News Posts matching #HBM

Return to Keyword Browsing

Marvell Develops Industry's First 2 nm Custom SRAM

Marvell Technology, Inc., a leader in data infrastructure semiconductor solutions expanded its custom technology platform with the launch of the industry's first 2 nm custom Static Random Access Memory (SRAM), designed to boost the performance of custom XPUs and devices powering cloud data centers and AI clusters. Combining advanced custom circuitry and software from Marvell with core SRAM and cutting-edge 2 nm process technology, Marvell custom SRAM delivers up to 6 gigabits of high-speed memory while significantly reducing memory power consumption and die area at comparable densities.

Custom SRAM is the latest Marvell innovation designed to enhance memory hierarchy performance within accelerated infrastructure. Marvell previously introduced its CXL technology for integration into custom silicon to add terabytes of memory and supplemental compute capacity to cloud servers and unveiled custom HBM technology that increases memory capacity by up to 33% while reducing the space and power required for dense high-bandwidth memory (HBM) stacks inside XPUs.

Micron HBM Designed into Leading AMD AI Platform

Micron Technology, Inc. today announced the integration of its HBM3E 36 GB 12-high offering into the upcoming AMD Instinct MI350 Series solutions. This collaboration highlights the critical role of power efficiency and performance in training large AI models, delivering high-throughput inference and handling complex HPC workloads such as data processing and computational modeling. Furthermore, it represents another significant milestone in HBM industry leadership for Micron, showcasing its robust execution and the value of its strong customer relationships.

Micron HBM3E 36 GB 12-high solution brings industry-leading memory technology to AMD Instinct MI350 Series GPU platforms, providing outstanding bandwidth and lower power consumption. The AMD Instinct MI350 Series GPU platforms, built on AMD advanced CDNA 4 architecture, integrate 288 GB of high-bandwidth HBM3E memory capacity, delivering up to 8 TB/s bandwidth for exceptional throughput. This immense memory capacity allows Instinct MI350 series GPUs to efficiently support AI models with up to 520 billion parameters—on a single GPU. In a full platform configuration, Instinct MI350 Series GPUs offers up to 2.3 TB of HBM3E memory and achieves peak theoretical performance of up to 161 PFLOPS at FP4 precision, with leadership energy efficiency and scalability for high-density AI workloads. This tightly integrated architecture, combined with Micron's power-efficient HBM3E, enables exceptional throughput for large language model training, inference and scientific simulation tasks—empowering data centers to scale seamlessly while maximizing compute performance per watt. This joint effort between Micron and AMD has enabled faster time to market for AI solutions.

AMD Previews 432 GB HBM4 Instinct MI400 GPUs and Helios Rack-Scale AI Solution

At its "Advancing AI 2025" event, AMD rolled out its new Instinct MI350 lineup on the CDNA 4 architecture and teased the upcoming UDNA-based AI accelerator. True to its roughly one-year refresh rhythm, the company confirmed that the Instinct MI400 series will land in early 2026, showcasing a huge leap in memory, interconnect bandwidth, and raw compute power. Each MI400 card features twelve HBM4 stacks, providing a whopping 432 GB of on-package memory and pushing nearly 19.6 TB/s of memory bandwidth. Those early HBM4 modules deliver approximately 1.6 TB/s each, just shy of the 2 TB/s mark. On the compute front, AMD pegs the MI400 at 20 PetaFLOPS of FP8 throughput and 40 PetaFLOPS of FP4, doubling the sparse-matrix performance of today's MI355X cards. But the real game-changer is how AMD is scaling those GPUs. Until now, you could connect up to eight cards via Infinity Fabric, and anything beyond that had to go over Ethernet.

The MI400's upgraded fabric link now offers 300 GB/s, nearly twice the speed of the MI350 series, allowing you to build full-rack clusters without relying on slower networks. That upgrade paves the way for "Helios," AMD's fully integrated AI rack solution. It combines upcoming EPYC "Venice" CPUs with MI400 GPUs and trim-to-fit networking gear, offering a turnkey setup for data center operators. AMD didn't shy away from comparisons, either. A Helios rack with 72 MI400 cards delivers approximately 3.1 ExaFLOPS of tensor performance and 31 TB of HBM4 memory. NVIDIA's Vera Rubin system, slated to feature 72 GPUs and 288 GB of memory each, is expected to achieve around 3.6 ExaFLOPS, with AMD's capabilities surpassing it in both bandwidth and capacity. And if that's not enough, whispers of a beefed-up MI450X IF128 system are already swirling. Due in late 2026, it would directly link 128 GPUs with Infinity Fabric at 1.8 TB/s bidirectional per device, unlocking truly massive rack-scale AI clusters.

