WealthPast logo WealthPast
People
Blog
Calculators
Proverbs & SayingsOn This DayHolidaysAboutContact

Gordon Moore’s Cost Curve: How a 1965 Forecast Shaped Semiconductor Investment

What Gordon Moore actually predicted in 1965, why he revised it in 1975, and how a cost curve became an investment target rather than a law of nature.

Written by Ibraham

Gordon Moore — historical photographic portrait
Gordon Moore — historical photographic portrait. Photo: Staff photographer. Wikimedia Commons source · CC BY-SA 3.0. No AI alteration.

Gordon Moore’s 1965 forecast became famous because it connected technical progress to economics. Writing when integrated circuits were still young, Moore observed that the number of components on a chip had been rising while the relative cost per component fell. He projected that the pattern would continue for about a decade. In 1975, he revised the expected doubling interval to roughly two years. The forecast was not a law of physics or a promise of stock returns; it became a planning target that helped coordinate research, manufacturing and customer expectations.

What Moore actually forecast

The Gordon and Betty Moore Foundation dates the original article to April 19, 1965. Moore was then leading research at Fairchild Semiconductor. His argument joined density and cost: putting more components on an integrated circuit could spread functions across a smaller, more economical device. That gave businesses a reason to invest not only in faster computing but in computing that could reach more applications at lower relative cost.

The popular phrase “computing power doubles every 18 months” compresses several versions and can obscure the source. The foundation’s historical account says the 1965 projection involved annual doubling, while the 1975 revision anticipated roughly two years. Treating one interval as timeless would contradict Moore’s own revision.

A forecast can coordinate capital

Semiconductor progress is expensive. New process technology requires laboratories, fabrication equipment, materials, design tools and trained teams. When suppliers and customers expect a recurring improvement curve, they can plan compatible investments. Equipment companies develop the next tools, chipmakers design the next process, and computer producers plan products that may use the resulting capacity.

This is why Moore’s Law mattered financially even though it did not compel nature. It acted as a shared benchmark. Missing the benchmark could weaken a producer’s position; attempting to meet it could demand enormous capital before the revenue arrived. The modern cost structure is visible in ASML’s equipment and service model, where advanced machines and an installed base connect long research cycles to recurring customer support.

Intel converted the observation into an operating challenge

Moore and Robert Noyce founded Intel in July 1968, after their work at Fairchild. Intel first emphasized memory and later centered its strategy on microprocessors. The observation did not choose those products. Managers still had to allocate capital, develop manufacturing processes and find customers. A cost curve can describe an opportunity; it cannot decide which company captures it.

The distinction also prevents a simplistic founder story. Moore contributed scientific and managerial judgment, while engineers, equipment makers and customers across the industry produced the actual sequence of improvements. Intel’s change of focus, explored in the Andrew Grove DRAM case, shows how a technology trend still requires difficult portfolio choices.

Lower cost per function is not the same as lower total spending

More economical transistors can expand demand so dramatically that total industry investment rises. A manufacturer may spend more on a new fabrication process even while the cost of a particular computing function falls. Consumers may also buy more devices and services rather than pocket the entire efficiency gain.

That pattern resembles a scale economy, but it is not automatic profit. Yield problems, delayed products, weak demand or a rival architecture can destroy returns. The NVIDIA supply-chain case illustrates another boundary: design economics and manufacturing economics can sit in different companies, so a successful chip still depends on partners and capacity.

Philanthropy separated wealth from operating companies

Gordon and Betty Moore created their foundation in 2000 to support science, environmental conservation, patient care and the San Francisco Bay Area. In 2001, Caltech and the foundation announced gifts totaling $600 million: $300 million from the couple over five years and $300 million in foundation grants over ten years. Those are documented commitments, not a complete measure of Moore’s personal fortune.

Foundation assets, Intel revenue and personal net worth are three different quantities. Combining them would overstate what one person owned and confuse operating business with philanthropic capital. The same discipline is essential when reading any list of historic fortunes.

The durable lesson

A good forecast can make a market more legible, but it can also become a dangerous assumption when treated as destiny. Investors and managers should ask which engineering steps, capital commitments and customer uses are required for the trend to continue. Moore’s most useful legacy is therefore not a magic growth formula. It is a dated, testable observation that helped an industry organize work.

Read the full Gordon Moore profile for the sourced timeline and limits around wealth estimates.

Sources

Share with