SK Hynix · one lens
Hamilton Helmer on SK Hynix
SK Hynix’s repeated HBM qualification lead suggests a process advantage, but Samsung’s catch-up and new NAND entrants leave the barrier unproven.
SK Hynix presents a Benefit candidate more readily than a Barrier. SK Hynix began 12-layer HBM3E production in September 2024, four months before Samsung secured Nvidia clearance in January 2025 for an 8-layer HBM3E part used in less powerful, China-tailored processors. That comparison supports a product-and-qualification lead: 12 layers for SK Hynix versus 8 layers for Samsung, with SK Hynix reaching production four months earlier. Neither company discloses the resulting price premium, cost saving, or free-cash-flow contribution, leaving Helmer’s Superior and Significant Benefit tests open.
Process Power is the strongest supported Power candidate, but the claim should remain confined to HBM3E. SK Hynix’s earlier production and Samsung’s later, narrower clearance suggest that engineering and customer-qualification routines mattered beyond access to equipment and capital. Samsung nevertheless qualified an HBM3E alternative within four months of SK Hynix’s production milestone. One product cycle cannot establish the long-lived, hard-to-copy learning Helmer requires, so Sustainability remains contested rather than demonstrated.
Possible Scale Economies, Switching Costs, and Cornered Resource are weaker. Bloomberg’s Indiana packaging report and Reuters’ U.S. manufacturing report establish expansion activity, but neither provides a unit-cost comparison showing that higher SK Hynix volume disadvantages a smaller rival. Nvidia’s qualification requirements and advance allocations could make supplier changes costly, yet Samsung’s clearance preserves another qualified route. Nvidia’s HBM4 allocation, for the generation succeeding HBM3E, favored SK Hynix, but no cited source identifies an exclusive asset or supply right unavailable to competitors at any price.
CXMT’s planned NAND entry sharpens the replication test outside HBM. Reuters reports intended competition with Samsung, SK Hynix, and YMTC, not successful replication of their capabilities.
TEXXR indexed 49 SK Hynix articles from the start of 2026’s third quarter through September 18, versus 10.6 articles per quarter across the preceding eight quarters. Seoul-listed shares rose 6.42%, or KRW 112,000, from KRW 1,745,000 on September 17 to KRW 1,857,000 on September 18. Those baselines indicate increased attention and a one-day market move, not stronger Power. SK Hynix’s HBM3E process lead remains plausible, while its economic magnitude and non-replicability remain unproven.
The seven Powers, one by one
| Power | Status | Why |
|---|---|---|
| Scale Economies | contested | capacity projects do not establish lower unit costs or a durable volume gap |
| Network Economies | absent | not argued in the record |
| Counter Positioning | absent | not argued in the record |
| Switching Costs | contested | Samsung’s eventual qualification shows that customer qualification friction is surmountable |
| Branding | absent | not argued in the record |
| Cornered Resource | contested | preferred Nvidia allocation is not shown to be exclusive or unavailable to rivals |
| Process Power | contested | repeated HBM3E and HBM4 execution leads coexist with Samsung’s eventual catch-up |
An editorial application of Hamilton Helmer's published framework — not their actual view, affiliation, or advice.
Where this lens comes from
Hamilton Helmer's framework is set out at the lens page, drawn from 7 Powers. This page applies it to SK Hynix and nothing else — the company's full record is in the dossier.