Iluvatar CoreX has reported benchmark results showing that its Tiangai 300 AI accelerator outperforms Nvidia's H200 on selected metrics, drawing renewed attention to China's progress in developing domestic alternatives to high-end GPUs. The commentary notes that the broader competition extends beyond raw processing power, as Chinese chipmakers seek to narrow the gap with leading international AI computing platforms through chip design, software optimisation and system integration.
AI acceleratorhigh-end GPUsIluvatar CoreX, Huawei and 3 totalDigiTimes Asia
According to reports, Elon Musk is reconsidering his approach to collaborating with TSMC on the Terafab project, signaling a potential shift from his original ambition to independently build and operate semiconductor fabs. This remains a reported, potential shift under consideration with no confirmed implementation, and the source does not state the cause.
According to Seoul Economic Daily, OpenAI CEO Sam Altman visited South Korea for the first time in about a year and met Samsung Electronics executives to discuss AI cooperation; Richard Ho, OpenAI's vice president overseeing in-house AI chip development, accompanied him, fueling speculation about potential collaboration on next-generation AI accelerators. The report says OpenAI's in-house AI accelerator, Jalapeño, is expected to use Samsung's next-generation HBM4 memory, and that Samsung signed a memorandum of understanding with Broadcom in July covering up to US$200 billion in memory and foundry cooperation through 2030; with Broadcom developing custom AI chips for OpenAI, Samsung could also gain foundry and packaging opportunities (reported/speculative, not confirmed). Per Business Korea, Altman also met SK Group Chairman Chey Tae-won, with SK hynix and SK Telecom executives reportedly attending; details were not disclosed, but the discussions are believed to have covered existing cooperation on HBM supply and AI data center development in South Korea along with potential future collaboration. Per The Asia Business Daily, OpenAI's Stargate team visited South Korea ahead of Altman to discuss semiconductor supply volumes, contract terms and AI infrastructure development with local companies, including HBM supply talks with Samsung. As background, per TrendForce, HBM's blended average selling price is projected to rise 121% YoY in 2027, driven by supply constraints and a growing share of higher-priced HBM4. In Taiwan, per Commercial Times, Foxconn Chairman Young Liu confirmed meeting with Altman, and Reuters reports Foxconn is working on a new collaboration with OpenAI; sources cited by Commercial Times say Altman met TSMC Chairman C.C. Wei and engaged AI server suppliers including Quanta and Wistron. In June 2026, OpenAI and Broadcom unveiled their first custom AI inference chip, Jalapeño, which took nine months from initial design to tape-out; per OpenAI, the chip has undergone performance testing, with initial deployment planned by the end of 2026 and expansion to gigawatt-scale capacity in the coming years. Per Commercial Times citing supply chain sources, OpenAI's first-generation ASIC uses TSMC's 3nm process, while its second-generation chip is reportedly planned to adopt A16 with backside power delivery to improve performance and energy efficiency.
HBMHBM4AI acceleratorOpenAI, Samsung Electronics and 9 totalTrendForcePartnership/M&A
Related manufacturers
OpenAI, Samsung Electronics, SK hynix, SK Telecom, Broadcom, Foxconn, TSMC, Quanta, Wistron
India's semiconductor push is accelerating: states are launching new subsidies, chipmakers are weighing fabs, global suppliers are considering manufacturing in the country, and engineering partnerships are deepening India's role in chip supply chains. The source gives no specific companies, products, amounts, or timelines; the fab and manufacturing moves remain at the weighing/considering stage, i.e., planned rather than confirmed.
On Saturday, October 31, Samsung hosted OpenAI CEO Sam Altman for an AI seminar with around 50 Samsung presidents at its training center in Yongin, with leaders from Samsung's finance and biotech arms joining the electronics side. Samsung presented it as part of a leadership seminar series for its company-wide AI push, and the talks covered where AI is headed, the markets it could open up, and how it changes the way large companies make decisions. According to Samsung, Altman said adopting AI alone doesn't give anyone an edge, and the payoff comes from using it to move faster and build better products. The source explicitly states that nothing came out of it in terms of a new supply contract, at least publicly, and Samsung's own write-up stuck to AI strategy. Background: Samsung signed a letter of intent with OpenAI back in October 2025 to act as a memory partner for Stargate, with OpenAI's demand projected at up to 900,000 DRAM wafers per month; separately, on October 31, SK Group chairman Chey Tae-won also reportedly met Altman in Seoul, with SK hynix and SK Telecom executives joining, per industry sources; last month, AMD's Lisa Su spent hours with Samsung's chip chief chasing HBM supply; and Micron's CEO expects the memory crunch to get worse in 2027 and 2028. The source does not state any direct impact of this meeting on memory supply and demand.
