Memory chip shortages prompt price increases; Israeli chip foundry; 2 acquisitions; Baidu’s AI chips; IBM’s new quantum processor; GF’s GaN push; 3D NAND scaling boosters; U.S. policy recommendations; +$500M in fundings; SiPho and SiGe capacity; EV joint venture.
Samsung reportedly is hiking memory chip prices by 30% to 60% due to high demand from AI data centers and constrained supplies. Those shortages are causing ripples elsewhere. SMIC, China’s largest foundry, said its customers are holding back orders for other types of semiconductor due to concerns about memory supplies.
Meanwhile, interest in photonics and power semiconductors is picking up, as evidenced by this week’s investments:
Venture capital fund Awz Ventures is partnering with the Israeli government to build a national semiconductor foundry in Ashkelon, Israel. The foundry will produce III-V semiconductors primarily for Israel’s defense sector.
Tower Semiconductor, meanwhile, is allocating an additional $300M for silicon photonics and silicon germanium technologies and an expansion of its Newport Beach FAB 3.
GlobalFoundries signed an agreement to license TSMC’s 650V and 80V GaN technology to further its own development and production of next-gen power devices at its Vermont facility. Production is expected in late 2026.
CadenceacquiredChipStack, a provider of a generative AI-driven platform for chip verification, including intelligent test planning, test generation, and AI-assisted debugging.
Arm intends to acquireDreamBig Semiconductor, a provider of remote direct memory access (RDMA) networking solutions, including NICs and a chiplet platform for near memory compute, for approximately $265M in cash. The deal is expected to close by early 2026.
New processors:
IBM debuted a new quantum processor that has 120 qubits linked together with 218 tunable couplers to enable the execution of circuits with higher complexity. It also introduced an experimental processor with key components needed for fault-tolerant QC. The company is also shifting fabrication of its quantum processors to 300mm wafers in NY Creates’ Albany NanoTech complex.
Baidu announced two new AI processor chips — the M100 large-scale inference chip will launch early next year, and the M300 chip, tailored for training very large multimodal models with trillions of parameters, will launch in early 2027.
Big Fundings:
d-Matrix raised $275M for its data center inference platform, which combines digital in-memory compute accelerators with NICs and inference-optimized software.
Ferroelectric Memory Co. raised €100M (~$116.4M) to commercialize its nonvolatile ferroelectric hafnium oxide memory, which it suggests could replace DRAM and cache SRAM in certain applications.
Majestic Labs emerged from stealth with over $100M in funding for its AI server architecture, which features custom accelerator and memory interface chips that disaggregate memory from compute.
The Council on Foreign Relations‘ 95-page report on U.S. Economic Security recommends a number of targeted government actions in the chip industry and other tech sectors, particularly in the area of critical minerals, workforce development and onshoring of the manufacturing of critical inputs and components for semiconductors, including chemicals, PCBs, and IC substrates
Chinese hackers used Anthropic’s agentic AI tool to perpetrate a large-scale cyber attack without significant human interaction.
The SIA weighed in on the proposed value-based patent fee structure, saying “imposing a new tax on innovation would impede the competitiveness of the U.S. semiconductor industry and make the U.S. a less attractive place to develop new technology.”
ITIF explained why U.S. policymakers should keep semiconductor export controls to a minimum.
Anthropic plans to spend $50B on new data centers in the US.
Asia:
YMTC is building a third NAND flash fab in Wuhan, China, aiming to be operational by 2027.
ProteanTecs opened a new office in Yokohama, Japan, and named Noritaka Kojima as Japan GM and country manager.
A Dutch government delegation plans to travel to China next week in hopes of resolving the Nexperia issue. Early this week, China granted some export controls exemptions to help relieve the shortage in the automotive industry.
Baya Systemsexpanded to Cambridge, U.K., with a new office that will serve as a regional headquarters and strategic collaboration hub.
Microsoft plans to invest $10B in a Portugal data center park, and Google will invest $6B in Germany for data centers and other operations.
The UK and the Netherlands signed an Innovation Partnership deal, supporting joint efforts on semiconductors, quantum and AI.
ESIA issued a position paper on the EU’s Circular Economy Act, recommending the EU “abstain from applying strict material recovery targets and recycling requirements to the semiconductor manufacturing sector.”
Synaptics’ AI-ready touch and fingerprint sensing will combine with Qualcomm’s compute and biometric security technology to promote secure and intuitive user experiences.
UMC will mass-produceMetalenz’ polarization-based biometric solution, leveraging UMC’s high-volume 12-inch capacity and wafer-on-wafer bonding.
yieldWerx will integrate its analytics platform directly into iTest‘s production environment for real-time semiconductor test analytics.
Poet Technologies and QCi will jointly develop 400G/Lane TFLN modulator-based 3.2Tbps engines.
AMD finalized its acquisition of MK1, a provider of high-speed inference and reasoning-based AI technologies optimized for large-scale AMD hardware deployments.
Fig.1: Metalenz’s facial authentication solution, manufactured by UMC.
Market Reports
Taiwan‘s third quarter IC revenue (including design, manufacturing, packaging and test) totaled US$52B, a nearly 21% increase from the prior year, reports TSIA.
