Supply chain trends in large Semiconductor industry
US government issued new rules to restrict investment in China, to take effect early next year
According to the news of the official website of the Treasury Department of the USA, on October 28, 2024, the Treasury Department released the Final Rule to finalized the three investment areas prohibited in the Addressing United States Investments in Certain National Security Technologies and Products in Countries of Concern（the Outbound Order), including semiconductors and microelectronics, artificial intelligence (AI), and quantum computing, and will implement the Order from January 2nd next year.
1. Semiconductor and microelectronics: Prohibited transactions: Covered transactions related to certain electronic design automation software; certain manufacturing or advanced packaging tools; the design or manufacturing of certain advanced integrated circuits; the advanced packaging technology for integrated circuits; and supercomputers are prohibited. Notifiable transactions: The transactions related to the design, manufacturing or packaging of integrated circuits not covered by the prohibited transactions are subject to the notification requirements.
2. Quantum information technology: Prohibited transactions: Covered transactions related to the development or manufacturing of any critical components required for quantum computers; The development or fabrication of certain quantum sensing platforms; and the development or fabrication of certain quantum networks or quantum communication systems.
3. Certain artificial intelligence (AI) systems: Prohibited transaction: Covered transactions related to the development of any AI system designed or intended to be used exclusively for certain end uses are prohibited. In addition, covered tractions related to the development of any AI system that is trained for more than 10 25 computational operations, or primarily using biometric data and trained for more than 1024 computational operations. Notifiable transactions: Covered transactions related to the development of any AI system not covered by the prohibited transactions, where such AI system refers to the system designed or intended for use in certain end uses or applications, or is trained for computing capacity greater than 10 23 computational operations are subject to notification requirements.
The implementation of the new rule shows that the US attaches importance to its leading position in the global science & technology field. The US government believes that restricting investment in China in key technology areas is to prevent US capital and expertise from being used to help China develop key technologies that may bring military advantages。This indicates that competition in science and technology has become closely linked to national security and has become an important topic in international relations.
The new rule could have far-reaching impact on the global supply chains. The intensified high-tech competition between China and the USA may lead to further fragmentation of the global supply chains and affect the landscape of the global market. The fragmentation could force other countries and enterprises to choose standing with China or the USA, thus affecting global trade and investment flows. Facing external pressures, China is likely to make more efforts to promote independent innovation and reduce its dependence on foreign technology and investment in the long term, this may encourage the Chinese government and enterprises to speed up the pace of independent R&D and promote the development of local science and technology industries.
Company Trend（October 29）
Infineon unveils the world’s thinnest silicon power wafer
According to the news of its official Wechat account, Infineon unveiled its breakthrough in treating and processing the world’s thinnest silicon power wafer on October 29. The wafer is 300mm in diameter and 20μm thick, only a quarter of the thickness of a hair, and half the thickness of the current most advanced 40-60μm wafer. Following unveiling the world's first 300mm GaN power semiconductor wafer and the construction of the world's largest 200mm SiC power semiconductor fab in Kulim, Malaysia, Infineon achieved a new milestone in semiconductor manufacturing technology again.
It is learned that since the thickness of the metal layer used for fixing chip onto a wafer is larger than 20μm, to overcome the technical obstacle of reducing the wafer thickness to 20μm, Infineon’s engineers have to create an innovative and unique wafer grinding method. This greatly affects the treating and processing of the back of thin wafers. In addition, the technical and production-related challenges such as warping and separation of wafers, have a significant impact on the back-end assembly process that ensures the stability and superior robustness of wafers. The 20μm thin wafer process, based on Infineon's existing manufacturing technology, ensures that the new technology can be seamlessly integrated into the existing mass Si production lines, without creating additional manufacturing complexity, ensuring the highest possible yield and supply safety.
