The Bottleneck Rotation Signal — Records Went to the Principals, Money Went to the Picks-and-Shovels, and Prosecutors Raided That Same Layer for Collusion

The Bottleneck Rotation — Records to the Principals, Money to the Picks-and-Shovels

Two mismatched scenes overlapped on Korea's chip tape this week. Samsung Electronics posted record Q2 2026 earnings and its stock fell sharply. TSMC beat consensus with $21.93B in net profit (up 77% YoY) and raised its 2026 capex guidance — and closed down anyway. Jim Cramer went on-air to flag the "butterfly effect" spillover to SK Hynix.

In the same week, Hanmi Semiconductor hit its daily upper price limit on record quarterly earnings. Chip inspection-equipment maker MIT cleared A-grade technical validation and formalized a H2 2026 KOSDAQ listing. Mekaro acquired Samil Tech for KRW 35.6 billion in an all-cash deal to widen its equipment footprint. Germany's ZEISS said it will open a semiconductor business base in Yongin. Two separate Korean news pieces put the divergence bluntly: Samsung and SK Hynix underperformed while semiconductor equipment and supply-chain ETFs rallied.

In one line: on the week the principals put up records, the money rotated into the picks-and-shovels layer.

Regulators Reached the Same Conclusion

The same week the market repriced the supplier layer, Korean prosecutors raided global semiconductor component suppliers to Samsung and SK Hynix on collusion allegations. Set aside how the case ends — the fact a collusion probe is even coherent is a market signal. A supplier can only try to collude when it has pricing power. If the principal is buyer-of-first-resort, suppliers don't need to collude, and couldn't if they tried. This week the enforcer effectively confirmed what the market was already pricing: pricing power has migrated to the supplier layer.

The "Bottleneck Beats Tech" Argument, Already Inside the Money Flow

Also this week, a Korean outlet ran an interview with a semiconductor-engineer-turned-investment influencer arguing that supply-chain bottleneck analysis outperforms product-cycle analysis. The thesis: don't read the principal's roadmap, read where the roadmap will jam. This week's tape shows that argument is already inside the money flow — not a forecast, an observation.

Why Did the Principals Sell Off?

Not because their stories deteriorated. Samsung's numbers were records. TSMC's 2026 capex went up. ASML posted Q2 revenue of €9.33B, above guidance, with Korea taking 43% of equipment shipments. That 43% is the tell. It measures how hard Korean principals are spending, and the winner of a capex cycle is always the layer receiving the capex, not the layer writing the checks. Principal margin gets compressed by the capex bill; supplier margin gets expanded by the capex bid. The old rule reasserted itself.