Why don't we just release a basic GPU with 128GB RAM and eat NVidia's local generative AI lunch?
The networking effect of all devs porting their LLMs etc. to that card would instantly put them as a major CUDA threat.
But finance folks running the show would never get such an idea...
Intel bagholder investors, view your typical Intel RCG and/or H1B logical thought process in OP. This is who Intel hired to replace the experienced engineers who left to Nvidia, AMD, and Apple. Any wonder why Intel is trading below book value?
Most knowledgeable intel engineer.
Amazon, Google, and a group of smaller companies are piling into the LLM training market with custom processors that will compete with NVidia, just as Apple did against Intel in the laptop processor space. Unfortunately, Intel never figured out how to successfully compete against NVidia, so it isn’t going to magically happen now.
The cluelessness is astounding.
Nvidia is way more than just a good GPU.
It’s a vertically integrated platform with software, gpu, cpu and networking all efficiently working together. You don’t become a 3.5 trillion dollar company with just “moar ram”!
Yes. Jensen has nurtured Nvidia's success. It's his baby, while Intel has been a Grifters Paradise
Nvidia’s software maturation. Proprietary and high throughput low latency gpu interconnect. Strong leadership and roadmap. You do know these disciplines take years to develop and execute while Intel, in its current form, is so sc--wed