AI News

Intel CEO Lip-Bu Tan: Joint chip development with NVIDIA will continue for years

Electronic Technology ApplicationUpdated Nov 16 20:46

AI summary

Intel CEO Lip-Bu Tan said that joint development between NVIDIA and Intel will continue for years, and the two sides will announce the products together once they are ready. The collaborative products include server chips that integrate Intel CPUs into NVIDIA AI GPUs, as well as PC products that integrate NVIDIA GPUs into Intel processors; the latter is codenamed Serpent Lake, with an optimistic forecast of debuting as early as next year, though 2028 is more likely.

Body · AI translation

News from November 14: NVIDIA and Intel have also had an on-again, off-again relationship, and in recent years the two have become competitors. However, in September last year NVIDIA announced a $5 billion investment in Intel, becoming one of the latter's major shareholders.

This collaboration is not merely a financial investment; it also involves deep cooperation on jointly developing chips, which inevitably brings to mind that 10 years ago Intel made an unprecedented move to partner with AMD to develop a Core processor integrating an AMD GPU—unfortunately, that was discontinued afterward.

Regarding the NVIDIA collaboration, Intel CEO Chen Liwu mentioned in an interview recently broadcast by Yahoo News that the joint development between the two sides will continue for many years, and he hopes everyone will keep paying attention.

He did not give a launch time for the products, only saying that once the products are ready to launch, the two sides will announce them together.

It is worth noting that this interview was conducted on October 21 and is only now being broadcast, so it will not contain Chen Liwu's comments on Huang Renxun's resignation.

But the relationship between the two has always been good, and last year's investment also came from that; the two veterans are old friends, and in this relationship of both competition and cooperation they can also find common interests.

As for the specific products of the collaboration, there are those suitable for the AI server market as well as those for the PC consumer market. The former will integrate Intel CPUs into NVIDIA's AI GPUs; although NVIDIA has its own self-developed ARM CPUs, x86 still occupies a core position in the server market.

Among the PC-oriented products, Intel will integrate NVIDIA GPUs into its own processors. This combination is more attractive than the Intel CPU + AMD GPU of that year; after all, N cards' dominance in this area is unmatched by AMD, and they can not only play games but also run AI.

If one must find a product analogy, the result of the two sides' collaboration may be similar to the current RTX Spark, which integrates a GPU unit at the RTX 5070 level, but the CPU is still a customized ARM architecture from MediaTek. In the future there will be an Intel x86 version, with more complete performance and ecosystem compatibility.

The codename of this product was leaked before: Serpent Lake. However, there are no specific details, nor any credible leaks, and there is no launch time for the product. Optimistic predictions say as early as next year, but 2028 is a more likely time point for launch.

Chen Liwu said that when the products are ready, the two sides will announce them together. If Huang Renxun had not resigned, then one or two years later the two would certainly release them together; it is just a pity that it is hard to say now, and it is unknown when NVIDIA's new CEO will be confirmed.

Of course, it cannot be ruled out that Huang Renxun will make a comeback in a year or two; after all, that would be the best thing for NVIDIA. For now, let's lie low for a while.

Source

Electronic Technology Application