From 4% to 22%: Why the Semiconductor Correction Is Technical — and Why 2027 Is the Opportunity of the Decade

Larry Fink, the chief executive of BlackRock — the largest asset manager in the world — described the current moment as the beginning of the greatest expansion of global financial markets in history. The mechanism he identifies is the tokenization of assets: a future in which everything that has value trades continuously, 24 hours a day, seven days a week, on global digital infrastructure. The implication for the underlying technology that makes this possible is straightforward.

José Luis Cava builds on this macro foundation to make a specific case about the semiconductor sector — one that distinguishes between the technical noise of the current correction and the structural signal beneath it.

The Structural Shift in Plain Numbers

In the decade from 2000 to 2010, semiconductor companies represented approximately 4% of the S&P 500 by market capitalization. Today that figure stands at 22%.

This is not a bubble in the traditional sense. A bubble is a temporary mispricing that reverts when the underlying demand proves imaginary. What the shift from 4% to 22% reflects is a genuine recomposition of the economy — a world in which the production of intelligence, automation, and connectivity has become the primary source of economic value, displacing the manufacturing and retail-heavy composition of previous decades.

Between late March and early July 2026, the Philadelphia Semiconductor Index rose 115%. That is a vertical move. Vertical moves correct. The correction currently underway is not evidence that the thesis was wrong. It is the normal consequence of a move that went too far, too fast, and attracted the wrong participants at the wrong prices.

Why the Correction Is Mechanical, Not Fundamental

Understanding who drove the final leg of the semiconductor rally explains why the correction is happening and how it will end.

The dominant buyers in the most recent phase were two groups that Cava identifies as structurally weak hands.

The first group is Korean retail investors who purchased semiconductor exposure on leverage. Retail investors using borrowed money to buy assets that have already tripled are the definition of late-cycle momentum participants. Their position sizes are larger than their conviction, because leverage amplifies both returns and losses. When the market pauses or reverses, they are forced to sell not because their thesis changed but because their margin accounts force them out. The selling pressure from this group is mechanical — it is not a signal about the fundamental value of semiconductors.

The second group is professional fund managers who entered the sector late. Fewer than 0.25% of active fund managers have managed to beat the S&P 500 this year. The majority of those who eventually bought semiconductor exposure did so after the move was largely complete, driven by the career risk of underperforming a benchmark that was being driven by names they did not own. When results disappoint relative to elevated expectations, these managers are also forced to reduce exposure. Their selling is not a fundamental judgment — it is a portfolio management response to performance pressure.

Both groups are being expelled from the market. When they are gone, the supply of forced sellers is exhausted. The buyers who remain are the ones who entered with conviction and at lower prices. That is when the next leg begins.

The Robotics Equation: 50 Times More Than a Smartphone

The AI data center buildout — the demand that drove semiconductor margins to historic highs and Micron's gross margins from 39% to 85% — is the first wave of structural demand.

The second wave is robotics, and its scale is different in kind rather than in degree.

An iPhone contains semiconductors. A humanoid robot consumes approximately 50 times more semiconductor content than an iPhone, across processors, memory, sensors, motor controllers, and communication chips. The iPhone was a device produced in quantities of hundreds of millions annually. The humanoid robot market, currently in its early commercial phase, is on a trajectory toward similar or greater unit volumes within this decade.

When the combination of AI software capability, falling manufacturing costs, and improving battery technology brings humanoid robots into industrial and eventually consumer deployment at scale, the demand for semiconductors does not increase by a multiple of the AI infrastructure wave. It increases by a multiple of the entire consumer electronics era.

Cava describes 2027 as a unique opportunity in decades precisely because this transition from prototype to mass production is beginning now. The investors who are being expelled from the market by the current correction are exiting at the moment when the structural case is becoming clearest.

The Entry Zones

For investors who accept this structural thesis and are waiting to add exposure at technically sound levels, Cava identifies specific zones based on prior support structures.

NVIDIA (NVDA): The zone between $164 and $170 corresponds to the March and April 2026 lows — levels at which the market previously absorbed selling pressure and reversed. A return to this zone would represent a correction of approximately 25-30% from the highs reached during the summer rally, sufficient to exhaust the leveraged and momentum-driven sellers.

AMD: The level of $409 is identified as the key support zone where a sustainable entry can be established.

Micron (MU): The primary target is approximately $712, with close attention to the May 2026 lows near $652. Micron's margin expansion — from 39% to 85% — reflects the pricing power that comes from genuine supply shortage. The question is whether that pricing power persists long enough to justify current valuations. At $712 and especially at $652, the risk-reward becomes more favorable.

SMH ETF (UCITS equivalent): For European investors seeking sector exposure without individual stock selection, the zone between $469 and $477 corresponds to May 2026 lows. This level represents the last major area of support before the vertical move began, and a return to it would reset the risk-reward profile for the sector as a whole.

The Longer Frame

The semiconductor sector's move from 4% to 22% of the S&P 500 is not complete. If the tokenization of assets that Fink describes becomes a reality — if global financial infrastructure becomes fully digital and continuously operating — the computational requirements for that infrastructure are orders of magnitude larger than what exists today.

The current correction is the market doing what it always does: removing the weak hands, resetting sentiment, and creating the entry point that the investors who were waiting for it have been anticipating. Historically, those who bought during corrections within structural bull markets — not at the absolute bottom, but somewhere in the correction zone — have been the beneficiaries.

The vertical move needed to correct. The structural change that drove it does not.

---

Analysis based on José Luis Cava's market commentary from July 16, 2026, incorporating perspectives from Larry Fink's BlackRock outlook. This post is for informational and educational purposes only and does not constitute investment advice.