Once, Intel intended to acquire NVIDIA; now, NVIDIA directly spent $5 billion on Intel. NVIDIA will subscribe for Intel common stock at a price of US$23.28 per share and become its majority shareholder. Technically, the two parties will achieve seamless interconnection of each other's CPU and GPU architectures through NVIDIA NVLink bus technology with a bandwidth 14 times that of PCIe and ultra-low latency, including the two major fields of PC and data center.
In the data center field, Intel will customize x86 processors for NVIDIA specifically for NVIDIA AI infrastructure platforms, forming a more competitive combination with powerful NVIDIA AI GPUs.
Of course, what everyone is more concerned about is the PC field.Intel will integrate NVIDIA RTX GPUs in its x86 processors in the form of chiplets, called "Intel x86 RTX SOC", which will be widely used in various PC products such as desktops and laptops, covering many fields such as consumption, creation, education, business, and games.
Obviously, a shocking change in the PC market has begun!
Especially in the field of notebook computers, the already close relationship between NVIDIA GPU and Intel CPU will become even deeper.
For example, for thin and light notebooks and mini machines, the deeply integrated Intel x86 RTX SOC will integrate and package powerful performance without the need for a separate graphics card. It can bring ultra-high GPU performance and energy efficiency while maintaining a compact size, just like AMD APU. Of course, this chip is likely to be used in more types of products in the future, and the possibilities can be said to be unlimited.
During the joint press conference,Huang Renxun said: "We believe this is an investment with great potential. This is an important cooperation that will bring huge benefits to both Intel and NVIDIA. We are developing revolutionary products, targeting a market with annual revenue of nearly 50 billion U.S. dollars, such as the data center CPU market with annual revenue of 25 billion U.S. dollars. Among consumer products, notebook computers alone sell 150 million units each year. We will cover most of the consumer PC market."
It is foreseeable that this transaction will be a business opportunity of US$25 billion to US$50 billion per year, and the cooperation between the two parties will further expand the market share of AI laptops and thin and light notebooks.
NVIDIA also emphasized that it is committed to promoting multiple generations of joint research and development product roadmaps, demonstrating its long-term and firm investment in the x86 ecosystem, and will also firmly promote the roadmaps of other product routes and architectures, including the original Arm architecture CPU.
As for the future of its own core graphics cards and independent graphics cards, an Intel spokesperson said that it will continue to provide GPU products. Obviously, at least in the short term, it will not give up easily. It is more likely to form a match with NVIDIA GPUs, complement each other's advantages, and adapt to the needs of different markets.
Regarding this investment cooperation, NVIDIA founder and CEO Jensen Huang said: "AI is driving a new industrial revolution and reshaping every layer of the computing stack, from chips to systems to software. The core of this change is NVIDIA's CUDA architecture. This historic cooperation closely integrates NVIDIA's AI and accelerated computing stack with Intel's CPUs and the huge x86 ecosystem, achieving the integration of two advanced platforms."
Intel CEO Chen Liwu said: "Intel x86 architecture has always been the cornerstone of modern computing for decades. Intel's advanced data center and client computing platform, combined with the strength of process technology, manufacturing, and advanced packaging, will complement NVIDIA's advantages in AI and accelerated computing, and jointly promote the industry to achieve new breakthroughs. Thank you Mr. Huang and the NVIDIA team for their trust and investment in us."
The announcement of this cooperation between the two parties was very sudden, beyond almost everyone's expectations, and there was no warning before.
But in fact,Huang Renxun revealed that Intel and NVIDIA had secretly reached a preliminary cooperation agreement about a year ago and began discussions on related technologies, including the architectural level.
Currently, three teams are working together to develop products, and both PC chips and data center chips will use NVIDIA NVLink as the interconnect technology.
Of course, considering the long cycle of chip product development, especially the need to deeply integrate different technologies from both parties and take into account different indicators, including but not limited to interconnection, packaging, performance, energy efficiency, and software optimization, it may take three or four years to see the first product, and it is initially expected to be by the end of 2027 or early 2028.
Interestingly, Huang Jenxun particularly emphasized: "The Trump administration has nothing to do with this partnership, but of course they would be very supportive.I have informed (U.S. Commerce Secretary) Lutnick about this today. He is very excited and supports the two American companies working side by side. "
After the press conference, Huang Renxun said very directly:"We are very happy to have invested in Intel. The return on this investment will be very impressive: we will become a very large customer of Intel CPUs."
Wall Street analysts also agree very much with this cooperation, believing that as AI widely enters PCs, NVIDIA may benefit from its growing influence on Intel's future product design.
on the other hand,AMD, TSMC, Arm and other companies may feel very hurt.
AMD will face the joint attack of the two giants alone, and everything from PCs to data centers will be under greater competitive pressure.
TSMC may lose orders for some NVIDIA products, ranging from gaming GPUs to AI GPUs, especially once Intel's advanced processes mature.
If Arm wants to enter the PC and data center market, the difficulty will suddenly increase, especially in the data center field. With Intel as its backer, NVIDIA may greatly reduce its dependence on Arm - Arm may be thinking, why didn't the acquisition of me fail at that time?
