AP Memory Technology reported record-breaking third-quarter 2026 revenue and net profit, driven by surging demand for its integrated passive device (IPD) silicon capacitors under the S-SiCap product family. Order visibility now extends through the second half of 2027. The company is also pursuing a dual-track strategy with UMC. It projects that, as manufacturing capacity expands, silicon capacitor revenue will experience multi-fold growth and surpass its core IoT RAM revenue in 2027 — a forecast, not an achieved result.
Taiwan's National Science and Technology Council (NSTC), on November 11, 2026, proposed expanding the list of national core key technologies from 42 to 46 items, with four additions covering satellite communications, AI chip substrates, drones, and biotechnology. The draft is still under public consultation and must be reviewed by a committee before the Executive Yuan can promulgate it.
Intel's Xeon 7 "Diamond Rapids" has made its first public appearance as actual silicon: at the Tencent Global Digital Ecosystem Conference in Shenzhen on October 29, Intel displayed two packages, one showing the heat spreader and one flipped to show the contact side, though photos are only now circulating. The chips are meant for the massive LGA 9324 socket on Intel's Oak Stream platform. The specs are not new, as Intel detailed the package design at HOT CHIPS back in August: the flagship packs up to 256 Panther Cove P-cores without SMT, spread across 16 compute chiplets built on Intel 18A-P, with 1.28 GB of last-level cache, 16 memory channels with MRDIMM support up to 12,800 MT/s, and 128 PCIe Gen 6 lanes. At the event, Intel pitched Diamond Rapids as a CPU designed for agentic AI infrastructure and reportedly announced joint development with Tencent Cloud for its CVM instances; however, the new Tencent Cloud instances launched at the same event run on Xeon 6+ "Clearwater Forest" chips.
TDK Corporation (TSE:6762) and Microchip Technology Inc. (NASDAQ: MCHP) are demonstrating a distributed sensor-node platform for next-generation robotic systems at electronica in Munich this week (TDK booth B5.520). The platform combines sensing technology from InvenSense, a TDK group company, with Microchip's dsPIC33A Digital Signal Controller (DSC) and LAN8651/LAN8679 based 10BASE-T1S single-pair Ethernet connectivity for real-time intelligence and edge AI processing for robotics applications. The demonstration board combines a Microchip dsPIC33A DSC, LAN8651/LAN8679 10BASE-T1S connectivity, a TDK InvenSense ICM-42688-P 6-axis inertial measurement unit (IMU), and a MikroE expansion connector for compatible additional TDK sensors. The dsPIC33A DSC can process and fuse the InvenSense IMU data locally before transmitting relevant information over the 10BASE-T1S network, reducing latency, network traffic, and the computational workload of the robot's central controller. The source states that today's humanoid robots may incorporate 100 or more sensors, and the platform is designed to help robotics developers integrate a greater number of intelligent sensor nodes while simplifying cabling and communications. Massimo Mascotto, director of product marketing for InvenSense, said the partnership with Microchip aims to solve the real customer need of easily integrating a multitude of sensors with dedicated, distributed processing capabilities for fast local intelligence. The Microchip dsPIC33A DSC combines a 200 MHz 32-bit core, DSP acceleration, and a double-precision floating-point unit with high-speed ADCs and high-resolution PWM capabilities. Additional TDK sensing solutions for smart perception in robotics, including TMR and Hall sensors for ultra-precise angle detection, MEMS microphones for acoustic activity detection, temperature sensors for environmental monitoring, pressure sensors for tactile sensing, and 9-axis IMU+TMR sensing for exact location and orientation detection, will be demonstrated at electronica, Munich, November 10-13.
