NVIDIA’s stock is up 25% after purchasing $5 billion worth of Intel and RTX iGPUs for x86.
Using the top design teams from both companies, NVIDIA and Intel today announced one of the biggest industry partnerships in recent years: working together to produce several generations of consumer PC and data center devices. The most recent announcement states that NVIDIA is investing $5 billion at a price of $23.28 per share, making it one of the biggest shareholders in Intel, holding a 4.9% stake in the entire business. This collaboration’s objective is divided into two verticals. One, and possibly the most intriguing, is the incorporation of NVIDIA RTX GPUs inside Intel x86 System-on-Chips (SoCs), which were once powered by Intel’s own Arc GPUs. With its “Kaby Lake G” concept, which never advanced past generation one, Intel even contemplated partnering with AMD. As the industry standard for integrated graphics, NVIDIA’s RTX GPUs will power millions of laptops, mobile devices, and potentially even desktop computers.
The creation of unique x86 CPUs for NVIDIA, which will be incorporated into NVIDIA’s AI infrastructure platforms, is the second pillar of this partnership. spanning several verticals, including NVIDIA SuperPODs, DGX workstations, and HGX servers. Some HGX systems presently use x86 CPUs in addition to NVIDIA’s in-house Arm-based Grace/Vera CPUs. For the time being, this has worked well for NVIDIA, and according to its data center plan, the company will keep shipping these customized CPUs as part of its product line. However, NVIDIA previously totally avoided both x86 vendors with proprietary Arm designs, but Intel will earn a substantial part in the AI training and inference infrastructure with special x86 CPUs optimized for NVIDIA. For instance, Intel’s designs, like the next Clearwater Forest Xeons, will use the 18A node to deliver up to 288 “Darkmont” E-cores on a large chiplet package. This should work well with NVIDIA’s accelerators, and we may even see some modified versions that run on NVIDIA’s SuperPODs with hundreds of GPUs but have fewer cores but higher frequency.
Lastly, by investing in Intel, NVIDIA is demonstrating that its design teams are prepared to collaborate with Intel Foundry, which will provide 18A and 14A nodes for upcoming NVIDIA GPUs. NVIDIA may avoid the majority of supply chain problems and risks associated with manufacturing its newest and best GPUs abroad by bringing production in-house. Additionally, Intel Foundry provides cutting-edge packaging services like Foveros 3D, which may pave the way for some extremely intriguing future designs. With NVIDIA as its lifeline, Intel Foundry goes from having virtually no major clients to having the biggest market capitalization firm as a client. “Intel’s x86 architecture has been foundational to modern computing for decades – and we are innovating across our portfolio to enable the workloads of the future,” stated Lip-Bu Tan, CEO of Intel. “With NVIDIA’s leadership in AI and accelerated computing, Intel’s top data center and client computing platforms, along with our manufacturing, process technology, and advanced packaging capabilities, will enable new industry breakthroughs.” We value the trust Jensen and the NVIDIA team have shown in us by investing in us, and we eagerly anticipate the work that lies ahead as we develop new products for clients and expand our company.”
Every product segment is seeing more competition. AMD must now contend with NVIDIA and unified Intel, but that demonstrates something else. AMD’s next product line appears to be quite competitive and robust, as seen by NVIDIA’s decision to use Intel to get the best performance out of its AI servers. At the open, Intel’s stock is up 25%. We will learn more about the news at a press conference today at 10 a.m. Pacific time (7 p.m. CET) with Jensen Huang, the CEO of NVIDIA, and Lip-Bu Tan, the CEO of Intel.
