CUDA is losing its AI dominance
The argument was made that NVIDIA's CUDA software platform is rapidly losing its competitive moat in both AI training and inference.
The argument
The guest pointed out that two of the three leading frontier models—Google's Gemini and Anthropic's Claude—are trained and served on non-CUDA platforms like TPUs and Trainium. Furthermore, migrating models from GPUs to alternative hardware can now be executed in as few as ten keystrokes, undermining CUDA's historical lock-in.
The thesis, stress-tested
✓ What validates it
- ✓Further migration of major LLMs to non-GPU hardware
- ✓Increased developer adoption of alternative APIs for model deployment
▸ Risks discussed
- ▸NVIDIA could introduce new proprietary features to lock in developers
- ▸OpenAI's GPT remains heavily reliant on the CUDA ecosystem
Hear it yourself
"the latest news from around the world delivered to you in an instant. Subscribe to Bloomberg News Now and get news when you want it on your schedule. These are short audio reports on the day's top stories that you can listen to in just a few minutes. Make Bloomberg News Now your trusted source for the latest global headlines with context. It's available twenty four hours a day, anytime you need it, right on your smartphone or smart speakers. Subscribe to Bloomberg News Now today on Apple Podcasts, Spotify, or anywhere you listen."
AFFILIATE LINK · ZORTIX MAY EARN A COMMISSION · NEVER A RECOMMENDATION TO TRADE