Compal Optimizes AI Workloads with AMD Instinct MI355X at AMD Advancing AI 2025 and International Supercomputing Conference 2025

As AI computing accelerates toward higher density and greater energy efficiency, Compal Electronics (Compal; Stock Ticker: 2324.TW), a global leader in IT and computing solutions, unveiled its latest high-performance server platform: SG720-2A/ OG720-2A at both AMD Advancing AI 2025 in the U.S. and the International Supercomputing Conference (ISC) 2025 in Europe. It features the AMD Instinct MI355X GPU architecture and offers both single-phase and two-phase liquid cooling configurations, showcasing Compal's leadership in thermal innovation and system integration. Tailored for next-generation generative AI and large language model (LLM) training, the SG720-2A/OG720-2A delivers exceptional flexibility and scalability for modern data center operations, drawing significant attention across the industry.

With generative AI and LLMs driving increasingly intensive compute demands, enterprises are placing greater emphasis on infrastructure that offers both performance and adaptability. The SG720-2A/OG720-2A emerges as a robust solution, combining high-density GPU integration and flexible liquid cooling options, positioning itself as an ideal platform for next-generation AI training and inference workloads.

TSMC Prepares "CoPoS": Next-Gen 310 × 310 mm Packages

As demand for ever-growing AI compute power continues to rise and manufacturing advanced nodes becomes more difficult, packaging is undergoing its golden era of development. Today's advanced accelerators often rely on TSMC's CoWoS modules, which are built on wafer cuts measuring no more than 120 × 150 mm in size. In response to the need for more space, TSMC has unveiled plans for CoPoS, or "Chips on Panel on Substrate," which could expand substrate dimensions to 310 × 310 mm and beyond. By shifting from round wafers to rectangular panels, CoPoS offers more than five times the usable area. This extra surface makes it possible to integrate additional high-bandwidth memory stacks, multiple I/O chiplets and compute dies in a single package. It also brings panel-level packaging (PLP) to the fore. Unlike wafer-level packaging (WLP), PLP assembles components on large, rectangular panels, delivering higher throughput and lower cost per unit. Systems with PLP will be actually viable for production runs and allow faster iterations over WLP.

TSMC will establish a CoPoS pilot line in 2026 at its Visionchip subsidiary. In 2027, the pilot facility will focus on refining the process, to meet partner requirements by the end of the year. Mass production is projected to begin between the end of 2028 and early 2029 at TSMC's Chiayi AP7 campus. That site, chosen for its modern infrastructure and ample space, is also slated to host production of multi-chip modules and System-on-Wafer technologies. NVIDIA is expected to be the launch partner for CoPoS. The company plans to leverage the larger panel area to accommodate up to 12 HBM4 chips alongside several GPU chiplets, offering significant performance gains for AI workloads. At the same time, AMD and Broadcom will continue using TSMC's CoWoS-L and CoWoS-R variants for their high-end products. Beyond simply increasing size, CoPoS and PLP may work in tandem with other emerging advances, such as glass substrates and silicon photonics. If development proceeds as planned, the first CoPoS-enabled devices could reach the market by late 2029.

DRAM Revenue Drops 5.5% in the First Quarter of 2025; SK hynix Overtakes Samsung for Top Spot

TrendForce's latest findings reveal that global revenue for the DRAM industry reached US$27.01 billion in 1Q25, marking a 5.5% QoQ decline. This downturn was driven by falling contract prices for conventional DRAM and a contraction in HBM shipment volumes. Samsung's redesign of its HBM3e products eased the HBM production squeezes. This prompted downstream players to clear inventories and extended the price declines seen since 4Q24.

Looking ahead to 2Q25, as PC OEMs and smartphone makers complete inventory corrections and ramp up system production ahead of the 90-day U.S. reciprocal tariff grace period, bit procurement momentum is expected to strengthen significantly. This will drive notable increases in supplier shipment volumes. On the pricing side, TrendForce forecasts a rebound across major application contract prices, with both conventional DRAM and combined DRAM (including HBM) contract prices expected to rise.

