Detailed · 11 events
A History of Semiconductors and Hardware
1940s
John Bardeen, Walter Brattain, and William Shockley at AT&T Bell Labs demonstrated amplification in a point-contact transistor made of germanium. The birth of solid-state electronic amplification—replacing the vacuum tube that had dominated since 1925—and the origin of the industrial revolution that made computing small, low-power, and mass-producible. The three shared the 1956 Nobel Prize in Physics.
1950s
Jack Kilby at Texas Instruments demonstrated the first working integrated circuit, fabricating several transistors, resistors, and capacitors onto a single germanium substrate. Six months later, Robert Noyce at Fairchild Semiconductor independently invented an IC on silicon using the planar process; Noyce's approach became the dominant manufacturing method. Kilby received the 2000 Nobel Prize in Physics.
1960s
Gordon Moore, head of R&D at Fairchild Semiconductor, contributed an article to Electronics magazine projecting that the number of components that could be integrated onto a chip would roughly double every two years or so. The observation became Moore's Law, the de facto roadmap of the semiconductor industry. In 1968 Moore and his colleague Robert Noyce co-founded Intel.
1970s
Designed by Federico Faggin, Masatoshi Shima, Ted Hoff, and Stan Mazor at Intel through a calculator-LSI contract from the Japanese company Busicom. 2,300 transistors, 108 kHz clock. The first commercial product to put general-purpose processing on a single chip, and the start of the lineage that ran through the Intel 8008, 8080, 8086, and into x86.
1980s
Acorn Computers in Cambridge, UK, completed ARM1—a RISC (Reduced Instruction Set Computing) processor designed in-house for the Acorn Archimedes. Its low power consumption and compact instruction set later became the default for embedded and mobile devices; from the 2007 iPhone onward, ARM dominated the world's smartphone market. By 2026, the majority of processors shipped globally use the ARM instruction set.
2000s
NVIDIA released the official version of CUDA (Compute Unified Device Architecture)—a programming model for using gaming GPUs as general-purpose parallel processors. Initially aimed at scientific computing, the alignment with deep learning demonstrated five years later by AlexNet (2012) turned NVIDIA into the central infrastructure company of AI computation.
2010s
The home button became a fingerprint sensor: Touch ID brought biometric unlock and purchase authorisation to the mainstream. At the same time the A7 became the first 64-bit smartphone processor in the world—a transition the industry had expected two years later, and one that took competitors including Qualcomm by surprise. Touch ID would later become the foundation for Apple Pay (2014).
At ImageNet 2012, Alex Krizhevsky, Ilya Sutskever, and Geoffrey Hinton of the University of Toronto reached a top-5 error rate of 15.3%—more than ten points ahead of the runner-up's 26.2% obtained by conventional methods. Their convolutional neural network, 'AlexNet', trained on two NVIDIA GTX 580 GPUs, proved the practical viability of deep learning overnight. Computer vision shifted, almost completely, from hand-engineered features to deep learning from that point forward.
Taiwan Semiconductor Manufacturing Company (TSMC) began volume production at 7 nm—the industry's first—and would continue through finer nodes at 5 and 3 nm. The first major customer was the Apple A12 in the iPhone XS. While Intel struggled to move to 10 nm, TSMC consolidated a near-monopoly on leading-edge semiconductor manufacturing. The chips of Apple, AMD, NVIDIA, and Qualcomm are now nearly all fabricated in TSMC's plants.
2020s
Apple announced the first Macs—MacBook Air, Mac mini, and 13-inch MacBook Pro—running the in-house ARM-based M1 chip, ending the fifteen-year run of Intel x86 in the Mac. The M1 outperformed the Intel Macs on both power and speed, the culmination of more than a decade of Apple's internal chip-design effort. With it, the Mac joined iPhone, iPad, and Apple Watch on a unified Apple Silicon base.
NVIDIA announced the H100 GPU—the de facto standard training and inference processor of the generative-AI era. Hopper architecture, 80 GB of HBM3 memory, FP8 support. As the demand surge following the November 2022 ChatGPT launch took hold, the H100 and its successors (H200, B200) became the core of global AI infrastructure investment, lifting NVIDIA's market capitalisation into the top tier worldwide.