Nvidia and Intel x86 AMD
Share the Story:
Intel & Nvidia join forces
With the Information Technology world ever shrinking, we see a big, but not unexpected, leap forward in the consolidation machine. Nvidia have never hidden their desire to add CPUs to their repertoire of hardware delivery. Announced earlier today Nvidia have teamed up with Intel, to the tune of $5billion, and secured a ‘licence’ for x86 CPUs to be part of their Server Farm and Hyperscaler business to push their NVLink solution further. The deal is very early in design and has likely been announced to secure further funding sources for Intel, which was only recently bailed out by the US Government to the cash register value of $9 Billion earlier this year. The reasons for the dire situation Intel find themselves in are not a surprise for those within the industry, and even those outside it.
But, what does this deal actually offer for both parties you ask? Well from the announcement (detailed here) it is a supplementary development stream for Nvidia, as they continue on their long standing ARM and other internal developments within their AI Cloud, car and robotic product line. Largely driven by their domination of the GPU market via their Geforce and RTX brands, creating AI leading software and hardware that has driven the technology market in almost all sectors (mine included). The only chink in their armour has been the lack of their own complete SoC options via a robust CPU pathway to enable them to be completely self-sufficient. Something this deal, and the prior ARM contract they have now resolves on a biblical scale.
The news is not only good for Nvidia though, Intel are on the ropes since their good, but late and lacking Battlemage GPU’s last year failed to make a dent into the GPU market place. Having been squeezed out of the home PC gaming CPU market more and more over recent years by AMD, they have found their once industry leading CPU dominance slip significantly. Both in the home and, even more impactful, the commercial DataCentre and manufacturing industries across the globe. With them losing ~15% of their business control to the US Government(10%) and now Nvidia(5%), this deal is almost the best case scenario for the American tech giant, one that has been on the cards for the past Two years or more. Over on Moore’s Law Is Dead YouTube channel, Tom leaked that the Battlemage GPU architecture and roadmap was dead last year. And this new deal confirms the accuracy of that leak concretely. The pairing of Intel CPU hardware with Nvidia’s GPU and software stack is a formidable combination though for AMD, Apple and Qualcomm, and this will certainly set many executives off into strategy and defense mode. Not to mention US Government monopoly commission.
Into Battle We Go
Although much is yet to be detailed and clarified from both companies on the roadmap ahead. We can already see early signs of where and what will come first, along with the fact this is not a new direction for Nvidia at all, no. More, a new route to market penetration and revenue growth in both purchase and sales. From this early announcement, we know that Nvidia plan to purchase a large stockpile of Intel x86 CPU’s which they will incorporate into their DataCentre (Cloud) infrastructure business. Nvidia have been using Intel Xeon CPUs (their server brand of chips) for many a year in DCs. But this new partnership means they are now strategically linked, enabling a deeper symbiosis of each others architecture than before. Current Servers still use the dated PCIe interconnect which keeps CPU and GPU, and memory, very distinct. Now they will leavearage the best of each design via their NVLink technology. This will enable them to have efficient, fully integrated, very powerful, and cost effective solutions to expand on that market with new scalability and potentially even better pricing, although that is far from a given.
Supplementing this, which is very much Nvidia’s roadmap already, is Intel’s roadmap of expanding their own Home PC, laptops, Mobile PC and corporate DataCentre hardware. Via a cost efficient and powerful SoC portfolio to rival AMD’s APU, NPUs and DataCentre stack. The combination of Intel’s undoubtedly, top tier x86 Architecture, assimilated with Nvidia’s top end AI and GPU hardware, creates a fearsome proposal to OEM manufacturer’s, PC builders and even console makers. Both Microsoft and Valve have prior and agreed roadmaps with AMD, but things can change and with “everything is an Xbox” strategy from Microsoft now. We could see a very high end, small and efficient PC or mobile device coming down that road in the not to distant future via any of the OEMs signing up to that deal. Valve could be their first customer if the rumoured Fremont home machine were to use these chips or even the, certain to be in development, Steam Deck 2. The result is a win, win for both teams but also a partnership that could disrupt multiple markets and companies in very short order. All these designs will be supplied via Nvidia RTX Chiplets and will be integrated into all of Intel’s architectures, replacing their current Battlemage GPU and AI NPU segments. Although NVlink was not announced to be part of this as yet, I would expect this will be a growth area for both to better embed and maximise die space, memory bandwidth and performance, specifically in the Laptop, Mobile gaming and even Phone space. This is all brand new and very early in the plan stage for both companies, but be sure to stay tuned on Binary Dinosaur website and YouTube channel for more news and analysis on this new venture.