TDK InvenSense sensors (ICM-42688-P 6-axis IMU)Microchip dsPIC33A Digital Signal Controller (DSC)LAN8651/LAN8679 10BASE-T1S single-pair Ethernet connectivityTDK Corporation (TSE:6762), Microchip Technology Inc. (NASDAQ: MCHP) and 3 totalEE JournalPartnership/M&A
Related manufacturers
TDK Corporation (TSE:6762), Microchip Technology Inc. (NASDAQ: MCHP), InvenSense (a TDK group company)
According to a report published 11 November 2026, AMD detailed its predictive memory engine MEXT: the technology offloads cold memory pages to flash storage and uses AI models, with a primary transformer-based LLM, to serve pages before they are needed. AMD and MEXT founder Gary Smerdon claim it can cut the physical DRAM required in a server by 20% to 75% depending on the application and performance target, with the missing memory still registering as installed DRAM to the OS. In internal workloads such as Synopsys PrimeTime, AMD says it achieves 95% of full-server performance at 50% DRAM capacity and 80% at 25% capacity; in a 3-node Azure cluster with 5th-gen EPYC CPUs running the Spark SQL workload from the TPC-DS benchmark, MEXT offers 99% of full-cluster performance at 50% DRAM capacity; for agentic AI, across a five-agent concurrent geomean AMD claims 99% performance at 50% memory capacity and 85% at 25%, and 90% and 88% for 50% and 25% capacity respectively with a "memory-hungry" agent. AMD has deployed MEXT internally on 145 servers and says it will expand to 2,000 servers next year (2027 per the headline); it has also announced a partnership with HPE to bring MEXT to customers. The rollout is currently in phase 1, with some EPYC Venice servers carrying a "MEXT Ready" badge and a dedicated NVMe drive or bay; phase 2 will bring MEXT licenses and the RAM View application to these servers, but AMD has not said when phase 2 will begin. MEXT will be sold as a license bundled with RAM View, is EPYC-exclusive for now, and AMD says it trains in under a second when first deployed, operates on one EPYC core without a GPU, and continually trains as the workload progresses. The report frames this against an AI-driven DRAM pricing surge affecting enterprises, but notes AMD has not shared the benchmark footnotes and has not said when MEXT will be ready for deployment beyond AMD's walls.
NVIDIA's rumored GeForce RTX 5080 24 GB model has reportedly been put on hold. According to a report from Board Channels, NVIDIA has already informed its add-in card (AIC) partners to officially cancel the January 2027 launch plan for the 24 GB model, citing economic challenges. The company is struggling to secure enough 3 GB GDDR7 memory modules and is instead opting for the less expensive 2 GB modules for the standard 16 GB version, which will remain available. The source states that, amid one of the worst supply chain shortages for memory modules, using the more expensive 3 GB GDDR7 modules would result in a significantly higher MSRP and consequently a higher price for customers. Reportedly no other changes were planned besides the memory configuration: the CUDA core count was to remain the same, and the 24 GB model would potentially have slightly higher TDP and clock speeds, with the main improvements coming from increased memory capacity and bandwidth. Since the revised model will apparently not arrive and there is no rumor of a "SUPER" overhaul anytime soon, the standard lineup will remain until the next-generation GeForce RTX 60-Series "Rubin" arrives.
Nexperia is expanding its gallium nitride (GaN) and silicon carbide (SiC) power semiconductor portfolios with new devices targeting industrial, automotive, and computing applications. On the GaN side, new ultra-low-resistance GaN FETs cover 18 V to 200 V: the flagship GANE0R8-100QTM is a 100 V device with typical RDS(on) of 0.75 mΩ in a 3 mm × 5 mm PQFN package and is now available, along with five other products spanning 25 V to 200 V with resistance values down to 0.37 mΩ; two further devices, including the 18 V GANE0R3-015QTL with typical RDS(on) of 0.32 mΩ, are scheduled for Q1 2027. On the SiC side, Nexperia has introduced 750 V SiC trench MOSFETs for EV charging infrastructure, photovoltaic inverters, industrial drives, AI and high-performance computing power supplies, onboard chargers, and DC-DC converters, and the family also introduces TOLL packaging to Nexperia's portfolio; the 750 V devices are available now in industrial and AEC-Q101-qualified automotive versions. The source does not state the cause (this is a product portfolio expansion announcement, involving no price movement or capacity figures).
