Rock's Law is Moore's Law's economic inverse: as chips get denser, the cost of building the fabs to make them rises exponentially

Rock's Law — sometimes called Moore's second law, and attributed to Arthur Rock, an early Intel investor — is the cost counterpart to the density gains Moore's Law describes. An ICLE white paper states it directly: "Rock's Law is the darker side of Moore's Law. It holds that, as chips get denser, the cost of making them rises exponentially. Advanced fabs now cost $10–20 billion, and a single high NA EUV scanner... runs north of $400 million." The commonly cited form of the law holds that the capital cost of a state-of-the-art fabrication plant roughly doubles every four years.

The economic shape is a fixed-versus-marginal-cost asymmetry: an enormous one-time capital outlay (fab, mask set, scanner) amortized over a large run of comparatively cheap wafers. That is the same structure the vault records for AI compute in claim-training-inference-compute-asymmetry-mechanism — a large one-time training cost amortized over many cheap inferences — which is the analogy that generated this note. See moc-inference-economics.

Because each process node demands a fresh, larger capital commitment, the ICLE authors frame semiconductor rivalry as "competition for the market" (node by node) rather than steady-state "competition in the market" — a recurring, high-stakes race that only a shrinking number of firms can afford to enter. The paper reports that TSMC's capital expenditure runs on the order of 30–50% of revenue, peaking above 50% in 2021 (recorded here as the capture's paraphrase of the ICLE paper; the exact ICLE wording was not captured, so this specific figure is [unverified-quant — needs primary]).