AMD Chairman and CEO Lisa Su unveiled the company’s latest AI chip, MI325X, at the opening of Computex Taipei on June 3. She emphasized that the MI325X boasts 30% faster computing speed compared to NVIDIA’s H200. According to a report from CNA, Su also announced that AMD plans to release a new generation of AI chips each respective year, hinting at a strong competitive stance against NVIDIA.

Lisa Su announced that the MI300 series is AMD’s fastest progressing product. The tech giant’s next-generation AI chip, MI325X, features HBM3e and is built on the CDNA 3 architecture.

According to Su, AMD MI325X outperforms NVIDIA’s H200 in performance and bandwidth, more than twice than that of NVIDIA’s H200. On the other hand, MI325X delivers 30% faster computing speed compared to the H200.

Furthermore, Su also announced that AMD will release MI350 in 2025, which will be manufactured with 3nm process, while MI400 is expected to follow, launched in 2026.

On June 3, Lisa Su stated that AMD will continue its collaboration with TSMC, advancing process technology to the 3nm and even 2nm nodes. Yet, Su did not directly address the previous market rumors suggesting that AMD might switch to Samsung’s 3nm technology.

Previously, as per a report on May 29th from The Korea Economic Daily, it has speculated that AMD is likely to become a customer of Samsung Electronics’ 3nm GAA process. Reportedly, during AMD CEO Lisa Su’s appearance at the 2024 ITF World, which was hosted by the Belgian microelectronics research center imec, Su revealed that AMD plans to use the 3nm GAA process for mass-producing next-generation chips.

Per the same report, Lisa Su stated that 3nm GAA transistors can enhance efficiency and performance, with improvements in packaging and interconnect technology. This will make AMD products more cost-effective and power-efficient. The report further addressed that, as Samsung is currently the only chip manufacturer with commercialized 3nm GAA process technology, Su’s comments were interpreted as indicating that AMD will officially collaborate with Samsung for 3nm production.

Read more

• [News] South Korean Media Suggests AMD Could Become Samsung’s 3nm GAA Process Customer
• [News] Samsung Reportedly Signs USD 3 Billion HBM3e Deal with AMD

(Photo credit: AMD)

According to a report from Chinese media outlet “Phoenix New Media,” Zhang Ping’an, Executive Director of Huawei and CEO of Huawei Cloud Computing Technologies, acknowledged that China is unlikely to achieve 3nm or 5nm processes and emphasized that solving issues relating to 7nm would be good enough.

Zhang pointed out that the semiconductor industry in China currently cannot directly compete with developed countries in cutting-edge processes, such as 3nm and 5nm. This is an indisputable fact, but it does not mean that China’s semiconductor industry has no prospects for development, the report noted.

Zhang believed that the semiconductor industry in China should be more focused on deepening efforts in relatively mature processes, such as 7nm, to enhance product performance and reliability, meeting the needs of the market and users.

Zhang further emphasized that the innovation direction of China’s semiconductor industry should not be limited to any single-point chip processes alone. Moreover, overemphasis on advanced processes may lead to the ignorance of optimization and innovation in system architecture, which potentially resulting to bottlenecks in overall performance.

He also stated that China’s semiconductor industry should pay more attention to innovation in system architecture. By optimizing the collaborative work between chips and systems, overall performance can be enhanced, thereby gaining greater competitive advantages in the global market.

Read more

• [News] Huawei, OrangePi Introduced Kunpeng Pro, with Mystery AI Processor Packaged Inside
• [News] NVIDIA Reportedly Facing Price Cut Pressure of H20 Chip in China Amid Competition with Huawei

(Photo credit: Huawei)

During NVIDIA founder and CEO Jensen Huang’s keynote speech on June 2, he shared insights on how the AI era is driving the development of a new global industrial revolution.

According to a report from TechNews, he covered various technologies and application areas, including advancements in accelerated computing, microservices, industrial digitalization, and consumer devices, which are expected to become key focus areas in the evolving AI market.

• Accelerated Computing

1. Collaboration between the computer industry and NVIDIA to build AI factories and data centers: NVIDIA and leading computer manufacturers worldwide announced today the launch of a series of systems based on the NVIDIA Blackwell architecture. These systems feature Grace CPUs, NVIDIA networking technologies, and infrastructures to assist enterprises in establishing AI factories and data centers.

2. Foxconn utilizes NVIDIA artificial intelligence and Omniverse technology to train robots and streamline assembly operations: Foxconn operates over 170 plants worldwide, with its latest being a virtual plant driving the latest developments in industrial automation technology.

The latest of Foxconn’s plant is a digital twin model of a new factory in Guadalajara, Mexico, a hub for the electronics industry. Engineers at Foxconn define processes and train robots in this virtual environment to enable physical factories to efficiently produce the next generation of accelerated computing engines, the NVIDIA Blackwell HGX system.

3. NVIDIA significantly strengthens Ethernet networks for generative artificial intelligence: NVIDIA announced widespread adoption of the NVIDIA Spectrum-X Ethernet platform and will accelerate the release of new products. CoreWeave, GMO Internet Group, Lambda, Scaleway, STPX Global, and Yotta are the first batch of AI cloud service providers to adopt NVIDIA Spectrum-X, bringing ultimate network performance to their AI infrastructure.