GaN FETs750 V SiC trench MOSFETsNexperiaPower Electronics News
According to the source, AMD is promoting software technology it obtained through its acquisition of Mext (acquired back in June; the source gives no year), claiming that buying a Mext license enables up to 99 percent of the performance with half the DRAM on Epyc servers, by using AI models to predictively load memory pages from cheaper NVMe flash storage back into DRAM before they are needed, transparently making flash appear as DRAM. According to former Mext CEO and now AMD CVP of business development Gary Smerdon, replacing costly RDIMMs with flash storage could cut enterprise server costs by as much as 83 percent; of the initial 145 systems the software has been deployed on so far, he estimates customers have already saved $5 million in avoided DRAM costs. He cited a Meta research paper showing that between 50 and 80 percent of memory pages had not been touched in five minutes, ten minutes or longer, and noted that flash is 50x cheaper per bit and 30x lower power per bit but 500x slower. On performance: in tests against Synopsys' PrimeTime EDA suite, half the memory can be offloaded to flash while achieving 95 percent of DRAM performance; in AMD's tests, with 25 percent DRAM and 75 percent flash, performance fell to 80 percent of the DRAM-only baseline; in Apache Spark scenarios, AMD's Mext team says it is seeing 99 percent of the performance even after replacing half the DRAM. The software will be exclusive to AMD hardware and offered as a license sold through OEM partners; Linux systems require only a kernel driver, a separate RAM View application is offered to help enterprises decide which systems to enable; no GPU is required, but in virtualized environments the software requires one core per VM or VPC. AMD is currently in the first phase of its rollout, working with OEMs like HPE to make systems so-called Mext ready; in the second phase it aims to begin selling licenses and shipping Epyc platforms with RAM View preinstalled. AMD is betting customers will accept some performance trade-off for substantial DRAM savings, as memory remains constrained thanks to the AI datacenter buildout.
German equipment maker SCHMID Group has received an order for wet-process equipment for a new glass core substrate plant in South Asia. The customer is a US company specialising in glass-based advanced packaging, which is now moving its technology from pilot production to industrial scale. According to the customer's own public announcements and SCHMID's press release, the plant will be South Asia's first advanced packaging site built around glass substrates and 3D heterogeneous integration; once fully ramped up, it is designed to produce around 70,000 glass panels a year, as well as glass interposers and 3D integration modules for AI, data centres, 5G and aerospace and defence. SCHMID will supply equipment for critical wet-process steps, which the company says are decisive for yield, reliability and cost, and the order will be booked in the fourth quarter of 2026. SCHMID names neither the customer nor the country, and does not disclose the value of the order. The stated capacity is a designed target, not an achieved result.
Intel announced it is expanding its CPU and GPU portfolio with the Intel GPU D-series, code-named Crescent Island, and next-generation Xeon processors, code-named Diamond Rapids, targeting large-model inference and enterprise high-performance orchestration workloads. The company outlined the new GPU D-series technology at the 2026 Open Compute Project (OCP) Global Summit as it pursues server CPU upgrade opportunities driven by agentic AI. This is a product roadmap disclosure (planned status); no specific supply, capacity or price impact is stated.
IEMC chairman Charles Kau said TSMC cannot make 3D DRAM, dismissing reports that TSMC is evaluating entry into the next-generation memory market at Nvidia's request. The report remains market speculation; the IEMC chairman's remark is a denial from one party, and TSMC itself declined to comment on market speculation and stressed that it remains focused on its core foundry business. The source does not state the cause.
