HPCwire had a chat with Dave Ditzel, the former CEO of Transmeta, who was in the thick of it during Nvidia’s early days of dabbling with server processors. According to Ditzel, Nvidia’s move to Arm stemmed from its licensing of Transmeta’s Tokamak, which could translate x86 code into a RISC instruction set.

“Nvidia brought out a product called Denver,” he explained. “It was that same design. It originally started as an x86 but through certain legal issues, had to turn itself into an Arm CPU.”

This decision set Nvidia on the path to embracing Arm architecture. Tokamak, developed by Transmeta, was meant to be the company’s third-gen x86 chip, following the Crusoe and Efficeon processors. But it never saw the light of day as a product. Instead, it was licensed to companies like Intel and Nvidia. Intel, for its part, acquired the design but didn’t launch a product based on it either.

Nvidia officially introduced Project Denver as an Arm-based CPU in 2011 and later rolled it into its Tegra lineup. While there was initial buzz around Arm servers, the software ecosystem wasn’t quite ready, which limited adoption.

Since then, Nvidia has shifted focus, unveiling its Grace CPU and stepping back from its attempt to buy Arm after regulators decided it wasn’t a great idea. Still, the Arm detour that started with Project Denver set the stage for Nvidia’s current trajectory, blending innovation with lessons learned from a bit of tech drama.