China's next-generation short-range wireless connectivity technology, NearLink, is actively being pushed into markets beyond the Huawei ecosystem. According to International SparkLink Alliance secretary-general Guosong Zeng, securing support from major chip vendors such as MediaTek, STMicroelectronics (ST), and NXP will be critical as the alliance works to encourage wider adoption among non-Huawei smartphone brands. The source does not state specific progress, a timeline, or any commitments already made by the chip vendors.
NearLinkHuawei, MediaTek and 4 totalDigiTimes Asia
In an interview, CircuitHub co-founder and CEO Andrew Seddon said the company is increasing capacity at its Massachusetts site fivefold and plans to open a UK factory in 2027. CircuitHub operates automated, software-driven factories for on-demand electronics manufacturing (The Grid, located in Massachusetts), and has shipped more than 2.2 million boards, placed over 150 million components and served more than 26,000 engineers; the company says it has now reached price parity with comparable manufacturers in China, with an average turnaround of four days versus nine for a typical Chinese quick-turn shop. It states around 95% of projects involve fewer than 10,000 units. The capacity expansion is a planned/ongoing company expansion, not a completed increase in output.
Reported 2026-11-02 (earnings call): Intel CFO Dave Zinsner said costs for Intel 3 and Intel 18A had fallen roughly 15% and 40%, respectively, since the first quarter, with yields also trending higher. 18A-based processors now account for nearly one-third of client CPU revenue, up from less than 5% in the first quarter. Demand is particularly strong for Wildcat Lake, Intel's cost-optimized 18A processor, with silicon unit volumes rising 70% sequentially and available supply already sold out. Intel is shifting some of its Oregon capacity toward Intel 3 to prioritize data center demand; the affected 18A wafers were largely earmarked for Panther Lake, and the move is expected to free up server CPU wafer capacity, though substrate supply remains another bottleneck. CEO Lip-Bu Tan said Clearwater Forest, Intel's first 18A server product, is already shipping for revenue, with significant demand expected in 2027. Intel reported double-digit yield improvements for 18A during the third quarter. Roadmap: Diamond Rapids, slated for 2027, will use Intel 18A-P, while its successor Coral Rapids is expected to adopt Intel 14A in 2028; Nova Lake desktop processors will mark the debut of 18A-P, progressing ahead of schedule with launch design wins projected to double those of Panther Lake, and more details expected at CES in January 2027. Market speculation (rumor) suggests Nova Lake may use a split-node approach pairing Intel 18A-P for the SoC tile with TSMC N2P for the compute tile, a configuration the report leaves open to question. Per Tom's Hardware, Intel 18A-P delivers a 9% performance boost at the same power or an 18% reduction in power consumption at equivalent speeds. Intel had warned that sluggish PC demand could persist into the first half of 2027.
This is an interview with Peter Jenkins, CEO of UK semiconductor metrology company Infinitesima, and is background information with no price movement or supply event. Key facts: Infinitesima is based near Oxford in the UK, founded in 2001 as a spin-out from the University of Bristol, and for the best part of twenty-five years has been developing technology enabling direct Angstrom-scale 3D measurement of chip features; its high-speed hybrid optical/AFM Rapid Probe Microscope (RPM) technology is being deployed in production as a fully automated in-line system, the Metron 3D; the company has a growing global team of over 140 and works with SK hynix, imec and ASML, with SK hynix having put the Metron 3D through a rigorous evaluation demonstrating its ability to meet high-volume manufacturing requirements. Before joining Infinitesima, Jenkins spent 27 years at ASML (stated as twenty-eight years elsewhere in the text), and previously began his career in semiconductor process engineering at Plessey Research and LSI Logic. Its strongest application area today is hybrid bonding, central to advanced packaging, high bandwidth memory and heterogeneous integration (copper pad recess and dishing measurement), and it runs a three-year programme with imec focused on Hybrid Wafer Bonding, High NA EUV and CFET transistor structures; it is also developing Multiprobe, which scans with many probes in parallel. The source does not state the cause of any event; the article contains no reported price, capacity or supply-demand changes.