Intel Reportedly Preparing HBM Alternative for AI Accelerators

Demand for AI accelerators has surged in recent years, putting pressure on suppliers to deliver more high-bandwidth memory to enable faster training and higher token throughput in inference. In response, Intel, SoftBank, and the University of Tokyo have quietly formed a new startup called "Saimemory" to develop an alternative to existing HBM solutions, by also using stacked DRAM. According to sources close to the effort, prototype chips are slated for 2027, with volume production aimed at 2030. The venture will combine Intel's extensive chip design experience with novel memory patents from the University of Tokyo, while SoftBank has pledged approximately ¥3 billion (approximately $21 million) to fund the research. Riken Research Institute and Shinko Electric Industries may also join as investors or technical partners, and the team plans to seek government support to accelerate development.

Traditional HBM relies on through-silicon vias (TSVs) to link multiple DRAM dies and uses a wide-bus interposer to achieve data rates above 1 TB/s. Saimemory's design reorganizes signal routing and refresh management to improve energy efficiency, latency, and performance. As readers may recall, there have been past efforts to introduce a rival stacked DRAM technology, but it has not been successful. For example, the Hybrid Memory Cube (HMC), co-developed by Samsung and Micron Technology around 2011, promised speeds up to fifteen times that of DDR3. Despite initial industry backing through the Hybrid Memory Cube Consortium, Micron discontinued HMC production in 2018 after it failed to gain market adoption. HMC's decline shows the challenge of displacing hard-rooted memory standards like HBM. If the Saimemory succeeds, Intel will likely be the first adopter with its upcoming AI accelerators. Others, such as AMD and NVIDIA, could also be approached by the consortium to get a trial chip. Still, the feasibility of mass deployment will largely depend on availability and yields.

Intel Details EMIB-T Advanced Packaging for HBM4 and UCIe

This week at the Electronic Components Technology Conference (ECTC), Intel introduced EMIB-T, an important upgrade to its embedded multi-die interconnect bridge packaging. First showcased at the Intel Foundry Direct Connect 2025 event, EMIB-T incorporates through-silicon vias (TSVs) and high-power metal-insulator-metal capacitors into the existing EMIB structure. According to Dr. Rahul Manepalli, Intel Fellow and vice president of Substrate Packaging Development, these changes allow a more reliable power supply and stronger communication between separate chiplets. Conventional EMIB designs have struggled with voltage drops because of their cantilevered power delivery paths. In contrast, EMIB-T routes power directly through TSVs from the package substrate to each chiplet connection. The integrated capacitors compensate for fast voltage fluctuations and preserve signal integrity.

This improvement will be critical for next-generation memory, such as HBM4 and HBM4e, where data rates of 32 Gb/s per pin or more are expected over a UCIe interface. Intel has confirmed that the first EMIB-T packages will match the current energy efficiency of around 0.25 picojoules per bit while offering higher interconnect density. The company plans to reduce the bump pitch below today's standard of 45 micrometers. Beginning in 2026, Intel intends to produce EMIB-based packages measuring 120 by 120 millimeters, roughly eight times the size of a single reticle. These large substrates could integrate up to twelve stacks of high-bandwidth memory alongside multiple compute chiplets, all connected by more than twenty EMIB bridges. Looking further ahead, Intel expects to push package dimensions to 120 by 180 millimeters by 2028. Such designs could accommodate more than 24 memory stacks, eight compute chiplets, and 38 or more EMIB bridges. These developments closely mirror similar plans announced by TSMC for its CoWoS technology. In addition to EMIB-T, Intel also presented a redesigned heat spreader that reduces voids in the thermal interface material by approximately 25%, as well as a new thermal-compression bonding process that minimizes warping in large package substrates.