Google Cloud is pairing its TPU 8i systems with Axion processors that are custom Arm Neoverse-based CPUs. With TPU 8i Max, Google is extending its use of Axion CPUs as the host compute for its AI accelerators, building on Axion's role in TPU 8t and TPU 8i. Beyond this, the source states that Google Cloud's continued investment reflects an industry shift toward pairing CPUs with accelerators to unlock the next era of AI inference, and that Google Axion also powers general-purpose cloud instances that run agent runtimes and orchestration frameworks.
Siemens is to buy Lauterbach, a debug and trace specialist, to combine complementary technologies for a unified view of chip and software behavior, enabling earlier development and verification of embedded software before physical chip availability.
Intel is expanding its computing platform strategy for agentic AI and enterprise workloads with the Intel Core Series 3 processor, codenamed Wildcat Lake, targeting mainstream consumer and commercial notebooks. The processor combines Intel 18A and TSMC 6nm processes in a two-chiplet design, using Universal Chiplet Interconnect Express (UCIe) standard chiplet interconnects and organic-substrate multi-chip packaging (MCP) to lower the cost barrier for bringing advanced process technology to mainstream PCs. This is a platform/product strategy disclosure of a planned nature; the source does not state a production timeline.
Intel Core Series 3 processor (codenamed Wildcat Lake)Intel, Taiwan Semiconductor Manufacturing Co. (TSMC)DigiTimes Asia
On Nov. 9 (local time), AMD announced that it will supply its EPYC server CPUs for Google's next-generation AI computing system. The system is Google's next-generation TPU 8tx, based on the eighth-generation Tensor Processing Unit, combining Google's TPU with AMD's fifth-generation EPYC processor, codenamed Turin, and ultra-high-speed networking. EPYC serves as the host CPU, overseeing overall system operations including AI workload orchestration, data preparation and delivery, and high-speed communications management. The system also incorporates AMD Secure Encrypted Virtualization (SEV), built into EPYC CPUs, extending protection that previously covered only the CPU and its directly connected memory to the connected TPUs: data remains encrypted while stored in CPU memory and during transfers between the CPU and TPU, with hardware isolation preventing external access during computations inside the TPU. AMD states its EPYC processors are used by more than 10 major cloud service providers (hyperscalers) and advanced AI research laboratories, support systems based on AMD Instinct GPUs, Nvidia GPUs and ASIC chips, and are compatible with the existing x86 software ecosystem. AMD and Google have previously collaborated on deploying multiple generations of EPYC processors in Google Cloud and on enterprise security technologies. This is an announcement of a supply arrangement; the source does not state the cause, and no shipment volumes or financial terms are reported.
EPYC server CPUTPU (Tensor Processing Unit)AMD, GoogleThe ElecPartnership/M&A
Mitsui Chemicals announced it will acquire semiconductor materials maker JSR in a deal valued at 1.84 trillion yen in total enterprise value including assumed debt and other obligations, the largest M&A deal in the history of Japan's chemical industry, surpassing Showa Denko's (now Resonac) approximately 964 billion yen acquisition of Hitachi Chemical in 2020. Mitsui Chemicals' board of directors resolved on Nov. 9 to jointly acquire JSR shares held by JIC Capital, a subsidiary of Japan Investment Corporation (JIC), together with the Development Bank of Japan (DBJ), and a share transfer agreement was signed the same day; completion is targeted for the second half of fiscal 2027, between October 2027 and March 2028. JSR's equity value is 1.3921 trillion yen. Mitsui Chemicals will acquire 100% of JSR's voting rights and make it a consolidated subsidiary, financing the acquisition through 1.64 trillion yen in short-term borrowings from financial institutions and a 200 billion yen investment from DBJ; after closing, DBJ's common shares will be converted into nonvoting preferred shares. JSR subsidiaries, including metal oxide resist (MOR) maker Inpria, biopharmaceutical CDMO KBI Biopharma, JSR North America Holdings and JSR Life Sciences, will also become subsidiaries. The sale comes approximately two years after JIC took JSR private in 2024 for approximately 900 billion yen and delisted the company. The two companies plan to combine front-end materials capabilities (photoresists and pellicles) with back-end materials such as thick-film photoresists and ICROS tape, pursue cross-selling and jointly develop new products including materials for three-dimensional and high-density packaging. Mitsui Chemicals aims to generate an additional 40 billion yen in annual profit by fiscal 2035 (20 billion yen from increased semiconductor materials market share and new product sales, and 20 billion yen from optimizing R&D organizations and business locations), and to raise company-wide core operating profit, excluding one-time gains and losses, to more than 300 billion yen by fiscal 2030.