This innovation will help significantly improve the energy efficiency, power density and reliability of power conversion solutions for AI data centers, as well as consumer, motor control and computing applications. Compared to traditional silicon wafer-based solutions, to half wafer thickness can reduce the substrate resistance by 50%, thus reducing power loss in the power system by more than 15%. For high-end AI server applications, larger current will demand more energy, so reducing the voltage from 230V to a processor voltage of below 1.8V is particularly important for power conversion. Ultra-thin wafer technology greatly facilitates the vertical power transmission design based on vertical trench MOSFET technology. This design realizes the tight connection between wafer and AI chip processor, improves overall efficiency, and reduces power loss.
Company Trend（October 30）
TSMC's Kaohsiung plant to mass produce 2nm chips in 2025
Recently, it was reported that TSMC's first 2nm plant (P1) in Kaohsiung is about to be completed. It is expected that multiple parties will be invited to participate in an inauguration ceremony on November 26 and the machine installation will be started on December 1.
TSMC did not respond positively to the news but just said that its Kaohsiung Wafer Plant was announced in November 2021 and started in 2022, and that currently its construction is in progress with public infrastructure available now. TSMC also stressed that the development of its 2nm process technology goes smoothly. The performance and yield of its 2nm chips were achieved as planned, and some even have better performance. Therefore, the 2nm process chips will be mass produced in 2025 as scheduled, with expected production curve similar to that of 3nm process.
Previously, TSMC's two 2nm production bases are located in Baoshan and Kaohsiung in Hsinchu Science Park. The initial monthly production capacity of the two fabs will be around 30~35 thousand wafers. Its Baoshan plant is expected to start the mini line by the end of the year, and will start mass production in Q4 2025, while its Kaohsiung plant will start mass production in H1 2026.
According to TSMC, its 2nm process will adopt a new GAA architecture, whose performance will increase by 10% to 15% and its power consumption down up to 30% compared to 3nm process. In addition, TSMC will release BSPR technology based on 2nm process node, which will further increase chip density and speed. It is expected that 2nm chip will be mass produced in 2026.
TSMC's progress in 2nm process node shows its strong technical strength and forward-looking in terms of semiconductor manufacturing. The mass production of 2nm chip will not only drive the growth of high-performance computing applications （especially to meet the needs of HPC), but also have a profound impact on AI, 5G and IoT. The strategic move of TSMC not only strengthens its position in the global semiconductor supply chain, but also sets a new technology benchmark for the industry as a whole.
TSMC's 2nm process node chips are expected to outpace its 3nm process node chips due to the strong market demand. Technology giants such as Apple, Nvidia and AMD are expected to be the first customers of TSMC’s 2nm node chips, and as their demand for high-performance chips grows, 2nm node chips are expected to be more widely used. In addition, TSMC's breakthrough in 2nm process node technology has aroused widespread interest, because it not only excites the scientific and technological world but also arouses heated discussions about the future of computing power across the world.
Company Trend（November 1）
Siemens to buy Altair for US$10.6bn
A few days ago, Siemens announced that the company has signed an agreement to acquire Altair Engineering, a leading software provider in the industrial simulation and analysis market, for Us$10.6bn, and the deal is expected to be completed in H2 2025.
According to the agreement, Altair will receive a quote of US$113 per share, which means that the company’s value is about US$10bn (US$ 10.6bn in fact). With this acquisition, Siemens consolidated its position as a leading technology company and its leading position in the field of industrial software. Ralf P. Thomas, CFO of Siemens AG, said that the transaction is anticipated to add to Siemens' earnings per share in about two years from the deal's closing.
The President of Siemens AG said the acquisition of Altair was a significant milestone for Siemens, namely to accelerate customers’ digital and sustainable transformation by combining the real world and the digital world. Altair's capabilities in simulation, high-performance computing, data science and AI combined with Siemens Xcelerator will create the world's most complete portfolio of AI design and simulation products.
This acquisition took place during a period of major changes in the design and simulation software industry. For example, Synopsys plans to acquire Ansys for US$35bn, which will bring together two big players in EDA and simulation software. This shows the importance of software in advanced manufacturing. For companies like Siemens traditionally focusing on heavy equipment and hardware, their transition to a robust software portfolio is a natural response to the digital trend of manufacturing. With this acquisition, Siemens will integrate Altair's capabilities in simulation, data science, and its own Xcelerator to create the world's most complete AI design and simulation portfolio. By integrating Altair's highly complementary simulation portfolio, Siemens can strengthen its integrated digital twin and provides a more comprehensive physical-based simulation portfolio.
