If you’ve been following the tech news over the last few hours, the announcement of the partnership between Intel and Nvidia came as a bolt from the blue. Promising on paper, the two CEOs expressed their ambitions in a video webcast. Let’s be objective: at this stage, no roadmap or product has really been announced, and it will be some time before we have a precise vision of what this collaboration will bring. However, as soon as the announcement was made, the two companies immediately mentioned the ambition of an x86 PC SoC integrating RTX graphics. Logically, two quick analyses emerged: what about ARC development at Intel and on the Nvidia side, and what future for collaboration with ARM, particularly in Windows PC SoCs?
The Soc X86 / RTX first product to come?
While both bosses remained nebulous about the product roadmap, Jensen Huang did say that the first product to come out of the Intel Nvidia collaboration would be a “GB10” chip, featuring x86 cores instead of the ARM cores previously used by MediaTek’s development department. He described this future chip exactly as the current GB10: an SoC with CPU cores coupled to an Nvidia GPU via NVLink. This leaves considerable doubt as to whether this famous N1 SoC will ever make it into Windows PCs.
However, Nvidia has made it clear that ARM will remain a solution used by the company. Nvidia will continue to develop its own processors, not only for servers, but also for cars, robots and more. The DGX Spark will also be launched with the N1…But the greens aren’t talking about it in the PC world anymore.On Intel’s side they explain that the use of RTXs would be “complementary” to the ARC range and that this deal won’t change the blues’ current roadmap: We’re not discussing specific roadmaps at the moment, but the collaboration is complementary to Intel’s roadmap and Intel will continue to offer GPU product offerings. (Statement by an Intel spokesperson).
Two issues continue to raise questions:
Firstly, this operation, which came as a surprise to many, is the fruit of discussions that date back several months, in theory even before Pat Gelsinger’s departure. Several people were involved in the discussions to assess the interoperability of the two companies’ products. 
The agreement between Intel & Nvidia was signed on September 13. This T-shirt shows the people involved in the deal[/caption] Finally, how are we to interpret Nvidia’s $5 billion investment in Intel shares when the industrial partnership didn’t really require it? It’s hard not to imagine that the secondary objective was not to produce chips at Intel. First of all, for this famous X86 / RTX SoC, for which it seems hard to imagine that Intel would agree to entrust its manufacture to TSMC, when with Foveros the company has one hell of an advantage. And secondly, how can we not think that Nvidia will try to loosen TSMC’s grip by producing elsewhere? Obviously, a lot remains to be seen at this point.