CXMT Reportedly Diversifying Manufacturing Footprint with HBM3 - Could Expand DDR5 Production

The rising profile of ChangXin Memory Technologies (CXMT) has supposedly attracted international scrutiny as-of-late. Despite dealing in commercial memory product lines—currently DDR5, DDR4, LPDDR5 and LPDDR4X—the Chinese manufacturer could be stepping up its game in the near future. According to a fresh DigiTimes Asia news report, the nation's "top DRAM supplier" could be freeing up production capacity—in favor of enterprise-grade third-gen High-Bandwidth Memory (HBM3). Industry moles believe that a major sacrifice will be made; namely CXMT's DDR4 line. Despite an alleged early 2025 ramping up of related activities, the firm's factories could refocus on new endeavors by mid-2026. Additionally, insiders reckon that company leadership is shifting commercial priorities: "by year-end 2025, DDR5 is expected to make up more than 60% of CXMT's output, alongside LPDDR4/5."

The move into more advanced memory technologies is reportedly the result of government instruction. DigiTimes outlined a new strategy: "as CXMT scales up, it's also shifting rapidly to DDR5. The company only began mass-producing DDR4 in late 2024, yet it's already expected to issue an end-of-life (EOL) notice by the third quarter of 2025. The speed of this pivot and retooling has surprised many across the industry. Industry sources say the sudden shift is policy-driven, as Beijing pushes key chipmakers to accelerate alignment with national goals, especially around AI and cloud infrastructure." Murmurs of CXMT's forthcoming exit from DDR4 production have spread across local chip making businesses; causing a sudden doubling of Nanya-branded 8 Gb DDR4 chip prices in China. The manufacturer's early journey into DDR5 territories looked promising on paper—around January 2025—but the latest DigiTimes report disclosed inside track info regarding troubled quality and yield issues. In particular, initial samples have reportedly exhibited unstable performance when crossing a 60°C (140°F) threshold.

SK hynix Presents Groundbreaking AI & Server Memory Solutions at DTW 2025

SK hynix presented its leading memory solutions optimized for AI servers and AI PCs at Dell Technologies World (DTW) 2025 in Las Vegas from May 19-22. Hosted by Dell Technologies, DTW is an annual conference which introduces future technology trends. In line with DTW 2025's theme of "Accelerate from Ideas to Innovation," a wide range of products and technologies aimed at driving AI innovation was showcased at the event.

Based on its close partnership with Dell, SK hynix has participated in the event every year to reinforce its leadership in AI. This year, the company organized its booth into six sections: HBM, CMM (CXL Memory Module)-DDR5, server DRAM, PC DRAM, eSSDs, and cSSDs. Featuring products with strong competitiveness across all areas of DRAM and NAND flash for the AI server, storage and PC markets, the booth garnered strong attention from visitors.

Marvell Custom Cloud Platform Upgraded with NVIDIA NVLink Fusion Tech

Marvell Technology, Inc., a leader in data infrastructure semiconductor solutions, today announced it is teaming with NVIDIA to offer NVLink Fusion technology to customers employing Marvell custom cloud platform silicon. NVLink Fusion is an innovative new offering from NVIDIA for integrating custom XPU silicon with NVIDIA NVLink connectivity, rack-scale hardware architecture, software and other technology, providing customers with greater flexibility and choice in developing next-generation AI infrastructure.

The Marvell custom platform strategy seeks to deliver breakthrough results through unique semiconductor designs and innovative approaches. By combining expertise in system and semiconductor design, advanced process manufacturing, and a comprehensive portfolio of semiconductor platform solutions and IP—including electrical and optical serializer/deserializers (SerDes), die-to-die interconnects for 2D and 3D devices, advanced packaging, silicon photonics, co-packaged copper, custom high-bandwidth memory (HBM), system-on-chip (SoC) fabrics, optical IO, and compute fabric interfaces such as PCIe Gen 7—Marvell is able to create platforms in collaboration with customers that transform infrastructure performance, efficiency and value.

NEO Semiconductor Unveils Breakthrough 1T1C and 3T0C IGZO-Based 3D X-DRAM Technology

NEO Semiconductor, a leading developer of innovative technologies for 3D NAND flash memory and 3D DRAM, announced today the latest advancement in its groundbreaking 3D X-DRAM technology family—the industry-first 1T1C- and 3T0C-based 3D X-DRAM cell, a transformative solution designed to deliver unprecedented density, power efficiency, and scalability for the most demanding data applications.