Additionally, NVIDIA’s partners have also released products utilizing the Spectrum platform, including ASRock Rack, ASUS, GIGABYTE Technology, Ingrasys Inc., Inventec, Quanta Cloud Technology, Wistron and Wiwynn. Moreover, Dell Technologies, Hewlett Packard Enterprise, Lenovo, and Super Micro Computer have collaborated with NVIDIA to incorporate the Spectrum platform into their respective products.

• Microservices

NVIDIA NIM has revolutionized deployment model: NVIDIA has announced that its inference microservice, NVIDIA NIM, optimized in container form, is now available for download by 28 million developers worldwide.

This allows deployment to cloud, data centers, or workstations, enabling developers to effortlessly build generative artificial intelligence applications for assisting partners, such as copilots and chatbots, within minutes, a process that previously took several weeks.

• Industrial Digitalization

1. Electronics manufacturers adopt NVIDIA AI and Omniverse to drive robotic factories and accelerate industrial digitization: NVIDIA announced that major Taiwanese electronics manufacturers, including Delta Electronics, Foxconn, Pegatron Corporation, and Wistron Corporation, are using NVIDIA’s technology to transform their factories into more autonomous production facilities through new reference workflows.

This workflow combines NVIDIA Metropolis visual artificial intelligence (AI) technology, NVIDIA Omniverse’ physically accurate rendering and simulation technology, and NVIDIA Isaac’s AI robot development and deployment technology.

2. Industry leaders adopt NVIDIA’s robotic technology to develop tens of millions of AI-supported autonomous machines, including BYD Electronics, Siemens, Teradyne Robotics, and Alphabet’s Intrinsic, among more than ten global leading companies in the robotics industry.

These companies integrate NVIDIA Isaac acceleration libraries, physically principled simulation content, and AI models into their software frameworks and robot models to enhance efficiency in factories, warehouses, and distribution centers. This enables human colleagues to work in safer environments and serves as intelligent assistants in executing repetitive or ultra-precise tasks.

3. NVIDIA introduces NVIDIA IGX with Holoscan support, enabling enterprise software to run medical, industrial, and scientific artificial intelligence applications in real-time at the edge: To meet the growing demand for real-time artificial intelligence computing technology at the industrial edge, NVIDIA announces the comprehensive launch of NVIDIA AI Enterprise-IGX software with Holoscan on the NVIDIA IGX platform.

• Consumer Devices

1. NVIDIA utilizes GeForce RTX AI PC to deliver the real AI assistant experience: NVIDIA announces the launch of the new NVIDIA RTX technology, designed to support AI assistants and digital human platforms running on new GeForce RTX AI laptops.

2. NVIDIA introduces Digital Human Microservices to lay the foundation for future generative AI digital avatars: NVIDIA announces the comprehensive rollout of NVIDIA ACE generative artificial intelligence microservices to accelerate the development of the next wave of digital humans and numerous breakthroughs in generative AI soon to be introduced on the platform.

Companies in customer service, gaming, and healthcare sectors are among the first to adopt ACE technology, making it easier to create, personalize, and interact with realistic digital humans. These microservices have broad applications in customer service, telehealth, gaming, and entertainment.

Read more

• [News] NVIDIA CEO Jensen Huang Announces the Latest Rubin Architecture – Rubin Ultra GPU to Feature 12 HBM4
• [News] Dell Leak Reveals NVIDIA’s Potential B200 Launch Next Year

(Photo credit: NVIDIA)

NVIDIA CEO Jensen Huang delivered a keynote speech at the NTU (National Taiwan University) Sports Center on June 2. As per a report from TechNews, during the speech, he unveiled the new generation Rubin architecture, showcasing NVIDIA’s accelerated rollout of new architectures, which became the highlight of the evening.

While discussing NVIDIA’s next-generation architecture, Huang mentioned the Blackwell Ultra GPU and indicated that it may continue to be upgraded. He then revealed that the next-generation architecture following Blackwell will be the Rubin architecture.

The Rubin GPU will feature 8 HBM4, while the Rubin Ultra GPU will come with 12 HBM4 chips, he noted.

The new architecture has been named after American astronomer Vera Rubin, who made significant contributions to our understanding of dark matter in the universe and conducted pioneering work on galaxy rotation rates.

Notably, despite the fact that NVIDIA has just launched the new Blackwell platform, it appears that NVIDIA keeps accelerating its roadmap. According to the latest information announced by Jensen Huang, the Rubin GPU will become part of the R series products and is expected to go into mass production in the fourth quarter of 2025. The Rubin GPU and its corresponding platform are anticipated to be released in 2026, followed by the Ultra version in 2027. NVIDIA also confirmed that the Rubin GPU will use HBM4.

Per a report from global media outlet Wccftech, NVIDIA’s Rubin GPU is expected to adopt a 4x reticle design and utilize TSMC’s CoWoS-L packaging technology, along with the N3 process. Moreover, NVIDIA will use next-generation HBM4 DRAM to power its Rubin GPU.

