Sandisk sees demand for roughly 60 to 90 gigabytes of NAND flash per CPU in certain AI inference and agent-orchestration applications. The source attributes this to customers rethinking how they use different memory and storage technologies across their AI systems. This is Sandisk's view of demand, not an achieved result; the source does not specify a time frame or list customers.
VeriSilicon has developed GLTiny, a lightweight 3D GPU programming model. According to EE Times Asia, it reduces driver memory requirements and CPU overhead while supporting 3D rendering on resource-constrained wearable devices. The source does not state the cause, provides no specific date, and gives no quantitative metrics or commercialization timeline.
3D GPU programming model (GLTiny)VeriSiliconEE Times Asia
According to the report published Friday 30 Oct 2026, Intel's share price plunged on Thursday after the company announced its Q3 results: shares started the day at $92.96 apiece, sank to $73 immediately after the results were announced, and reached $70.87 in after-hours trading at the time of writing, a 23.7 percent slump. The Q3 results included a 21 percent year-on-year revenue jump to $16.5 billion, at the high end of Intel's revenue guidance of $15.8-16.8 billion, helped by 78 percent growth for datacenter and AI products, and gross margins of 43 percent, which beat Intel's predictions by two points; however, revenue for client computing and physical AI products reached $8.1 billion, falling five percent year-over-year and declining nine percent from last quarter. In prepared remarks, CEO Lip Bu Tan said that in recent weeks customers began signaling a PC slowdown based on memory prices and availability, that the pace of the PC correction has accelerated over the last two months as OEM partners aggressively manage inventory to match softer sell-through, and he predicted the market for PC chips will shrink by 12 percent next year and remain tough until memory availability and prices begin to normalize. CFO Dave Zinsner declined on the earnings call to offer guidance on whether Intel can meet server CPU demand early in Intel's next financial year, saying Intel will likely not fully meet what customers require and that the situation will be more challenging in the first half of the year than in the second half. Tan also predicted, as a forecast rather than an achieved result, that server CPU TAM, largely driven by agentic workloads, could increase as much as 5X over five years, from $40B to more than $200B. The report also notes that SK Hynix this week predicted memory shortages will persist until the year 2030 (a prediction by SK Hynix, not an Intel statement).
PC processorsdatacenter and AI productsmemoryIntelThe RegisterPrice Movement
Eternal Materials posted earnings before taxes (EBT) of NT$388 million (approx. US$11.92 million) in September 2026, surging 158.28% year-over-year and up 44.47% from August, as revenue from semiconductor advanced packaging materials and precision equipment both reached record highs for a single month. The source states no cause beyond the record single-month revenue in these two businesses; the core fact is the reported September results.
semiconductor advanced packaging materialsprecision equipmentEternal MaterialsDigiTimes Asia
Intel CEO Lip-Bu Tan confirmed on the company's Q3 2026 earnings call that the next-gen Nova Lake desktop processors will be the first client product to use Intel's enhanced 18A-P node, that Nova Lake is tracking ahead of schedule with design wins expecting to be twice those of Panther Lake at launch, and that more details will be shared at CES in January (confirmation/planning stage, not an implemented capacity result). CFO David Zinsner said Intel intends to take 18A wafers — which were generally going to build Panther Lake products — offline to put Intel 3 into Oregon, using some of the freed-up capacity to produce more data center wafers (Granite Rapids); this reallocation is a stated plan/intent, not yet a completed action. Background figures: 18A-P offers a 9% speed improvement at iso-power or an 18% power reduction at iso-speed versus 18A; 18A products now make up a third of client CPU revenue, up from less than 5% in Q1; Tan said demand for Wildcat Lake is outstripping supply; OEM partners are aggressively managing inventory to match softer sell-through, and the company expects its client business to decline by more than 20% in the next quarter; Zinsner expects the 2026 PC unit TAM to decline approximately 12% with second-half units down more than 20%, plans a roughly 10% PC unit TAM decline for 2027, and does not expect a sustained recovery until memory availability improves and pricing begins to normalize.
