How Nvidia Took Over the World With Video Game Chips

Nvidia spent most of its life making graphics chips for video games. In October 2025, it became the first company in history worth $5 trillion. This is the story of how chips built to make games look better ended up running the AI revolution. It starts in 2012, in a bedroom in Toronto, where a graduate student used two gaming graphics cards to teach a computer to see. Then we go back to the beginning: three engineers in a Denny's booth, a first chip that drew the wrong shape, a confession to Sega that bought six months to live, and a million dollar gamble on a machine that pretended to be a chip. Then the "science project" called CUDA that Wall Street hated for years, until AI showed what it was for. Then ChatGPT, an AI gold rush where Nvidia sells the shovels, and the cracks that could still hurt it: Taiwan, rivals' own chips, China, and $700 billion of spending that has to keep going. Luck is part of it. But when the future finally showed up, it was already speaking Nvidia's language. Chapters 0:00 A bedroom in Toronto 1:08 The first $5 trillion company 2:10 Three engineers at a Denny's 3:18 Why games needed a new kind of chip 4:16 The craftsman and the army of apprentices 5:37 Forty thousand dollars and a name 6:57 The chip that drew the wrong shape 8:06 Telling Sega the truth 8:46 Thirty days from going out of business 10:16 The first GPU 11:54 Scientists disguise math as pictures 13:43 CUDA 15:14 The bet Wall Street hated 16:10 How a machine learns 18:17 ImageNet and AlexNet 20:27 Nvidia goes all in on AI 21:25 Transformers and buying progress 22:26 Why couldn't anyone catch up? 24:00 When chips have to talk 25:19 A crypto crash, then ChatGPT 26:26 The AI gold rush 28:34 What could go wrong 31:24 So why Nvidia? Sources Nvidia Blog, the Denny's founding: https://blogs.nvidia.com/blog/nvidia-... Acquired, Nvidia Part I: https://www.acquiredbriefing.com/p/nv... NV1: https://en.wikipedia.org/wiki/NV1 Jensen Huang, NTU commencement speech (2023): https://interconnected.blog/jensen-hu... EE Times, the Ikos emulator order: https://www.eetimes.com/ikos-announce... HPCwire, Ian Buck: https://www.hpcwire.com/35-hpc-legend... Lex Fridman Podcast, Jensen Huang transcript: https://lexfridman.com/jensen-huang-t... The Register (2008): https://www.theregister.com/2008/07/0... AlexNet: https://en.wikipedia.org/wiki/AlexNet Krizhevsky, A., Sutskever, I. and Hinton, G. E. (2012). ImageNet Classification with Deep Convolutional Neural Networks. NeurIPS. Deng, J. et al. (2009). ImageNet: A Large-Scale Hierarchical Image Database. CVPR. Metz, C. (2021). Genius Makers. Vaswani, A. et al. (2017). Attention Is All You Need. Kaplan, J. et al. (2020). Scaling Laws for Neural Language Models. Nvidia annual and quarterly financial results, fiscal 2023 to fiscal 2027. Nvidia GTC 2026 (CUDA developer count). The Modern Machine makes animated documentaries about the companies and technologies that built the modern world. #Nvidia #AI #Documentary