Built on a 3D NAND-like architecture and with proof-of-concept test chips expected in 2026, the new 1T1C and 3T0C designs combine the performance of DRAM with the manufacturability of NAND, enabling cost-effective, high-yield production with densities up to 512 Gb—a 10x improvement over conventional DRAM.
"With the introduction of the 1T1C and 3T0C 3D X-DRAM, we are redefining what's possible in memory technology," said Andy Hsu, Founder & CEO of NEO Semiconductor. "This innovation pushes past the scaling limitations of today's DRAM and positions NEO as a frontrunner in next-generation memory."

China Rumored to Acquire 12-High HBM3E Bonders Through Korean Companies

China is pushing forward with its HBM (High Bandwidth Memory) progress as part of its plan to be self-sufficient in the semiconductor industry. JCET Group, China's top semiconductor packaging firm, has bought advanced TC (thermal compression) bonders that are usually used for 12-high stacks of HBM3E chips, according to Money Today Korea (MTN). This state-of-the-art equipment comes from Korean companies where export rules are less strict. This lets China jump from its current HBM2 technology to more advanced memory solutions. However, even if China isn't yet going to produce HBM3E chips soon, having this equipment is useful to boost manufacturing yields even for lower-spec HBM products.

China's desire to make HBM chips at home is a reaction to U.S. rules and taxes meant to hold back its chip abilities. These steps haven't slowed progress; instead, they have made China more determined to stand on its own in chip-making. The AI chip design market hit $18.4 billion last year and is set to grow 28% each year until 2032. The plan aims to supply Chinese-made HBM chips to big tech firms like Huawei, Tencent, and DeepSeek. This helps China get around U.S. export limits while moving its chip industry up in value. Choi Jae-hyeok, Professor of Electrical and Information Engineering, Seoul National University says, "In China's case, it is government-led... In the case of DDR, they are making up to DDR4 and DDR5. China has always wanted to move from low-value-added products to high-value-added products. The next direction is HBM..."

Samsung Electronics Announces First Quarter 2025 Results

Samsung Electronics today reported financial results for the first quarter ended March 31, 2025. The Company posted KRW 79.14 trillion in consolidated revenue, an all-time quarterly high, on the back of strong sales of flagship Galaxy S25 smartphones and high-value-added products. Operating profit increased to KRW 6.7 trillion despite headwinds for the DS Division, which experienced a decrease in quarterly revenue.

The Company has allocated its highest-ever annual R&D expenditure for 2024, and in the first quarter of this year, it has also increased its R&D expenditure by 16% compared to the same period last year, amounting to 9 trillion won. Despite the growing macroeconomic uncertainties due to recent global trade tensions and slowing global economic growth, making it difficult to predict future performance, the Company will continue to make various efforts to secure growth. Additionally, assuming that the uncertainties are diminished, it expects its performance to improve in the second half of the year.

TSMC Outlines Roadmap for Wafer-Scale Packaging and Bigger AI Packages

At this year's Technology Symposium, TSMC unveiled an engaging multi-year roadmap for its packaging technologies. TSMC's strategy splits into two main categories: Advanced Packaging and System-on-Wafer. Back in 2016, CoWoS-S debuted with four HBM stacks paired to N16 compute dies on a 1.5× reticle-limited interposer, which was an impressive feat at the time. Fast forward to 2025, and CoWoS-S now routinely supports eight HBM chips alongside N5 and N4 compute tiles within a 3.3× reticle budget. Its successor, CoWoS-R, increases interconnect bandwidth and brings N3-node compatibility without changing that reticle constraint. Looking toward 2027, TSMC will launch CoWoS-L. First up are large N3-node chiplets, followed by N2-node tiles, multiple I/O dies, and up to a dozen HBM3E or HBM4 stacks—all housed within a 5.5× reticle ceiling. It's hard to believe that eight HBM stacks once sounded ambitious—now they're just the starting point for next-gen AI accelerators inspired by AMD's Instinct MI450X and NVIDIA's Vera Rubin.

Integrated Fan-Out, or InFO, adds another dimension with flexible 3D assemblies. The original InFO bridge is already powering AMD's Instinct cards. Later this year, InFO-POP (package-on-package) and InFO-2.5D arrive, promising even denser chip stacking and unlocking new scaling potential on a single package, away from the 2D and 2.5D packaging we were used to, going into the third dimension. On the wafer scale, TSMC's System-on-Wafer lineup—SoW-P and SoW-X—has grown from specialized AI engines into a comprehensive roadmap mirroring logic-node progress. This year marks the first SoIC stacks from N3 to N4, with each tile up to 830 mm² and no hard limit on top-die size. That trajectory points to massive, ultra-dense packages, which is exactly what HPC and AI data centers will demand in the coming years.

