Taiwan announced its weight-level news on the evening of the 18th that NVIDIA announced its stake in Intel. At 1 a.m. on the 19th, both parties jointly organized the online reporters' meeting. NVIDIA founder Huang Renqiong and Intel executive di...
Taiwan announced its weight-level news on the evening of the 18th that NVIDIA announced its stake in Intel. At 1 a.m. on the 19th, both parties jointly organized the online reporters' meeting. NVIDIA founder Huang Renqiong and Intel executive director Chen Liwu went to the Fire Line to share the new news about this cooperation.
"60 years ago, IBM System/360 (a large computer launched by IBM in 1964) opened the era of universal computing, but now the CPU-led Morrown Law has reached the limit," Huang Rensheng said back to history. Now, NVIDIA uses GPU accelerated computing to break through the bottleneck, increasing its efficiency by hundreds or thousands of times, while reducing energy consumption and costs, thus creating a big AI explosion. In the next step, NVIDIA will combine Intel x86 CPUs and NVIDIA GPUs to deeply integrate NVLink (Dual Direct Interconnect Technology for CPU and GPUs), and welcome the new era of "accelerated computing and AI computing".
According to the announcement, both parties will jointly develop Intel x86 CPUs, which are specialized in the NVIDIA AI-based facility platform, and promote a new generation of PC SoCs, integrating NVIDIA GPU chiplets. This means that in the future, not only data center servers, but also personal computers will also usher in a brand new experience. Huang Renqiu bluntly stated, "This is a historic cooperation, which symbolizes that the calculation method has been completely changed."
Chen Liwu called out that he had known Huang Rensheng for more than 30 years, witnessed the long-term vision that NVIDIA has laid since the CUDA era, and created an extraordinary AI platform to promote a new market opportunity. Intel was very happy to have this cooperation opportunity. At the same time, he also emphasized, "The cooperation between the two parties is a cooperation to change the rules of the game and an important milestone."
At the same time, Chen Liwu further proposed the strategic meaning of three points of cooperation between the two parties. First of all, NVIDIA is a leading position in the field of AI accelerated computing. Intel has been bullying in the data center and PC CPU market for a long time, and the two strong hands will bring the best combination to the industry. Secondly, this also symbolizes the release of x86 in the new era. Through the combination of NVLink interconnection, unified memory and advanced packaging, x86 will become an important cornerstone of next generation computing. More importantly, both parties unanimously emphasized the concept of "customer-centeredness". "Our belief is simple: to make big products to satisfy customers." Chen Liwu emphasized that the two teams have been put in place and will fully promote this cooperation to achieve success.
NEWS: @NVIDIA and @Intel to develop AI infrastructure and personal computing products.
Read the announcement: https://t.co/Gl28iWwSZc pic.twitter.com/srOhEnr0Ja
— NVIDIA Newsroom (@nvidianewsroom) September 18, 2025
Where does the Taiwan Iron and Television character go?However, the most sensitive question to Taiwanese and stock market investors is: Is the status of the Taiwan Power Prosecution in Shenshan being challenged? In the past, almost all NVIDIA flagship GPU chips were manufactured by Telco. With advanced 3nm and 2nm processes, Telco is the world's top AI chip foundry. Will this cooperation make Intel not only a CPU supplier, but also directly undertake part of NVIDIA's foundry?
In this regard, Huang Renxian will bluntly say in reporters that "Taiwan is still a world-class foundry manufacturer and is irreplaceable in terms of process, production and flexibility." He even described that "Taiwan's Magic no longer requires him to be deliberate." However, Intel's foundry services are rapidly promoting the 18A and 14A processes, and using Foveros multiple packaging technology to try to integrate NVIDIA GPU small chips with Intel CPUs to create a new system across processes.
Called this statement, Chen Liwu emphasized that the focus of cooperation is on products, which does not mean that the existing OEM relationship will be replaced immediately, but it is not ruled out that the evaluation will be carried out step by step in the future.
How can legal person observe? Industry experts pointed out that NVIDIA and Intel have started working on Gaming GPU case last year, and are expected to launch films at the end of 2026. However, it seems that high-level GPUs will still be in NTD, and its data center, PC CPU and GPU are still in the early stages of R&D, and NTD advanced process orders will not be affected by quality.
In addition, the legal person believes that NVIDIA's next-generation AI GPUs still deeply value NTD CoWoS and SoIC technologies, and Intel's own packaging technologies Foveros and EMIB's yield and technical stability are still to be observed. Therefore, it believes that the probability of AI GPUs being transferred to Intel by 2027 is extremely low, and it is optimistic that Telco will maintain a stable position in advanced processing and packaging.
NVIDIA alliance Intel is both a challenge and a firming force for NTU. If Telco continues to be the first in advanced processes and further expands its cooperation with NVIDIA, AMD (Ultra Micro) and even Intel itself, it will continue to be the core supplier of AI chips. It is certain that this "Double Alliance" will reshape the industrial power version of the AI era and also bring new changes to Taiwan's semiconductors in the global market..
Huang Rensha explains why we should cooperate with Intel? Arm and x86 must haveAnother focus is that since NVIDIA has deeply invested in Arm processors, why should it still cooperate with Intel?
Huang Renyue explains that Arm and x86 have different market positioning. He emphasized that the cooperation with Arm will continue to advance, and both parties are developing the next generation of Grace processor "Vera", "Thor", which is used for robots and self-driving cars, and the N1 used by DGX Spark, are all being promoted step by step.
The alliance with Intel, "is to provide the most advanced AI-based facilities with NVLink expansion capabilities in the enterprise and cloud market that still mainstream x86," said Huang Renfeng. In other words, NVIDIA bets on both Arm and x86 routes to ensure that it will not be blocked by a single life system.
Huang Rensha said that NVIDIA's architecture and accelerated computing cover almost all CPU architectures. The point is that NVIDIA hopes to accelerate it to extreme no matter which general computing platform with market impact. Today, his team can cooperate with Intel on the CPU platform. This is the world's largest CPU platform in the enterprise, industry, cloud and consumer markets. Therefore, it is a very exciting cooperation. Both parties will innovate this general computing platform through the integration of NVIDIA's accelerated computing and AI architecture.
Overall, the cooperation between NVIDIA and Intel is a double win. DIGITIMES analyst Chen Zejia believes that for NVIDIA, the purpose is to strengthen its NVLink ecosystem in the data center market, and it is also conducive to the development of RTX series products in the PC market. From Intel's perspective, this cooperation provides additional financial resources for its advanced processes, and at the same time, it also helps the company maintain CPU market share in the data center and PC markets to compete with AMD's products and further expand its market share.
According to current expert analysis, NVIDIA has become the global AI industry overlord who has been calling for a slump. Now it has invested in Intel, and it seems more difficult for its competitors to surpass it in the future.
Extended reading: NVIDIA and Intel cooperate to strengthen the principle of addition and do not change the original plan, and the cooperation with NTU will continue. Spend USD 5 billion to invest in Intel! It is not about crystal OEM, what does NVIDIA want?