photoresist (PR)pelliclesthick-film photoresistsMitsui Chemicals, JSR and 8 totalThe ElecPartnership/M&A
Related manufacturers
Mitsui Chemicals, JSR, JIC Capital, Development Bank of Japan (DBJ), Inpria, KBI Biopharma, JSR North America Holdings, JSR Life Sciences
LG Display announced on Nov. 10 that it received the King Sejong Award, the top honor at the 2026 Patent Technology Awards organized by the Ministry of Intellectual Property, for its patent covering a highly integrated gate driving circuit for OLED TVs. The technology reduces OLED TV bezel width by approximately 34%: on a 65-inch OLED TV, bezel width decreases from 6.5mm to 4.3mm. Technically, LG Display consolidated redundant circuits in conventional GIP designs into shared core circuits, reducing the number of transistors in the circuit by 74% and the GIP area by approximately 40%. The technology has been applied to all OLED TV panels ranging from 42 to 97 inches since 2022, cumulative mass production has reached 28.3 million units, and it has generated cumulative cost improvements of approximately 385 billion won to date. Since last year, LG Display has also been expanding its application to OLED panels for gaming monitors.
OLED TV panelsGIP (Gate In Panel) circuitsgate driving circuitLG DisplayThe Elec
STMicroelectronics (ST) unveiled the STM32H7P industrial microcontroller (MCU) on Nov. 10, featuring hardware accelerators for digital signal processing (DSP) and artificial intelligence (AI) workloads. The new MCU combines a 520MHz Arm Cortex-M7 processor with a dedicated hardware signal processor (HSP) and a neural processing unit (NPU), and can be used in industrial equipment for applications including digital power, motor control and condition monitoring. ST said the HSP can reduce processing time by up to 80% compared with conventional CPU-based software processing; the NPU delivers AI computing performance of up to 85 giga operations per second (GOPS); and AI inference for arc fault detection in solar power systems runs 40 times faster than conventional software-based processing. The STM32H7P supports up to 2MB of embedded flash memory, integrates industrial peripherals including analog-to-digital converters (ADCs), digital-to-analog converters (DACs) and timers, and is designed to operate at temperatures of up to 125 degrees Celsius. ST is currently supplying samples to original equipment manufacturers (OEMs), with mass production scheduled to begin in January 2027; pricing for the LQFP100 package with 1MB of flash memory starts at $4.43 per unit for orders of 10,000 units.
Acbel Polytech expanded its operations in the third quarter of 2026, driven primarily by growth in enterprise power supplies as demand for fuel-cell and data-center power products increased. Per the title, its enterprise power share rose to 59%. Revenue growth and an improved expense ratio lifted operating profit significantly from the previous quarter, signaling a gradual recovery in core operations despite product-mix pressure on gross margin. The source does not state details beyond this cause, and no capacity, price or customer specifics are mentioned.
enterprise power suppliesfuel-cell power productsdata-center power productsAcbel PolytechDigiTimes Asia
Anthropic has filed for an initial public offering (IPO) to list on the Nasdaq under the ticker symbol ANT. The source describes this IPO filing as a defining milestone for both the company and the entire artificial intelligence landscape. The headline further references strategic tech alliances, massive compute commitments, and the dual realities of AI growth, but the body supplies only the core fact of the IPO filing; no specific chips, suppliers, quantities or timeline are provided, and the source does not state the cause or further details. This is a filed plan to go public and does not demonstrate that the listing has been completed.