Domestic News（October 28）
Breakthrough achieved in key technology of photoresist based on domestic raw materials and self-designed formula
According to the news of the official Wechat account of Huazhong University of Science and Technology, recently, the research team of Wuhan National Laboratory for Optoelectronics of Huazhong University of Science and Technology of the university first solved the problem of raw materials and formula for photoresist, to help China break the bottleneck of raw materials for domestic chip making. According to the report, its self-developed T150A series photoresist products have passed the verification of semiconductor mass production and realized homemade raw materials for photoresist based on completely self-designed formula, which is expected to create a new future for homemade semiconductor photo-lithography.
The semiconductor-specific photoresist developed by the research team is to match international top enterprise's mainstream KrF photoresist series. Compared to a foreign product of the same series, T150A exhibits a maximum resolution of 120nm in the lithography process and shows greater process width tolerance, greater stability, better membrane retention, and even better performance in the etching process. The verification showed that the sidewall perpendicularity of the dense pattern in T150A perform excellent after etching. The team leader said that they developed the product by studying the molecular basis and raw materials of photo-lithography technology and finally obtained the formula with independent intellectual property, and this is only the beginning of their research road.
Domestic News（October 28）
Homemade 7nm autopilot chip powered on successfully!
SiEngine Technology recently announced that its all-scenario high-order autopilot chip "Star 1” was powered on successfully, and its performance exceeded all design goals. The chip will be mass produced in 2025 and will be used on a large scale in 2026.
"Star 1" comprehensively aligns with the world’s current most advanced autopilot product, and all its key indicators such as CPU performance, AI computing power, ISP processing power, and NPU local storage capacity surpass those of international advanced mainstream products. The chip is based on 7nm process node for automobiles, which meets AEC-Q100, and its multi-core heterogeneous architecture makes intelligent driving more powerful. Its CPU computing power is up to 250 KDMIPS, its NPU computing power is up to 512 TOPS, and it can realize up to 2048 TOPS through multi-chip collaboration. In terms of hardware configuration, Star 1 integrates high-performance VACC and ISP, built-in ASIL-D as well as rich interfaces to meet the L2 to L4 intelligent driving needs.
Previously, SiEngine Technology has released the 7nm intelligent cabin chip “LongYing 1". Data show that from January to August 2024, among the installed capacity of passenger vehicle cabin control chips in China, "Longying 1”as a homemade cabin chip, ranks first.
Domestic News（November 1）
EnnoCAD works with Tsinghua University to develop homemade EDA
According to the news of the official Wechat account of EnnoCAD, recently, EnnoCAD worked with Tsinghua University to jointly promote the in-depth integration of industry-university-research. As for low-power IC design, the team of EnnoCAD held a special lecture to make in-depth exchanges with the students and teachers at the classroom of the School of Integrated Circuits, Tsinghua University.
In this co-lecturing, the team of EnnoCAD gave an in-depth, high-level presentation on low-power IC design techniques. The lecturers started with the theoretical foundation of power consumption of digital circuits and gradually introduced the methodology and practical operation of low-power IC design. The students learned not only how to reduce power consumption by optimizing IC design but also knew the latest technology and trends in low power IC design. The purpose of this university-enterprise cooperation is to provide students with an in-depth understanding of the latest IC design technology and industry trends, and to bring more force to the local EDA industry.
The teaching and research staff of the School of Integrated Circuits of Tsinghua University expressed satisfaction and recognition for the application of relevant low-power consumption products in the design of IC and said that the technical strength of EnnoCAD is impressive, and said that the lecturing of EnnoCAD provided a rare learning opportunity for the students of Tsinghua University. In the future, Tsinghua University will carry out more cooperation with EnnoCAD to jointly promote the development of domestic EDA.
编辑 | 泓明数字营销部
声明：本公众平台发布内容（文字、图片及影音等素材）部分来源于网络，转载内容不代表本平台观点，如涉及版权请直接与我们联系。
