[sanjayjc]

Genuine question: given LLMs' inexorable commoditization of software, how soon before NVDA's CUDA moat is breached too? Is CUDA somehow fundamentally different from other kinds of software or firmware?

[tomrod]

Current Gen LLMs are not breaching the moat yet.

[fzzzy]

Yeah they are. llama.cpp has had good performance on cpu, amd, and apple metal for at least a year now.

[tomrod]

Thw hardware is not the issue. It's the model architectures leading to cascading errors