Chip Industry Week In Review

Applied Materials may scale back or cancel its $4 billion new Silicon Valley R&D facility in light of the U.S. government’s recent announcement to reduce funding for construction, modernization, or expansion of semiconductor research and development (R&D) facilities in the United States, according to the San Francisco Chronicle.

TSMC could receive up to $6.6 billion in direct funding under the CHIPS and Science Act, per an agreement with the U.S. Department of Commerce. The proposed funding would support TSMC’s investment of more than $65 billion in its Arizona fabs. “The CHIPS and Science Act provides TSMC the opportunity to make this unprecedented investment and to offer our foundry service of the most advanced manufacturing technologies in the United States,” said TSMC Chairman Mark Liu.

Arm, Intel, Google, and Meta introduced new chips for the AI market.

• Arm’s Ethos-U85 NPU includes a 4X performance enhancement and 20% greater power efficiency, as well as 2,048 MAC units, while providing support for AI frameworks including TensorFlow Lite and PyTorch.
• Google’s Axion processor is the company’s first Arm-based CPUs for data centers. Built with the Arm Neoverse V2 CPU, the processor has purpose-built silicon microcontrollers called Titanium , which offloads storage I/O processing to Google’s block storage service.
• Intel’s Gaudi 3 AI accelerator offers a 1.5X increase in memory bandwidth compared to its predecessor, while doubling networking bandwidth and four times the AI compute for BF16. The chip is manufactured on a 5nm process and allows activation of all engines in parallel.
• Meta’s Training and Inference Accelerator (MTIA) v1 was first announced last year. While exact specs were not released, the company said the chip was designed as a workhorse, as more Meta products begin incorporating AI and doubles the compute and memory bandwidth of their previous chips.

South Korea will invest nearly $7 billion into the country’s AI sector by 2027. The funding will include more than $1 billion to promote the growth of AI semiconductor firms.

Apple plans to overhaul its entire Mac line with its new in-house AI-focused M4 processor chips, according to Bloomberg.

Tesla settled its lawsuit over a 2018 fatal crash involving its Autopilot technology. Apple engineer Walter Huang died when his Model X SUV crashed into a highway barrier in California, with Autopilot switched on.

Quick links to more news:

Market Reports
Global
In-Depth
Education and Training
Security
Product News
Research
Events
Further Reading

Markets and money

SEMI reports worldwide sales of semiconductor manufacturing equipment edged down 1.3% to $106.3 billion in 2023 from an all-time record of $107.6 billion in 2022

EDA and IP revenue increased 14% YoY to $4.4 billion in Q4 of 2023. “This was very solid, well into the mid-double digits across all geographies and most categories,” said Walden Rhines, executive sponsor of the SEMI Electronic Design Market Report. “The one weak spot was layout, which was essentially flat. The place it hit hardest was the Asia/Pacific region, and particularly China.”

Fig. 1: Q4 revenue by segment and region, in $M. Source: SEMI Electronic Design Market Data

Revenue in the processor sector could hit $242 billion by 2028, Yole Group reported, basing its projection on an annual growth rate of 8%. The introduction of genAI into new smartphones and PCs could be a major growth driver.

TrendForce said the actual impact of the U.S.’s updated semiconductor export controls will be “insignificant.”

Global

In Portugal, Infineon and Amkor will operate a dedicated packaging and test center at Amkor’s manufacturing site. Operations are expected to commence in the first half of 2025.

Japan activities in the tech industry continue to heat up:

• Prime Minister Kishida of Japan met with U.S. President Biden to further their alliance, including joint efforts to address supply chain challenges in current and mature-node semiconductors.
• Rapidus formed a U.S. subsidiary and opened an office in Santa Clara, CA and named Henri Richard as general manager and president.
• Microsoft will invest US$2.9 billion over two years to increase its hyperscale cloud computing and AI infrastructure in Japan, plus expand digital skilling programs.
• Microchip and TSMC expanded their partnership to enable a specialized 40nm manufacturing capacity at JASM (Japan Advanced Semiconductor Manufacturing), TSMC’s majority-owned manufacturing subsidiary in Kumamoto Prefecture, Japan.
• Renesas restarted a previously shutter plant in Japan to meet power chip demand.
• Shin-Etsu Chemical will invest about 83 billion yen, or nearly US $550 million in a new plant to increase Japan’s semiconductor supply chain.
• Advantest has entered into a technical partnership with Toray Engineering to produce mini/microLED displays, including integrated data linkage solutions for manufacturing.

Taiwanese electronics manufacturers are increasing production of genAI servers, as smartphone and computer sales lag.

China’s Huawei says it will use Intel‘s AI chip in its PCs, despite blacklisting.

In the U.S. House of Representative, Adam Schiff (D-Calif.) introduced the Generative AI Copyright Disclosure Act, requiring notice of all copyrighted works used in building or altering the training dataset of an AI system, to also apply retroactively.

In-Depth

Semiconductor Engineering published its Test, Measurement and Analytics and Low Power-High Performance newsletters this week, featuring these articles:

• IC Test and Quality Requirements Drive New Collaboration
• Progress in Wafer And Package Level Defect Inspection
• New Strategies For Interpreting Data Variability
• Memory On Logic: The Good And Bad
• Using AI/ML to Minimize IR Drop
• Linear Drive Optics May Reduce Data Latency
• Startup Funding: March 2024.

