Today, NVIDIA announced that Jensen Huang will step down as CEO (NVIDIA Announces CEO Transition | NVIDIA Newsroom). The reasons are publicly known and I will not dive deeper into that: Kai Fu Lee’s 2018 publication „AI Superpowers“ (Book review: AI superpowers – China, Silicon Valley, and the new world order – Thinking Ahead Institute). You easily see why. What’s more important is to value the contributions Huang made to the AI industry. Here are three essential ones that should not be forgotten.
CUDA. „The Compute Unified Device Architecture“. The enabler for graphics chips to run general-purpose computing. NVIDIA investion billions in the technology while few believed that the investment would pay off. It made products more costly, though. When AI reseachers picket it up, things began to take off. A risky bet, and a breakthrough for the AI industry. The ecosystem built around CUDA has not been matched to date. In 2023 Yahoo Finance summarized this achievement: Going all-in with Nvidia: How Jensen Huang’s high-stakes bets paid off.
Going all-in on Deep Learning. Huang ensured that NVIDIA embraced Deep Learning, and intelligently connected NVIDIA’s technology to the essential players in the emerging AI industry. Remember when Huang personally handed over a new supercomputer for Deep Learning, the DGX-1 to OpenAI for free? It was a publicity stunt, but also so much more than that (Nvidia Just Gave A Supercomputer to Elon Musk-backed OpenAI | Fortune). Huang succeeded in positioning NVIDIA as a central partner for the industry, and NVIDIA has kept that position all the way through. NVIDIA built what AI researchers needed. As simple as that.
Scaling from chips to data centers. In 2019, NVIDIA bought Mellanox (NVIDIA Completes Acquisition of Mellanox, Creating Major Force Driving Next-Gen Data Centers | NVIDIA Newsroom), enabling NVIDIA to scale from mere chips to data centers. The industry had matured by the end of the 2010s. Training LLM models demanded massive amounts of GPUs. The bottleneck was ensuring an efficient way how chips talked to each other. With the newly acquired networking layer, NVIDIA could offer computing power on a „AI factory“ level scale. The networking segment has become a major revenue driver since then (Nvidia Q2 2025: Networking is booming but your networks cost nothing – DCD).
Sure, all of this is not the work of one man. But Huang was the one who betted strategically, took risks, and saw things through from a CEO perspective. One can rightfully argue that the current AI industry would not exist as it does without Huang’s contribution. While his era at NVIDIA ends today, there is likely much more to come.