Every AI lab and trillion-dollar tech giant traces back to a single lab and one invention: the transistor. This is the lineage — the spin-offs, acquisitions and generational leaps that carried computing from a germanium sliver in 1947 to the AI era.
Each era was built on the last — and each threw off the companies that would dominate the next.
It begins with the transistor at Bell Labs (1947). William Shockley founded a lab in California in 1956; a year later eight of his engineers — the "traitorous eight" — left to start Fairchild Semiconductor. Fairchild in turn spawned Intel (1968) and AMD (1969), seeding the entire Silicon Valley chip industry.
Once chips were cheap, value moved to software. Microsoft (1975) and Apple (1976) turned the microprocessor into the personal computer; MS-DOS then Windows became the platform the world ran on, while Apple defined the graphical, consumer machine.
Networking the PCs created a new layer. Amazon (1994) and Google (1998) were built to sell and search the web; the browser, the data center and the ad-funded free service became the dominant models, and cloud computing (AWS, 2006) turned computing itself into a utility.
The iPhone (2007) put a supercomputer in every pocket and made Apple and Google (Android) the gatekeepers of computing. Social platforms (Meta) and the smartphone sensor boom generated the vast datasets that the next era would learn from.
Cheap GPUs met big data and deep learning. NVIDIA's chips became the engine; labs like DeepMind (2010), OpenAI (2015) and Anthropic (2021) built the models; and ChatGPT (2022) put generative AI in front of a billion people, kicking off the build-out decoded across this whole site.
The pattern repeats: a new hardware layer (chip → PC → web → phone → AI) commoditises the old winners and mints new ones. Today's giants are racing not to be the incumbents that the AI era leaves behind.
Silicon Valley's family tree, traced from one lab. Monospace, the way it was drawn.
Bell Labs — the transistor (1947)
│
Shockley Semiconductor (1956)
└─ [the "traitorous eight" leave, 1957]
└─ Fairchild Semiconductor (1957)
├─► Intel (1968) ── 4004 · x86 · Core · today's CPUs
├─► AMD (1969) ── x86 rival · Ryzen · EPYC · Radeon
├─► National Semiconductor
└─► [the "Fairchildren"] ── dozens of chip startups
└─ venture capital (Kleiner Perkins, Sequoia)
└─ funded Apple, Google, Amazon, NVIDIA…
Software layer → Microsoft (1975) · Apple (1976)
Internet layer → Amazon (1994) · Google (1998)
Mobile layer → iPhone (2007) · Android
Accelerator layer → NVIDIA (1993) — GPUs → the AI engine
AI-lab layer → DeepMind (2010) · OpenAI (2015) · Anthropic (2021)
└─ ChatGPT (2022) → the modern AI boom
| Company | Founded | Born from | Why it matters today |
|---|---|---|---|
| Intel | 1968 | Fairchild spin-off | Defined the CPU; now fighting to regain the leading edge. |
| AMD | 1969 | Fairchild spin-off | x86 rival; MI-series GPUs chase NVIDIA in AI. |
| Microsoft | 1975 | PC software | Windows/Office; now the biggest backer of OpenAI. |
| Apple | 1976 | Personal computing | iPhone + in-house M-series & A-series silicon. |
| NVIDIA | 1993 | PC graphics | GPUs became the engine of the entire AI boom. |
| Amazon | 1994 | E-commerce → cloud | AWS runs much of the internet; Trainium AI chips. |
| 1998 | Web search | Search, Android, TPUs, and DeepMind/Gemini. | |
| TSMC | 1987 | Foundry model | Fabricates nearly every leading-edge AI chip on Earth. |
| DeepMind | 2010 | AI research | AlphaGo, AlphaFold; now Google's AI core. |
| OpenAI | 2015 | AI research lab | GPT models and ChatGPT ignited the current boom. |
| Anthropic | 2021 | OpenAI alumni | Claude models; safety-focused frontier lab. |
> A simplified lineage; many companies have tangled, overlapping origins. Founding dates are widely documented.