More than Moore

Accelerating Moore’s Law should accelerate economic growth because falling computing costs generate new demand, higher utilization and rising corporate productivity rather than reducing total spending.

• Intel’s 1985 decision to exit memory chips and focus on microprocessors preceded revenue growth of approximately 29% per year from 1986–1992 and 34% per year from 1992–1997, before growth slowed to 1.4% from 1997–2001 as the franchise matured.
• The historical mechanism is Jevons’ Paradox: cheaper computing expanded usage from offices to homes and smartphones, so falling unit costs coexisted with rising revenue for more than a decade.
• AI computing capacity is now doubling in roughly 10 months instead of 24, while installed AI computing is growing by 2–3 times per year versus about 58% annual growth in total computing capacity during the earlier Moore’s Law era.
• Global semiconductor revenue reached $796 billion in 2025, up 26%, and the industry forecasts $1.5 trillion in 2026, a 90% increase; three-year-old AI chips reportedly rent for more than they did the previous autumn despite record shipments of newer chips.
• Machine intelligence costs are said to be falling by roughly 90% per year, while human cognitive labor becomes more expensive; cash profit per employee at the 900 largest US public companies exceeds $120,000, nearly double its level a decade ago.

Takeaway: Investors should expect AI-driven efficiency to appear more clearly in corporate profits and profit per employee than in GDP, although weak productivity statistics could still lead skeptics to underestimate the boom.