Nova LakePanther LakeGranite RapidsIntelTom's HardwareCapacity Change
Reported on 2026-10-30, Intel's 3Q26 results were mixed: Data Center and AI (DCAI) revenue surged 78% to $7.3 billion and operating income more than tripled to $3.04 billion; however, according to Reuters, a fatal explosion in October at Intel's Kiryat Gat fab in Israel, which primarily produces data center processors using 10nm technology, is expected to cost roughly $600 million in revenue over two quarters, and sluggish PC demand could persist through 1H27. Reflecting these headwinds, Intel forecast fourth-quarter revenue of $15 billion to $16 billion, missing the $17.07 billion LSEG consensus, with adjusted earnings projected at 25 cents per share versus analysts' 43-cent expectation. CFO David Zinsner said the incident would reduce fourth-quarter revenue by roughly $300 million and shave 200 basis points off gross margin, with a similar revenue hit expected in 1Q27; after a two-week shutdown, Intel aims to begin restoring production at the Israeli facility by the end of this week. In the PC business, 3Q26 sales fell 5% to $8.10 billion, missing analysts' $8.98 billion estimate, and Zinsner expects fourth-quarter PC business revenue to plunge by a percentage in the mid-20s. CEO Lip-Bu Tan said he was disappointed the company could not meet strong customer demand, citing manufacturing supply shortages, an unprecedented memory crunch and resulting PC demand weakness as near-term challenges; he also noted that despite rapidly ramping Intel 3 wafer output, an industry-wide shortage of advanced organic substrates for EMIB packaging continues to limit production scaling. Zinsner also warned the chipmaker is unlikely to generate positive cash flow in 2026; Intel had raised its capital expenditure guidance from $18 billion to $20 billion in July, with a further substantial increase expected in 2027 to support its manufacturing buildout. On the positive side, Zinsner expects industry-wide server CPU unit shipments to grow more than 30% in both 2026 and 2027; Intel Foundry revenue surged 58% to $6.7 billion in 3Q26 including internal sales, with its operating loss narrowing to $1.85 billion from $2.32 billion a year earlier. Tan said a couple of customers will be ready to announce adoption of Intel's EMIB/EMIB-T advanced packaging over the next two quarters, and that in October Intel delivered its industry-standard 0.9 process design kit (PDK) for the 14A process to select customers as planned. TrendForce expects Google to adopt EMIB-T in 2027, while AWS is also testing EMIB technology; industry sources earlier suggested Google's TPU v9 lineup includes three variants: Broadcom-designed WhaleFish (expected to use TSMC's CoWoS-L with mass production slated for 2H27), MediaTek-designed HumuFish (set for 1H28 mass production relying exclusively on Intel's EMIB-T) and TriggerFish (also planned to use EMIB-T, targeted for 2H28 or later).
EMIBEMIB-Tadvanced organic substratesIntel, Google and 5 totalTrendForceDisaster/Disruption
TrendForce reported on October 29, 2026 that the high-capacitance MLCC shortage persists; with NVIDIA Vera Rubin ramping up and AMD sourcing MI450 components, ODMs expect high-end MLCC shortages to peak in Q1 and Q2 2027. TrendForce also noted that AI inference drives CPU and DDR5 demand, with high-capacitance MLCC shortages expected to persist.