Security

VU Amsterdam reported a new cross-privilege Spectre v2 vulnerability. “CPU hardware utilizing speculative execution that are vulnerable to Spectre v2 branch history injection (BHI) are likely affected,” states the CERT release.

University of Florida researchers presented a chiplet hardware security module to mitigate security vulnerabilities in SiP systems. Boise State University and Rensselaer Polytechnic Institute researchers discovered a covert channel exploiting CPU clock modulation.

The Cybersecurity and Infrastructure Security Agency (CISA) updated its malware analysis system, Malware Next-Gen, allowing organizations to submit malware samples and other suspicious artifacts for analysis, and issued other alerts.

Infineon announced its PSoC Edge E8x MCUs meet the highest certification level of the Platform Security Architecture (PSA) Certified program for IoT embedded security.

Ansys joined BAE Systems’ Mission Advantage Program to help the U.S. Department of Defense (DoD) move away from manual processes and adopt digital engineering and model-based systems engineering (MBSE) for security and defense purposes.

Synopsys released a software composition analysis, called Black Duck Supply Chain Edition, providing a view of software risks inherited from open-source, third-party, and AI-generated code, enabling teams to identify and resolve security vulnerabilities.

Education and Training

SIA announced its Workforce Policy Blueprint, with recommendations for: building the supply of engineers and scientists; improving and simplifying training of skilled technicians; and reducing financial barriers to entering the industry.

Japan and the U.S. jointly announced  a new AI partnership with University of Washington and University of Tsukuba and between Carnegie Mellon University and Keio University, supported by $110 million in combined private sector investments from NVIDIA, Amazon, Arm and other companies.

Japan, the Philippines and the U.S. announced a semiconductor workforce development initiative with students from the Philippines to train at American and Japanese universities.

NVIDIA partnered with Georgia Tech’s College of Engineering on a new program giving students access to artificial intelligence tools rarely found in the classroom. Students will work with NVIDIA AI Enterprise software, which is based on AI infrastructure from Penguin Solutions.

Deloitte Digital launched an academy for Apple Vision Pro to prepare for spatial computing.

Product News

Flex Logix is developing its EFLX eFPGA generation 3.0 for TSMC N5/4 and N3.

Infineon had several product announcements this week:

• PSoC 4 HVMS automotive MCUs with a 12 V-regulator and LIN/CXPI-transceiver in small footprint QFN packages with wettable flanks;
• Rust ecosystem expansion for its AURIX MCUs;
• AURIX TC4x MCUs for embedded AI and automotive applications, achieving SAFE AI compliance proposed by Fraunhofer ICS;
• XENSIV Sensor Shield for Arduino to evaluate smart sensor systems in smart home and consumer applications.

Renesas introduced RA0 ultra-low power MCUs based on the Arm Cortex-M23 processor, targeted at battery-operated consumer electronics devices, small appliances, industrial system control, and building automation. The company also expanded its cloud-based Quick Connect Studio for rapid prototypes and co-optimization of software and hardware.

QP Technologies‘ Escondido, CA facility was certified ANSI/ESD S20.20 compliant, a standard  for establishing, implementing, and maintaining an electrostatic discharge (ESD) control program.

Ansys released AnsysGPT, its AI-powered virtual assistant, trained on the company’s data to provide rapid responses to client’s questions.

Micron released its automotive grade quad-port SSD for software-defined vehicles (SDVs), and announced that it has qualified automotive-grade memory and storage solutions for Qualcomm’s Snapdragon Digital Chassis.

AI semiconductor startup Rebellions selected Arteris’ FlexNoC interconnect IP and Magillem Connectivity and Registers for its new AI Hardware Accelerator Neural Processing Unit. The units are aimed at increasing data processing power with minimum power consumption while also cutting down on latency found in Generative AI and LLM applications.

AMAZEMET adopted Siemens Xcelerator portfolio to help build its closed-loop production chain for metal additive manufacturing materials and support post-processing equipment.

NTT announced the successful demonstration of an All-Photonics Network (APN)-driven hyper low-latency connections between data centers in the United States and United Kingdom.

Research

KU Leuven, imec, and other researchers proposed a new semiconductor defect inspection system for high-NA EUV.

Researchers at the Lawrence Berkeley National Laboratory were able to take atomic-level images of a chiral interface, in which electrons travel in just one direction, reducing resistance. The images could represent an important breakthrough in quantum computing.

Argonne National Laboratory researchers developed a robotic pendant drop to study the nanoscale structure of liquids without using a container.

Researchers from Improbable AI Lab at MIT and the MIT-IBM Watson AI Lab used ML to improve red-teaming to prevent an AI chatbot from giving toxic responses.

Events

Find upcoming chip industry events here, including:

Upcoming webinars are here.

Further Reading

Read the latest special reports and top stories, or check out the latest newsletters:

Systems and Design
Low Power-High Performance
Test, Measurement and Analytics
Manufacturing, Packaging and Materials
Automotive, Security and Pervasive Computing