According to Wccftech, currently, NVIDIA employs the fastest HBM3e in its B100 GPU, and it plans to update these chips with the HBM4 version when HBM4 solutions are produced in large quantities by the end of 2025.

Read more

• [News] Dell Leak Reveals NVIDIA’s Potential B200 Launch Next Year
• [News] TSMC’s Advanced Packaging Capacity Fully Booked by NVIDIA and AMD Through Next Year

(Photo credit: NVIDIA)

In the past two years, the semiconductor industry has experienced a market downturn, a recovery slower than expected, and a cash crunch. Major companies such as Intel, TSMC, and Samsung, while continuing to advance their expansion projects, have been constantly adjusting and slowing down the pace and schedule of their fab construction to better serve their long-term development goals. It’s found that seven fabs worldwide are projected to delay construction.

• Intel’s 1nm Chip Plant in Germany and Ohio Plant in the US Delay Construction

According to a report from global media outlet Volksstimme, the construction of Intel’s Fab 29.1 and Fab 29.2 near Magdeburg, Germany, has been postponed due to pending approval of EU subsidies and the need to remove and reuse black soil. The date of commencement has been pushed from summer 2024 to May 2025.

Earlier reports indicated that the construction of this chip planr was initially expected to begin in 1H23, but due to subsidy delays, construction was put off to summer 2024. Moreover, the topsoil at the construction site cannot be cleared until May 2025 at the earliest.

It is reported that Intel’s Fab 29.1 and Fab 29.2 were originally scheduled to start operations by late 2027 and were expected to employ advanced manufacturing processes, potentially Intel 14A (1.4nm) and Intel 10A (1nm) process nodes. However, Intel now estimates that it will take four to five years to build these two plants, and production is expected to commence between 2029 and 2030.

• Samsung’s Chip Plant in Pyeongtaek, South Korea and Fab in Taylor, US Delay Construction

In February 2024, Samsung revealed that it had partially halted the construction of its fifth semiconductor plant in Pyeongtaek, Gyeonggi Province. Samsung originally planned to build six semiconductor plants on an 855,000 square meter site in Pyeongtaek, creating the world’s largest semiconductor hub. Currently, the P1, P2, and P3 plants at the Pyeongtaek park house the most advanced DRAM, NAND flash memory, and foundry production lines, while the P4 and P5 plants are under construction.

Samsung stated that the halt was for further inspection. However, industry sources have revealed that Samsung’s adjustment of the new production lines for P4 and P5 fabs is to prioritize the construction of the PH2 production line at P4 fab. It is reported that P4 plant might build PH3 production line to produce high-end DRAM to meet market demands.

Besides, South Korean media Businesskorea also revealed Samsung has postponed the mass production timeline of the fab in Taylor, Texas, US from late 2024 to 2026, which is possibly due to a slowdown in the wafer foundry market growth, and the delay was attributed to U.S. government subsidies and issues related to the complexities in gaining permits.

• TSMC Postponed the Production of Two Plants in Arizona, US

On April 9, TSMC announced the plan to build a third fab in Arizona. Once completed, this fab will use 2nm process or even more advanced technologies to manufacture wafers for customers. With this addition, TSMC’s total capital expenditure in Phoenix, Arizona, will exceed USD 65 billion.

Meanwhile, TSMC disclosed that their first fab in Arizona will start production in 1H25, using 4nm process. The second fab, initially announced to use 3nm process, will also incorporate the more advanced 2nm process, with mass production set to begin in 2028. This fab was announced in December 2020, which was originally scheduled to start mass production using 3nm process in 2026, primarily, but the latest schedule represents a delay of nearly two years from the original one.

As to the third fab planned to set up in Arizona, TSMC has not yet disclosed the date for construction. However, they mentioned that it will use 2nm process or more advanced ones, with production expected to commence in the late 2030s.

• Wolfspeed’s 8-Inch SiC Fab might Delay Construction to 2025

Wolfspeed’s 8-inch SiC fab in Ensdorf, Saarland planned to invest about EUR 2.75 billion, but the construction has been postponed. The project has already secured subsidies of EUR 360 million from the German federal government and EUR155 million from the Saarland government. In addition, Wolfspeed is also seeking financial assistance from the European Chips Act. ZF will provide Wolfspeed with several hundred million dollars of financial investment in exchange for a minority stake in the plant.

Industry sources indicate that Wolfspeed aims to secure more funding before the groundbreaking ceremony. If it fails to gain financial assistance from the European Chips Act, the project is very likely to be delayed. The plant was initially scheduled to start construction in summer 2024, but Wolfspeed CEO Gregg Lowe revealed that it might now begin in 2025.

Read more

• [News] Intel’s 1nm-class Fabs in Germany Reportedly Delayed Due to Black Soil Concerns and Pending EU Subsidy Approval
• [News] TSMC’s Nanjing Plant Reportedly Pushes for Indefinite Exemption From US Before the May 31 Deadline
  

(Photo credit: TSMC)