high-capacitance MLCChigh-end MLCCCPUNVIDIA, AMDTrendForce Official (Market Data)Inventory Shift
A research commentary article discusses the 2023 paper 'Machine Learning-Based Microarchitecture-Level Power Modeling of CPUs', published in the IEEE Transactions on Computers with authors from UT Austin plus Hyperstone and U Stuttgart in Germany; the paper has 18 citations and extends a 2021 learning-based power modeling paper (which covered fixed function accelerators) to programmable in-order and out-of-order CPUs, with the learning method based on DNNs. The authors train decision-tree models to predict power for a single clock cycle on low level CPU building blocks such as fetch, decode, and load-store, trained using golden power data from a post-P&R gate-level netlist and simulation traces, then assemble the models hierarchically to construct a cycle-accurate power profile from a cycle-accurate micro-architectural simulation trace of the complete design. Results on two open source designs (a simple in-order CPU and the complex out-of-order CPU BOOM), implemented on an open source 45nm process and library, show that with only 20k cycles of training data using 8 very simple test programs (quick sort, Towers of Hanoi, Dhrystone, ...), power estimation for BOOM on the CoreMark benchmark has an average error in any clock cycle of less than 4%; at the resolution of individual building blocks, 80% of the blocks are within a 15% error in any clock cycle; and the runtime overhead of processing the decision trees is negligible. The commentary also cites that gate-level reference power estimation of the BOOM processor (approx. 460K standard cells) took a total of 40-65 hours running at 2-3 cycles/s versus 7-9 hours using the learned models and Verilator-based simulation, a speedup of roughly 5-9x; that the 3.6% MAE on CoreMark is normalized to the average power of the entire CPU and does not mean power is predicted within 3.6% in every cycle, with accuracy varying strongly across blocks (ALU 15% MAE, issue unit 18%, memory issue unit 23%); and that the decision trees are calibrated against a particular synthesized and placed-and-routed design, with the authors providing no evidence that characterized blocks preserve accuracy across different configurations, so applicability to design-space exploration remains unclear.
Infineon Technologies AG is expanding its collaboration with NVIDIA on 800 V direct-current (VDC) power architectures for AI data centers, with the latest development focusing on high-voltage intermediate bus converters (IBCs), a stage in the power delivery chain that steps down the main distribution voltage to levels suitable for downstream conversion and processor power delivery. The collaboration builds on work announced in May 2025 and further detailed in October 2025, which addressed serviceable, hot-swappable server racks operating with 800 VDC power distribution. At the IBC stage, switching frequencies up to 1 MHz enable smaller magnetic components and passive devices, reducing converter size and board height; Infineon combines its GaN portfolio with PSoC™ Control C3 controllers and EiceDRIVER™ gate-driver products to provide integrated IBC solutions, supporting conversion ratios including 800 V to 50 V, 800 V to 12 V, and 800 V to 6 V. The collaboration with NVIDIA also extends beyond intermediate bus conversion to power-path protection, including hot-swap controller functionality intended to support safe servicing and the deployment of future server boards in 800 VDC architectures.
GaN power deviceshigh-voltage intermediate bus converters (IBCs)PSoC™ Control C3 controllersInfineon Technologies AG, NVIDIAPower Electronics NewsPartnership/M&A
This is an opinion/analysis article arguing that AI processing, specifically inference, remains mostly cloud-centric but is shifting toward edge devices, and that storage and memory requirements will keep rising in both cloud and edge scenarios. According to the article, data center build-outs are redirecting supplies of SSDs (and HDDs) and DRAM, while the required capacities in PCs, smartphones and the like for today's edge AI processing are increasing in lockstep: the Gemini AI models embedded in latest-generation Chrome browser installs take up at least 4 GBytes of storage space; Apple Intelligence included with iOS 27 is now required versus optional with the version 26 predecessor and uses up to 14 GBytes; the required-by-default AI model suite included with MacOS 27 is 12-30 GBytes in size. PrismML compressed Alibaba's Qwen3.8 27B open-source model by 9-10x while still matching 98% of Qwen's aggregate benchmark scores, but the resultant model was still 5.9 GBytes; Perplexity's Portable Computer for local on-computer model storage and processing uses 20 GBytes. The article also states that the training run for OpenAI's latest-generation "Astra" GPT-6 fully consumed more than 100,000 GPUs, and argues that hybrid approaches (local base-level inference with cloud tapped as needed) will become increasingly common, with the decode portion of inference being memory-intensive from both capacity and bandwidth standpoints, rationalizing advanced HBM approaches. All points are the author's analysis; the source reports no price changes.