OKI Develops 124-Layer PCB Technology for Next-Generation Semiconductor Testing Equipment

OKI Circuit Technology, the OKI Group printed circuit board (PCB) company, has successfully developed 124-layer PCB technology for wafer inspection equipment designed for next-generation high bandwidth memory, such as HBM mounted on AI semiconductors. This is a roughly 15% increase in the number of layers over conventional 108-layer designs. OTC is seeking to establish mass production technology by October 2025 at its Joetsu Plant in Joetsu City, Niigata Prefecture, which has a proven track record and advanced development and production capabilities in the field of high multilayer, high-precision, large-format PCBs for semiconductor inspection equipment.

AI processing requires the transmission of vast data volumes between graphics processing unit (GPU) semiconductors and memory. As semiconductor performance increases, the memory installed is also required to have high-speed, high-frequency, and high-density data transfer capabilities. HBM features a stacked DRAM structure, requiring technology capable of fabricating wafers even more thinly and precisely. This configuration also requires that the PCBs used in inspection equipment meet even higher levels of performance and quality.

TSMC Unveils Next-Generation A14 Process at North America Technology Symposium

TSMC today unveiled its next cutting-edge logic process technology, A14, at the Company's North America Technology Symposium. Representing a significant advancement from TSMC's industry-leading N2 process, A14 is designed to drive AI transformation forward by delivering faster computing and greater power efficiency. It is also expected to enhance smartphones by improving their on-board AI capabilities, making them even smarter. Planned to enter production in 2028, the current A14 development is progressing smoothly with yield performance ahead of schedule.

Compared with the N2 process, which is about to enter volume production later this year, A14 will offer up to 15% speed improvement at the same power, or up to 30% power reduction at the same speed, along with more than 20% increase in logic density. Leveraging the Company's experience in design-technology co-optimization for nanosheet transistor, TSMC is also evolving its TSMC NanoFlex standard cell architecture to NanoFlex Pro, enabling greater performance, power efficiency and design flexibility.

Micron Announces Business Unit Reorganization to Capitalize on AI Growth Across All Market Segments

Micron Technology, Inc. (Nasdaq: MU), a leader in innovative memory and storage solutions, today announced a market segment-based reorganization of its business units to capitalize on the transformative growth driven by AI, from data centers to edge devices.

Micron has maintained multiple generations of industry leadership in DRAM and NAND technology and has the strongest competitive positioning in its history. Micron's industry-leading product portfolio, combined with world-class manufacturing execution enables the development of differentiated solutions for its customers across end markets. As high-performance memory and storage become increasingly vital to drive the growth of AI, this Business Unit reorganization will allow Micron to stay at the forefront of innovation in each market segment through deeper customer engagement to address the dynamic needs of the industry.

US Bans Export of NVIDIA H20 Accelerators to China, a Potential $5.5 Billion Loss for NVIDIA

President Trump's administration has announced that NVIDIA's H20 AI chip will require a special export license for any shipment to China, Hong Kong, or Macau for the indefinite future. The Commerce Department delivered the news to NVIDIA on April 14, 2025, citing worries that the H20 could be redirected into Chinese supercomputers with potential military applications. NVIDIA designed the H20 specifically to comply with earlier US curbs by scaling back performance from its flagship H100 model. The H20 features 96 GB of HBM3 memory running at up to 4.0 TB/s, delivers roughly 296 TeraFLOPS of mixed-precision compute power, and offers a performance density of about 2.9 TeraFLOPS per die. Its single-precision (FP32) throughput is around 74 TeraFLOPS, with FP16 performance reaching approximately 148 TeraFLOPS. In a regulatory filing on April 15, NVIDIA warned that it will record about $5.5 billion in writedowns this quarter related to H20 inventory and purchase commitments now blocked by the license requirement.

