In June 1963, a 30-year-old engineer at Fairchild Semiconductor filed a patent that sounds like a yawn: US Patent 3,356,858, "Low Stand-By Power Complementary Field Effect Circuitry." Granted in December 1967, it was assigned to Fairchild for, by some accounts, a $1 token payment. The engineer was Frank Wanlass, and what he had figured out was how to make a logic circuit that consumed essentially no power when it wasn't switching.

Today we call it CMOS — Complementary Metal-Oxide-Semiconductor. It is the architecture inside the chip in your phone, your laptop, your car, your watch, your earbuds, your refrigerator, and the data center serving this page.

What he actually invented

The MOSFET had already been demonstrated by Atalla and Kahng at Bell Labs in 1960. A MOSFET comes in two flavors: n-channel (conducts when the gate is high) and p-channel (conducts when the gate is low). Existing logic families used one type at a time, with resistors as "pull-up" loads. The resistor was always burning power — even when the gate was just sitting there holding a 0 or a 1.

Wanlass's insight was to stack one of each. A p-channel transistor on top, an n-channel on the bottom, gates tied together. For any steady input, exactly one is off — and an "off" MOSFET conducts almost nothing. The circuit only draws current during the brief instant it switches states. Static power consumption: nanowatts instead of milliwatts. A million-fold improvement.

He built test chips at Fairchild that ran on power so low you could light them with a hand-cranked generator. At a 1963 IEEE conference, he reportedly demonstrated CMOS gates drawing nanoamps when audiences expected milliamps.

Why nobody cared (at first)

CMOS was slow. Bipolar TTL logic ran circles around it in speed, and in 1963 the market wanted speed. CMOS was also harder to fabricate — two transistor types meant extra masking steps. For a decade it was a niche technology used mostly in wristwatches (RCA's 1970s CD4000 series), pacemakers, and military gear where battery life mattered more than gigahertz.

The flip came in the 1980s. As transistors shrank, bipolar's power dissipation became unmanageable — you literally could not cool a million-transistor bipolar chip. CMOS's near-zero static power suddenly looked like the only path forward. Intel switched the 80386 to CMOS in 1985. By 1990, CMOS had eaten the world.

Why this patent quietly runs civilization

• Every modern CPU and GPU is CMOS. Apple Silicon, NVIDIA H100s, AMD EPYC — all complementary pairs, just shrunk to 3nm and stacked billions deep.
• Every smartphone exists because of it. A bipolar phone SoC would need a car battery and a fan. Wanlass made the pocket computer thermally possible.
• "Dennard scaling" assumed CMOS. Robert Dennard's 1974 scaling rules — the engine of Moore's Law for 30 years — only work because CMOS power scales with frequency, not just sits there burning.
• Image sensors too. The CMOS sensor in every phone camera is the same architecture repurposed for photodetection, displacing CCDs around 2010.

Wanlass moved to General Microelectronics, then to a string of startups, and is credited with founding the basis of low-power digital design. He died in 2010, holding around 50 patents. He never became famous like Noyce or Kilby. But the next time your phone lasts a full day on a charge, that's '858 doing its work — one transistor on, one transistor off, drawing almost nothing in between.