According to DIGITIMES Intelligence, global server CPU shipments are projected to surpass the 50 million-unit milestone for the first time in 2027, reaching 52.6 million units, a surge of 46%, as demand accelerates alongside the industry-wide shift from AI training to inference. The projection also states that Intel is expected to retain its shipment crown, while rival AMD is closing in fast (the title notes AMD eyeing the revenue crown). This is a forecast, not an achieved shipment result.
On Oct. 29, 2026, Samsung Electronics reported results for the third quarter ended Sept. 30, 2026: consolidated revenue of KRW 195.3 trillion, an all-time quarterly high, up 14% QoQ, and operating profit of KRW 107.4 trillion, making it the world's first tech company to exceed KRW 100 trillion in quarterly operating profit. The DS Division posted revenue of KRW 147.1 trillion and operating profit of KRW 106.5 trillion, with QoQ sales up 15%; the Memory Business set an all-time high for quarterly revenue and operating profit, managing its product mix toward AI products amid capacity constraints, with strong server demand leading to higher market prices across all applications, and significantly scaled up HBM4 sales. The DX Division posted QoQ revenue growth of 8% and an operating loss of KRW 0.8 trillion. MX and Networks posted combined revenue of KRW 37.7 trillion with an operating loss of KRW 0.6 trillion; SDC posted revenue of KRW 8.0 trillion and operating profit of KRW 1.1 trillion; VD/DA posted combined revenue of KRW 13.9 trillion with an operating loss of KRW 0.2 trillion. Q3 capital expenditure was KRW 29 trillion, up KRW 12.2 trillion QoQ and KRW 19.8 trillion YoY; the January-September cumulative total was KRW 56.9 trillion, up KRW 24.7 trillion YoY; full-year 2026 capex is expected at approximately KRW 84.4 trillion (KRW 76.8 trillion for DS, KRW 3.9 trillion for SDC) — an expectation, not an achieved result. The company projects that in Q4 2026 memory demand growth will revolve around new CPU and GPU platform-related products and surging QLC SSD demand, and that in 2027 AI-driven demand will far exceed industry supply compared with 2026; these are forward-looking outlooks.
This is a background technology column published on October 29, 2026 (authored by Professor Younghyun Kim of Hanyang University ERICA), explaining — not reporting — industry developments; it contains no price, capacity or policy event. The article states that AI data center bottlenecks have shifted from GPU compute speed to data movement: copper interconnects suffer signal loss, heat and repeater power overhead at high speeds and longer distances, so networks are shifting to silicon-photonics-based optical interconnects; co-packaged optics (CPO) places optical engines next to the switch ASIC to shorten electrical paths, increase bandwidth density and save power. It cites Meta's evaluation of a CPO switch system presented at ECOC 2025, in which power consumption in the optical interconnect portion was approximately 65% lower than a comparable pluggable architecture, with connection stability confirmed under the specific test conditions; companies such as NVIDIA and Broadcom are accelerating commercialization by introducing network products incorporating CPO technology. The article also describes LPO (removable modules with the DSP eliminated) and NPO (optical engines close to but outside the switch package) as alternative approaches, explains that HBM and CPO reduce GPU data-wait time at the memory and network levels respectively, and notes remaining CPO challenges including laser light supply, fiber connection, heat from ASICs, and yield improvement.
