Semiconductors
The transistor is the most manufactured object in human history - over 10 sextillion made since 1947.
What is a Semiconductor?
A semiconductor is a material (usually silicon) that conducts electricity under some conditions but not others. By doping silicon with small amounts of other elements (phosphorus for n-type, boron for p-type), engineers create regions that can be switched between conducting and insulating. This is the transistor - the fundamental building block of all modern electronics.
The Transistor Invention (1947)
John Bardeen, Walter Brattain, and William Shockley at Bell Labs invented the transistor on December 16, 1947, winning the 1956 Nobel Prize in Physics. It replaced the vacuum tube - smaller, more reliable, cheaper, and using far less power. Texas Instruments introduced the first silicon transistor in 1954. The integrated circuit (multiple transistors on one chip) followed in 1958.
Moore's Law
Gordon Moore observed in 1965 that the number of transistors on a chip doubled roughly every two years while cost halved. This "law" held remarkably well for 50 years. A 1971 Intel 4004 had 2,300 transistors. A modern Apple M3 Max chip has 92 billion. Moore's Law is now slowing as transistors approach atomic scale (current chips use 3nm nodes, approaching the size of individual silicon atoms).
How Chips are Made
Semiconductor fabrication is the most complex manufacturing process in human history. A silicon wafer is coated with photoresist, then exposed to extreme ultraviolet light through a mask, etching circuit patterns. This is repeated hundreds of times with nanometer precision. TSMC's most advanced fabs cost $20+ billion to build and require machines (ASML's EUV lithography systems) that exist nowhere else and take years to produce.
The Global Chip Supply Chain
No single country can make advanced chips alone. Dutch company ASML makes the only EUV machines in the world. Japan supplies specialty chemicals and materials. Taiwan (TSMC) and South Korea (Samsung) do the most advanced fabrication. The USA designs many chips (Intel, AMD, Nvidia, Apple, Qualcomm) but manufactures fewer. This interdependence creates massive geopolitical tension, particularly between the USA and China.
GPUs and the AI Boom
Graphics Processing Units (GPUs) were designed for rendering 3D games by processing millions of simple calculations in parallel. Researchers discovered this parallel architecture was perfect for training neural networks. Nvidia's CUDA platform (2006) let researchers use GPUs for general computation. Nvidia's market cap went from $300 billion in 2022 to over $3 trillion by 2024, driven entirely by AI demand for its H100 chips.
Chips and Nuclear Technology
Radiation hardened chips are essential for nuclear power plants, weapons systems, and space applications. Normal chips are disrupted by ionizing radiation which flips bits (single-event upsets). Rad-hard designs use larger transistors, redundant circuits, and special materials to resist this. The electronics inside nuclear reactors must function reliably for 40-60 years in a high-radiation environment.