# NVIDIA launches CUDA

Date: 2006

## The fact

In November 2006 NVIDIA launched CUDA, a completed programming-model instrument under Huang that later became the default AI-training stack.

## Context

Neighbouring GeForce is the hardware. Neighbouring 2024 French study is later scrutiny.

GeForce 1999 is neighbouring hardware. 2024 CUDA-lock-in inquiry is later concentration.

[Jensen Huang](/people/jensen-huang) has been NVIDIA’s co-founder and CEO since April 1993. He does not write a BIS Entity List, a Commission DMA article, a TSMC fab act, a Chinese CAC circular, or a French Autorité decision. A GTC keynote is not a completed export licence. A subpoena is not a conviction. CUDA is not a court finding of monopoly unless a court so found. TSMC fabricates; hyperscalers buy; Beijing and Washington write the licence. Keep those dates separate. A Greenpeace cake is not a unique ppm. A DOE supercomputer is not a unique arsenal coefficient as his personal tally.

## Evidence and sources

NVIDIA carries the official instrument or communique. Reuters records the same dated act independently.

## Significance

## What is established and what is not

Established: November 2006 CUDA launch, independently reported.

Not established: that rivals could not write other models, or a unique later FLOP coefficient from the first toolkit.

What would change the picture: a contradictory original with a different date or name, a later repeal, a court that vacated the instrument, or a measured series that can be argued, with lag and alternatives, as a consequence of this instrument rather than of later institutions.

The date is the date of this instrument, not a licence to back-date every later harm or benefit.

Developer tools shipped. Researchers adopted. Later cloud buyers standardised. Courts had not yet sat on CUDA as Section 2.. A later repeal, court stay, or successor order would be a different Fact with a different date. That pair identifies NVIDIA launches CUDA, not a later speech about NVIDIA launches CUDA.

Co-authors and later implementers named in Context remain on the file. NVIDIA launches CUDA on 2006-11-08 is not a licence to treat every later NVIDIA communique as this same act. If a later repeal, stay, or successor instrument exists, it needs its own date and its own row. This dossier refuses to average NVIDIA launches CUDA into a personality. The contribution described on this Person row is limited to He launched as CEO. Developer tools shipped. Researchers adopted. Later cloud buyers standardised. Courts had not yet sat on CUDA as Section 2. That limit is the alternative to treating office-holding as authorship of every later series.

## People

- Jensen Huang: https://factrail.ai/people/jensen-huang

## Drivers

- Industrial-Policy Activism: https://factrail.ai/drivers/industrial-policy-activism
- Digital Learning Access: https://factrail.ai/drivers/edu-digital-learning-access

## Sources

- https://www.reuters.com/technology/nvidia-cuda-2006/
- https://developer.nvidia.com/cuda-zone

## Methodology

https://factrail.ai/methodology

Canonical: https://factrail.ai/facts/nvda-cuda-2006