co-packaged optics (CPO)silicon photonicsHBMNVIDIA, Broadcom and 3 totalSK Hynix Newsroom
In a technical explainer, Samsung states that zHBM, currently under development, is a next-generation AI memory architecture designed around AI inference, reasoning, long-context processing and agentic AI workloads that place greater demands on memory through more generated tokens and larger KV caches. Moving beyond conventional HBM architectures in which memory sits beside the AI chip, zHBM directly stacks memory on the processor. The article says zHBM uses direct 3D integration and a Distributed I/O architecture to create shorter, denser connections intended to deliver high bandwidth with greater energy efficiency; core technologies include Wafer-on-Wafer (WoW) stacking combined with Hybrid Copper Bonding (HCB) for direct copper-to-copper connections at fine pitches, and reduced size and pitch of Through Silicon Vias (TSVs) for higher interconnect density, with shorter interconnects lowering data-transfer energy. The article notes that direct stacking creates new thermal challenges, as heat generated by the processor may need to pass through the memory stack, and that Samsung is exploring stack-height optimization; it emphasizes co-design and co-optimization of memory with the AI chip from early development stages and aims to work with customers and ecosystem partners on zHBM solutions tailored to their AI chips' performance, power, thermal and architectural requirements. The article is informational only, with no specific timeline, capacity or pricing figures and no statement of market supply-demand conditions.
According to Kinsus Interconnect Technology R&D vice president Ting-hao Lin at the 2026 TPCA Show, Kinsus is exploring aluminum nitride (AlN) ceramics as a core material for substrates used in large-scale advanced packaging, seeking an alternative to glass. The exploration of a glass alternative is set against ongoing brittleness and reliability challenges of glass. The effort remains at the exploration stage, i.e., planned/research status, not an achieved mass-production or commercialized result.
aluminum nitride (AlN) ceramicsglass core materialsubstratesKinsus Interconnect TechnologyDigiTimes Asia
Semiconductor test socket maker ISC has doubled its share of test socket supply at its largest North American smartphone customer to 40% from 20%. The source-stated cause is that the customer diversified orders after a major incumbent supplier failed to meet delivery deadlines. Speaking on its Oct. 27 earnings conference call, ISC said shipments to the customer's mass-production lines began increasing significantly in the third quarter and that it has secured orders for the first half of next year. All sockets ISC has supplied to the customer so far have been pogo-pin sockets, primarily used in mass-production testing of mobile application processors (APs); ISC's core rubber socket product had not entered this market. The customer has also commissioned ISC to develop a product that would replace existing pogo-pin sockets with rubber sockets, and once development is completed, pogo-pin socket orders are expected to shift to ISC's core product. ISC generated approximately 20 billion won in revenue from the customer last year, a figure expected to rise to as much as 35 billion won this year. ISC reported third-quarter consolidated revenue of 71.8 billion won and operating profit of 21.8 billion won, up 11.2% and 25.4% year-over-year, with revenue down 1.6% and operating profit up 2.4% quarter-over-quarter and an operating margin of 30.4%. Third-quarter mobile revenue totaled 12.4 billion won, down 8% year-over-year and 15% quarter-over-quarter, reflecting a socket back-end production bottleneck that pushed approximately 11 billion won in orders into the fourth quarter. ISC expects record-high revenue and operating profit in the fourth quarter and forecasts full-year revenue of approximately 300 billion won and operating profit of 100 billion won. On capacity, the first Vietnam plant completed an expansion in the second quarter and began operations in October; ISC aims to start operations at its second Vietnam plant in the second quarter of 2027 and its Songdo plant in the first quarter of 2028, with annual capacity set to rise from the current 2 million units to 3 million in 2027, 4 million in 2028 and 5 million in 2029.
test socketpogo-pin socketrubber socketISCThe Elec
On Oct. 27 local time at Oracle AI World, Intel said it plans to increase shipments of its current-generation Xeon server CPUs to Oracle Cloud Infrastructure (OCI) and to supply its next-generation Xeon processors, codenamed Diamond Rapids. OCI's 14th-generation server platform (X14) will feature Intel's Diamond Rapids processors to improve AI acceleration and general computing performance. OCI X14 will incorporate enhanced Intel Advanced Matrix Extensions (AMX), which support reduced-precision floating-point formats including FP8, FP16 and BF16, allowing AI inference applications to run more efficiently and at higher speeds. The platform will also feature Intel Advanced Performance Extensions (APX), which improve execution speed by reducing instruction overhead. These technologies will enhance the efficiency of embedding generation, reranking of search results, and small language model (SLM) inference workloads. This is a planned expansion of the supply partnership; the source does not state the cause.