Shares of NVIDIA fell roughly 6 percent in after-hours trading on April 15, triggering a wider sell-off in semiconductor stocks from the US to Japan. South Korea's Samsung and SK Hynix each slid about 3 percent, while AMD also dropped on concerns about broader chip-export curbs. Analysts at Bloomberg Intelligence project that, if the restrictions persist, NVIDIA's China-related data center revenue could shrink to low- or mid-single digits as a percentage of total sales, down from roughly 13 percent in fiscal 2024. Chinese AI players such as Huawei stand to gain as customers seek alternative inference accelerators. Commerce Secretary Howard Lutnick has pledged to maintain a tough stance on chip exports to China even as NVIDIA commits up to $500 billion in US AI infrastructure investments over the next four years. Everyone is now watching closely to see whether any H20 export licenses are approved and how long the ban might remain in place.

Huawei CloudMatrix 384 System Outperforms NVIDIA GB200 NVL72

Huawei announced its CloudMatrix 384 system super node, which the company touts as its own domestic alternative to NVIDIA's GB200 NVL72 system, with more overall system performance but worse per-chip performance and higher power consumption. While NVIDIA's GB200 NVL72 uses 36 Grace CPUs paired with 72 "Blackwell" GB200 GPUs, the Huawei CloudMatrix 384 system employs 384 Huawei Ascend 910C accelerators to beat NVIDIA's GB200 NVL72 system. It takes roughly five times more Ascend 910C accelerators to deliver nearly twice the GB200 NVL system performance, which is not good on per-accelerator bias, but excellent on per-system level of deployment. SemiAnalysis argues that Huawei is a generation behind in chip performance but ahead of NVIDIA in scale-up system design and deployment.

When you look at individual chips, NVIDIA's GB200 NVL72 clearly outshines Huawei's Ascend 910C, delivering over three times the BF16 performance (2,500 TeraFLOPS vs. 780 TeraFLOPS), more on-chip memory (192 GB vs. 128 GB), and faster bandwidth (8 TB/s vs. 3.2 TB/s). In other words, NVIDIA has the raw power and efficiency advantage at the chip level. But flip the switch to the system level, and Huawei's CloudMatrix CM384 takes the lead. It cranks out 1.7× the overall PetaFLOPS, packs in 3.6× more total HBM capacity, and supports over five times the number of GPUs and the associated bandwidth of NVIDIA's NVL72 cluster. However, that scalability does come with a trade-off, as Huawei's setup draws nearly four times more total power. A single GB200 NVL72 draws 145 kW of power, while a single Huawei CloudMatrix 384 draws ~560 kW. So, NVIDIA is your go-to if you need peak efficiency in a single GPU. If you're building a massive AI supercluster where total throughput and interconnect speed matter most, Huawei's solution actually makes a lot of sense. Thanks to its all-to-all topology, Huawei has delivered an AI training and inference system worth purchasing. When SMIC, the maker of Huawei's chips, gets to a more advanced manufacturing node, the efficiency of these systems will also increase.

Safe Superintelligence Inc. Uses Google TPUs Instead of Regular GPUs for Next-Generation Models

It seems like Google aims to grab a bit of the market share from NVIDIA and AMD by offering startups large compute deals and allowing them to train their massive AI models on the Google Cloud Platform (GCP). One such case is the OpenAI co-founder Ilya Sutskever's Safe Superintelligence Inc. (SSI) startup. According to a GCP post, SSI is "partnering with Google Cloud to use TPUs to accelerate its research and development efforts toward building a safe, superintelligent AI." Google's latest TPU v7p, codenamed Ironwood, was released yesterday. Carrying 4,614 TeraFLOPS of FP8 precision and 192 GB of HBM memory, these TPUs are interconnected using Google's custom ICI infrastructure and are scaled to configurations in pods of 9,216 chips, where Ironwood delivers 42.5 ExaFLOPS of total computing power.

For AI training, this massive power will allow AI models to quickly go over training, accelerating research iterations and ultimately accelerating model development. For SSI, the end goal is a simple mission: achieving ASI with safety at the front. "We approach safety and capabilities in tandem, as technical problems to be solved through revolutionary engineering and scientific breakthroughs. We plan to advance capabilities as fast as possible while making sure our safety always remains ahead," notes the SSI website, adding that "Our singular focus means no distraction by management overhead or product cycles, and our business model means safety, security, and progress are all insulated from short-term commercial pressures."

