A dark day for Nvidia? Here’s what analysts are saying about Jensen Huang’s departure and CUDA’s lasting strength

Buck’s long work on CUDA

Nick Buck has worked alongside Huang since 2004, when he began building CUDA. The software effort started with two people and two libraries; it now supports more than 900 libraries and artificial intelligence models, with each library linked to an industry using Nvidia technology.

Buck’s interest in graphics began during his undergraduate years at Princeton in the late 1990s. While earning a Ph.D. at Stanford, he connected graphics processing units to push their capabilities and pursued ways to use them beyond graphics.

Turning GPUs into general tools

Stephen Witt, author of “The Thinking Machine,” said Buck and John Nickolls shared an intense drive to make computers faster. Witt also said Huang had expressed frustration that Nickolls did not receive the recognition he deserved after his death.

CUDA became Nvidia’s defense

The platform helped Nvidia reach a 90% share of AI computing at one point and has become a way to protect its market position. Its strength also comes from millions of developers who have improved the platform over decades.

A culture built on directness

The report describes Nvidia under Huang as a company that avoids one-on-one meetings and makes disagreements and mistakes public. People who worked with Buck at Stanford, Nvidia or both described him as sociable and easygoing, but intense when pursuing goals and more interested in results than theory.

At a GTC press conference, Buck discussed convolutional neural networks and described proteins as “complicated 3D squigglies in your body.” Harris said CUDA’s libraries help businesses avoid having to learn the software themselves. The report says competitors are advancing more quickly, though none has yet caught Nvidia.