Virtual Twins: Layers Of Challenges: Lam Research’s David Fried discusses the complexity of connecting different layers to create high-fidelity models, what those layers need to include, and why standards are needed to maximize their benefits and simplify their creation.
Research
Imec is developing technologies to enable z-pitch scaling boosters for 3D NAND flash to allow more layers to be stacked at an economical cost.
Cornell researchers outlined an action roadmap to significantly curb environmental impacts of AI growth, highlighting that new data centers should be located in “windbelt” states and low-water stress states—Texas, Montana, Nebraska, and S. Dakota.
Abhishek Shendre of D2S describes how pixel-level dose correction (PLDC) computing can make curvy mask shapes manufacturable for improved wafer quality in a summary of his paper from SPIE Photomask.
Tachyum announced details and specifications for its 2nm Prodigy Universal Processor, which includes up to 21.3x higher AI rack performance than Nvidia Rubin Ultra NVL576.
Tsavorite Scalable Intelligenceemerged from stealth with a composable compute architecture that it claims unifies CPU, GPU, memory, and scale up/scale out connectivity in a single device.
Adoptions/Certifications:
BlaizeadoptedArteris’ FlexNoC 5 interconnect IP in its programmable platform for AI inference across a range of edge and hybrid infrastructure use cases.
Intrinsic SemiconductoradoptedCadence’s Spectre FX FastSPICE simulator to verify RRAM memory arrays.
Enphase Energy is using Infineon‘s CoolGaN technology for its next-gen solar microinverters.
Panmnesia announced the sampling availability of its PCIe 6.0/CXL 3.2 fabric switch that supports port-based routing.
Security and Privacy
Google released Private AI Compute in the cloud, an AI processing platform that combines Gemini models from the cloud with security and privacy assurances from on-device processing.
The Cyber Security and Resilience Bill was introduced to the U.K. Parliament. The proposed laws would provide better national security and cyber protection to digital and essential services, including health care, transport, energy and water.
Security Research:
Row Hammer Effect and Floating Body Effect of Monolithic 3D Stackable 1T1C DRAM (Georgia Tech)
Rivian and Volkswagen inked a transaction agreement to create a joint venture, Rivian and Volkswagen Group Technologies, with a total deal size of up to $5.8B. The companies plan to create next-generation electrical architecture and software technology for future EVs.
Swedish autonomous electric trucking firm Einride, valued at approximately $1.8B, is going public through a SPAC deal with Legato Merger Corp.
Toyota began production at its $13.9B North Carolina battery plant as it ramps up hybrid production.
Waymoexpanded its service range to freeways across the San Francisco Bay Area, Phoenix, and Los Angeles. The company plans to extend services to Austin and Atlanta in the future.
Yole Group conducted an analysis on automotive radar technology, market dynamics, and chipset design, concluding that despite volume normalization, growth remains steady as 166 million radar modules shipped globally in 2024.
Product news:
Infineon unveiled an advanced MCU designed for high-voltage Li-ion battery management in EVs, which will support zonal architectures and the transition to SDVs.
Teradar emerged from stealth with the first commercial sensing technology capable of seeing in the terahertz band of the electromagnetic spectrum. The technology provides up to 20 times the resolution of today’s automotive radar.
Micron announced shipping of qualification samples of its automotive UFS 4.1 solution to customers globally.
Quantum
Rigetti Computing expects to deliver a 100+ qubit chiplet-based quantum system by the end of 2025, scaling to a 1,000+ qubit system by the end of 2027.
IQM Quantum Computers is planning to release a 150-qubit quantum computer with error correction functionality by the end of 2026.
Post-quantum security BTQ Technologies will acquireQPerfect, which provides a neutral atom quantum emulator and control framework.
Microsoft opened a new quantum lab in Denmark to further develop topological qubits.
InP and SiPho join CMOS as critical technologies. Lasers, CPO and OCS will be everywhere (indium phosphide, silicon photonics, co-packaged optics, optical circuit switch).
Reducing variation in manufacturing, monitoring behavior over time, and targeting specific workloads can have a big impact on power, performance, and area/cost.
With SRAM failing to scale in recent process nodes, the industry must assess its impact on all forms of computing. There are no easy solutions on the horizon.
This website uses cookies to improve your experience while you navigate through the website. The cookies that are categorized as necessary are stored on your browser as they are essential for the working of basic functionalities of the website. We also use third-party cookies that help us analyze and understand how you use this website. We do not sell any personal information.
By continuing to use our website, you consent to our Privacy Policy. If you access other websites using the links provided, please be aware they may have their own privacy policies, and we do not accept any responsibility or liability for these policies or for any personal data which may be collected through these sites. Please check these policies before you submit any personal information to these sites.
Necessary cookies are absolutely essential for the website to function properly. This category only includes cookies that ensures basic functionalities and security features of the website. These cookies do not store any personal information.
Any cookies that may not be particularly necessary for the website to function and is used specifically to collect user personal data via analytics, ads, other embedded contents are termed as non-necessary cookies. It is mandatory to procure user consent prior to running these cookies on your website.
Leave a Reply