, and the highest-probability path: an orderly but punishing rotation. The mega-IPOs price at or below their target ranges, passive flows do the forced selling in discrete rebalance events, and the semiconductor complex grinds through a multiple reset rather than a crash. Chips stabilize at lower forward multiples that reflect cyclical hardware economics instead of monopoly economics. The platform layer becomes the new index anchor. Yields stay sticky near 4.5% and cap any quick recovery in long-duration names. This is the path the June 5 session is already pricing.
Bull case, lower probability: the labor data softens into the fall, the Fed signals patience, yields retreat below 4.3%, and the IPO cohort is absorbed into a falling-rate tape that lets both the new listings and the derated chip names re-rate together. In this scenario the drain is real but the incoming liquidity from a dovish pivot offsets it, and the rotation becomes a broadening rather than a zero-sum transfer. This requires the macro to cooperate, and the macro has not been cooperating.
Bear case, the tail that is no longer remote: the listings come to a market that cannot absorb $200 billion without indiscriminate selling. SpaceX’s Fast Entry inclusion forces a violent Nasdaq-100 rebalance, the S&P seasoning change accelerates rather than spaces the pressure, and a hot inflation print converts higher-for-longer into an actual hike. In that world the semiconductor derating is not a rotation, it is the first leg of a broader de-grossing, and the speculative tail, including the quantum names, is the first thing sold and the last thing bought back.
The chip selloff and the IPO calendar are not two stories. They are the same capital being asked to occupy two places at once, and it cannot. The market is deciding which layer of the technology stack it will pay a premium to own, and it has started by taking the premium away from the names that built the infrastructure for the very companies now coming to take their seats. The picks and shovels are being repriced to fund the gold rush they made possible.
