At Intel's 2022 Investor Conference, Intel CEO Pat M. Kissinger and business leaders in various departments outlined a blueprint for Intel's future development plans. Among them, the development of Intel's emerging IFS foundry services and the next step in the implementation of IDM 2.0's grand plan are particularly eye-catching.
In four years, five process nodes, Intel wants to "continue" Moore's Law for ten years
With the development of the chip industry in recent years, Moore's Law has a tendency to gradually fail. Many people in the industry believe that Moore's law has come to an end, and the number of transistors in the future development of the chip industry will no longer follow Moore's law.
Today, when the process of process improvement is becoming more and more difficult, Intel has also stagnated in the process process, and even the outside world has suspected that Intel will abandon the IDM model and withdraw from the chip manufacturing industry. Today, there are still many doubts about whether Intel's process can make a breakthrough.
In response to questions from the outside world, at this investor conference, Intel announced its plans for the development of the process in the next few years. According to Intel's plan, in the next four years, Intel will cross five process nodes. The Chips from the Intel4 Process will be in production in the second half of 2022, and the Intel4 will use EUV technology, and its transistor performance will be increased by about 20% per watt. The more advanced Intel3 chips will be in production in the second half of 2023 and will be used in Xeon processors released in the same year. The Intel3 will have more features and a performance increase of about 18%.
In 2024, Intel will fully enter the Amy era, launching Intel20A and Intel18A process chips based on the new transistor architecture RibbonFET. Intel is confident in achieving breakthroughs in its future processes and has made predictions that Moore's Law will continue for a decade.
x86 "embraces" the industrial chain and establishes a more open chip ecosystem
IFS, as a new business that Intel just launched last year, has been attracting attention. The British media The Register recently exposed the news that Intel will license the x86 architecture to manufacturers in the future in order to promote the further development of the business.
At the Investor Conference on Feb. 18, Intel officially announced the news. Although Intel did not disclose more information at the investor conference, according to The Register' revelations, Intel's authorization for the x86 architecture includes soft and hard cores, and may even include Xeon cores in terms of core types. However, Intel's licensing model is not the same as the common pattern: Intel does not send design files (GDS II) to customers that can be factory-founded. Instead, it offers customers the possibility of mixing the x86 architecture and other architectures. Customers can pick the corresponding x86 core, mix and match with arm, RISC-V and other cores, and be OEM by Intel IFS services.
Intel said the foundry service will use the Intel16 process for production, and in the future it will be transferred to more advanced Intel3 and Intel18A process processes. This program is designed to meet the customization needs of more customers to advance Intel's IFS business.
In addition, Intel also disclosed at the investor conference that it is working to create an open ecosystem that is "open, optional, and trustworthy". Intel showed investors the IP, packaging, and foundry ecosystem platforms that Intel will build in the next few years.
Among them, IP includes mainstream IP such as x86, Arm, RISC-V and other special chip IP. The foundry aspect not only includes the Intel16, Intel3, and Intel18A process processes that Intel will put into production in the next few years, but also the mature foundry processes provided by the tower semiconductors that Intel has just acquired. At present, the mature process of high-tower semiconductors can already be used in chip foundries of 0.5 micron to 45 nanometer processes.
Accelerate the preparation of new business for the future, and it is expected to ship 4 million yuan this year
pat. Kissinger divided Intel's current business into traditional and emerging businesses according to the nature of the business. Traditional businesses include the Data Center and Artificial Intelligence (DCAI) Division, the Client Computing Division (CCG), and the Networking and Edge Division (NEX). Emerging businesses include Accelerated Computing Systems and Graphics (AXG), Intel Foundry Services (IFS), and Mobileye.
According to Intel's plans for the future, the total revenue of emerging businesses will account for more than half of Intel's revenue in the future, with the Accelerated Computing and Graphics division expected to generate more than $10 billion in revenue in 2026.
Intel also announced the release plan of its discrete graphics card "Ruixuan" at this conference. According to estimates from the Accelerated Computing and Graphics division, Intel will be able to ship more than 4 million independent GPUs this year. As early as the first quarter of 2022, OEMs will release laptops equipped with Intel Sharp Graphics. In addition, Intel has planned Celestial for the super enthusiast market, and the architecture development of the GPU has begun.
At the same time, Intel is also working to expand the scope of its IFS foundry services. The current trend of the automotive industry's transformation to intelligence has brought considerable growth to the automotive semiconductor industry, which is estimated to double over the next decade to $115 billion by 2030. At present, the supply chain of the industry is not yet perfect, and it cannot meet the needs of car companies that are easy to grow. Intel has noticed the gap in this market and is forming a dedicated automotive chip foundry team in the IFS division.
The IFS division will develop a high-performance, open automotive computing platform to help automotive OEMs build next-generation solutions. This open architecture will combine the building blocks of a chiplet with Intel's advanced packaging technology to meet the computing needs of the next generation of autonomous vehicles. At the same time, ifS is working with Mobileye to combine advanced process processes, technology optimization and advanced packaging to the vehicle-grade foundry platform.
Intel will also provide design services and IP licensing to automakers to enable them to use Intel's chip and system design capabilities. This is an extension of Intel's IFS acceleration program announced last year. The program is designed to give automotive chipmakers access to advanced process processes and packaging technologies.
summary
Since Intel unveiled its new business for IDM2.0 planning and IFS foundry services last year, it has been moving forward: from setting up a $1 billion fund to advance the design and development of advanced process chips to becoming a senior member of the RISC-V Foundation and joining the RISC-V architecture track. Then to the acquisition of February 15, 5.4 billion completed the acquisition of tower semiconductors, and opened the x86 instruction set authorization in ifS foundry services.
In this kind of action, it is not difficult to find Intel's firmness in the path of IDM2.0. At present, Intel has built an IP platform containing x86, Arm and RISC-V in the foundry industry, and has established cooperative relations with Cadence, Synopsys, Arm, Siemens EDA and other manufacturers, which shows Intel's future ambitions in the foundry industry.
At today's investor conference, Intel further revealed its future development in CPU, GPU, autonomous driving and other fields. It is not difficult to see that Intel's IFS business will benefit from these developments in the future.
Intel CEO Pat. Kissinger believes that Intel will usher in explosive growth in the foundry business in the next five years. He expects Intel to achieve a growth rate of 12% to 16% by 2026.
Whether Intel can finally achieve this goal, and whether it can truly achieve the grand blueprint of IDM 2.0, let us wait and see. Leifeng networkLeifeng networkLeifeng network