Qualcomm said on Oct. 27 local time that it will supply its Snapdragon X platform for the Blue Chip Project. The Blue Chip Project aims to design and assemble computer hardware for various public-sector applications in the US, with the goal of rebuilding domestic computer manufacturing capabilities. Qualcomm will supply its Snapdragon X series, a system-on-chip (SoC) platform for PCs that serves as the core processor for the Blue Chip product lineup; the platform integrates a CPU for general computing tasks, a GPU for graphics rendering and an NPU for AI workloads. Anduril will develop computers, including laptops, based on Snapdragon X chipsets, intended for government, defense, emergency response and other public-sector applications. Anduril has already completed a prototype of the first Blue Chip laptop; components sourced from US and allied-country suppliers account for more than 90% of total component procurement costs. Official orders will begin next year. PCB assembly will take place at Anduril's Arsenal-1 facility in Ohio. The project will proceed in stages: the first phase focuses on sourcing US-made components and expanding domestic manufacturing, the second aims to minimize dependence on foreign suppliers, and the ultimate goal is a domestic supply chain handling everything from raw materials to final product assembly in the US. The source does not state a cause beyond this context.
Snapdragon X platformCPUGPUQualcomm, AndurilThe ElecPartnership/M&A
SK ecoplant Materials (SK Materials) said on Oct. 28 that it signed a memorandum of understanding (MOU) with Japan's Furukawa Electric Co. to cooperate on semiconductor materials for artificial intelligence (AI) data centers. Furukawa Electric will develop new products for semiconductor and data center applications, while SK Materials will evaluate the new materials and expand their adoption in the South Korean market. The companies plan to establish a working group for regular technical exchanges and business discussions, and will jointly assess potential applications for various materials based on customer demand. The collaboration aims to address increasingly stringent performance and reliability requirements for materials used in packaging, thermal management, insulation and communications infrastructure. The source does not state a cause beyond this cooperation context.
semiconductor materials for AI data centersSK ecoplant Materials (SK Materials), Furukawa Electric Co.The ElecPartnership/M&A
Related manufacturers
SK ecoplant Materials (SK Materials), Furukawa Electric Co.
AP Systems said on Oct. 28 that it will participate in a third-party allotment of new shares by SD Optics, acquiring a stake in the provider of ultra-precision optical inspection solutions through convertible preferred shares (CPS). This is a planned equity investment. Through the investment, AP Systems plans to integrate SD Optics' optical technologies into its equipment, explore new markets and establish a strategic partnership, jointly targeting the next-generation semiconductor and display inspection and metrology markets. The companies will target advanced packaging applications including high-bandwidth memory (HBM), fan-out panel-level packaging (FOPLP) and through-glass vias (TGV) in semiconductor glass substrates, and jointly develop metrology and inspection equipment. Founded in 2010, SD Optics became the first company in the world to commercialize MEMS-based Mirror Array Lens System (MALS) technology; MALS controls an array of microscopic mirrors at a frequency of 12,000 Hz, enabling real-time focus adjustments without mechanical vibration, and is optimized for 3D surface profile measurements of samples with significant height variations and ultra-high-speed vision inspection. The technology has been adopted in ZEISS microscope systems, and SD Optics has supplied inspection modules and equipment to TGV, advanced packaging and display companies. Integrating MALS-based 3D line-scan modules into AP Systems equipment would enable precise, real-time measurement of product surface profiles, height differences and residues.
ultra-precision optical inspection solutionsmetrology and inspection equipmentMALS-based 3D line-scan modulesAP Systems, SD OpticsThe ElecPartnership/M&A