Google Unveils Seventh-Generation AI Processor: Ironwood

Google has rolled out its seventh-generation AI chip, Ironwood, which aims to boost AI application performance. This processor focuses on "inference" computing—the quick calculations needed for chatbot answers and other AI outputs. Ironwood stands as one of the few real options to NVIDIA's leading AI processors coming from Google's ten-year multi-billion-dollar push to develop it. These tensor processing units (TPUs) are exclusively available through Google's cloud service or to its internal engineers.

According to Google Vice President Amin Vahdat, Ironwood combines functions from previously separate designs while increasing memory capacity. The chip can operate in groups of up to 9,216 processors and delivers twice the performance-per-energy ratio compared to last year's Trillium chip. When configured in pods of 9,216 chips, Ironwood delivers 42.5 Exaflops of computing power. This is more than 24 times the computational capacity of El Capitan, currently the world's largest supercomputer, which provides only 1.7 Exaflops per pod.

Altera Starts Production Shipments of Agilex 7 FPGA M-Series

Altera Corporation, a leader in FPGA innovations, today announced production shipments of its Agilex 7 FPGA M-Series, the industry's first high-end, high-density FPGA to feature integrated high bandwidth memory and support for DDR5 and LPDDR5 memory technologies. Offering over 3.8 million logic elements, Agilex 7 FPGA M-Series is optimized for applications that demand the highest performance and highest memory bandwidth, including AI, data centers, next-generation firewalls, 5G communications infrastructure and 8K broadcast equipment.

As data traffic continues to increase exponentially due to the growth of AI, cloud computing and video streaming services, the demand for higher memory bandwidth, increased capacity, and improved power efficiency has never been greater. Agilex 7 FPGA M-Series addresses these challenges by offering users high logic densities, a high-performance fabric and a memory interface that accelerates data throughput speeds while reducing memory bottlenecks and latency.

Micron Announces Memory Price Increases for 2025-2026 Amid Supply Constraints

In a letter to customers, Micron has announced upcoming memory price increases extending through 2025 and 2026, citing persistent supply constraints coupled with accelerating demand across its product portfolio. The manufacturer points to significant demand growth in DRAM, NAND flash, and high-bandwidth memory (HBM) segments as key drivers behind the pricing strategy. The memory market is rebounding from a prolonged oversupply cycle that previously depressed revenues industry-wide. Strategic production capacity reductions implemented by major suppliers have contributed to price stabilization and subsequent increases over the past twelve months. This pricing trajectory is expected to continue as data center operators, AI deployments, and consumer electronics manufacturers compete for limited memory allocation.

In communications to channel partners, Micron emphasized AI and HPC requirements as critical factors necessitating the price adjustments. The company has requested detailed forecast submissions from partners to optimize production planning and supply chain stability during the constrained market period. With its pricing announcement, Micron disclosed a $7 billion investment in a Singapore-based HBM assembly facility. The plant will begin operations in 2026 and will focus on HBM3E, HBM4, and HBM4E production—advanced memory technologies essential for next-generation AI accelerators and high-performance computing applications from NVIDIA, AMD, Intel, and other companies. The price increases could have cascading effects across the AI and GPU sector, potentially raising costs for products ranging from consumer gaming systems to enterprise data infrastructure. We are monitoring how these adjustments will impact hardware refresh cycles and technology adoption rates as manufacturers pass incremental costs to end customers.

YES Pioneers Semiconductor Equipment Production in India

Yield Engineering Systems, Inc. (YES), a global leader in materials and interface engineering equipment solutions, proudly announces the shipment of the first commercial VeroTherm Formic Acid Reflow tool to a leading global semiconductor manufacturer from its Sulur, Coimbatore, manufacturing facility. This landmark achievement signifies a pivotal moment for YES and the burgeoning semiconductor ecosystem in India, as it represents the first equipment produced in India for advanced semiconductor applications like High Bandwidth Memory (HBM), which is critical for AI and High-Performance Computing (HPC) applications worldwide.

YES commenced operations in September 2024 at this state-of-the-art manufacturing facility in Sulur, Coimbatore, India, located at 96/3 Vadakku Sambala Thottam, Trichy Road, Kannampalayam, Sulur Taluk. This facility is integral to YES's strategic expansion plan, aimed at serving its global customers' operations in India and the world with greater efficiency.
Return to Keyword Browsing
Nov 17th, 2026 13:49 CST change timezone

New Forum Posts

Popular Reviews

TPU on YouTube

Controversial News Posts