angstrom-scale process control technologyRapid Probe Microscope (RPM) technologyMetron 3DInfinitesimaSemiWiki
At the Taiwan Semiconductor Industry Association's (TSIA) 30th anniversary gathering in Hsinchu, industry pioneer Shih Chin-Tay offered a different way to understand Taiwan's semiconductor advantage: not as the product of one breakthrough, one company or even one technology cycle, but as the cumulative result of six decades of learning. The article notes that Taiwan's semiconductor strength is often described today through leading-edge process nodes, advanced packaging capacity, and its pivotal role in the global AI supply chain. This is background industry analysis; no price movement, capacity change, or specific supply-demand event is reported, and the source does not state the cause.
Taiwan semiconductor ecosystemTaiwan Semiconductor Industry Association (TSIA)DigiTimes Asia
According to a report published 31 October 2026, Intel is reallocating some of its capacity from client to data center CPUs in the face of what CEO Lip-Bu Tan called 'one of the most significant supply shortages in history,' driven by unprecedented demand for Super Intelligence. The company recently began expanding capacity at Fab 24 near Leixlip, Ireland to increase output of Intel 3-based Xeon 6 'Granite Rapids' processors, but this capacity will only come online by the end of 2027 unless accelerated. Intel also plans to convert some manufacturing resources at its Oregon facilities from Intel 18A to Intel 3 — resources originally intended for Panther Lake client processors — with the additional Intel 3 capacity expected to enter production in the second half of 2027. Intel is accelerating equipment installation into Fab 62 in Arizona, set to become its second high-volume manufacturing facility for 18A/18A-P, and is expanding advanced packaging capacity in New Mexico and Malaysia. Intel 3 manufacturing costs have declined by approximately 15% since the first quarter of 2026. Advanced organic substrates remain a bottleneck, with approximately 40% additional substrate capacity expected in the second half of 2027. Intel's DCAI business generated $7.3 billion in revenue in Q3 2026, up 17% sequentially and nearly 80% year-over-year; the company expects CapEx of approximately $20 billion in 2026, increasing significantly in 2027, with firm purchase orders to equipment suppliers up nearly 40% sequentially in Q3. Intel expects supply constraints to remain particularly severe in the first half of 2027 and demand for data center CPUs to remain strong in 2027–2028.
server CPUsXeon CPUsXeon 6 'Granite Rapids' processorsIntelTom's HardwareCapacity Change
This is a background article stating that Taiwan's semiconductor industry has earned its reputation for chip dominance through decades of sustained investment in solving some of the world's hardest technological challenges, and that this persistence has built a manufacturing ecosystem whose industrial density, specialization, and resilience remain difficult for competitors to replicate. The title references TSIA 30 and the manpower problem behind Taiwan's semiconductor resilience. The source provides no specific price, capacity, or policy event, and does not state a cause or dates.
AUO is accelerating its strategic push into optical communications and advanced semiconductor packaging, establishing optical engines and glass core substrates (GCS) as its two primary technological pillars. According to AUO chairman and CEO Paul Peng, these technologies are still in their early phases, specific operational targets have yet to be finalized, but they are projected to make significant contributions to overall revenue within two to three years. The source does not state the cause.
optical enginesglass core substrates (GCS)AUODigiTimes Asia
According to Tom's Hardware on 30 October 2026, Intel's shares plunged 30% on Friday, wiping out nearly $150 billion of market capitalization in a single day — the largest single-day loss in the company's history — following a disappointing financial outlook largely driven by an anticipated drop of more than 20% in sales at Intel's client computing and physical AI group (CCPG). According to intraday estimates by Benzinga, approximately $1 trillion in semiconductor market capitalization was wiped out on Friday. The PHLX Semiconductor Sector Index (SOX), which tracks 30 major semiconductor companies, plunged 9.14%, or 1,052.82 points, to 10,465.38 points as of 2:31 PM ET. Stock moves: Arm -12.68%, AMD -12.48%, Qualcomm -12.14%, Applied Materials -8.73%, ASML -8.13%, Lam Research -8%, Micron -7.28%, Broadcom -7.1%; Nvidia lost 5.66%, about $270 billion, more in absolute dollar terms than Intel. The downturn extended to server makers and cloud companies: Dell -12.67%, HPE -8.10%, Supermicro -6.96%, Oracle -6.88%; Apple fell only 1.66%. Fundamentals: CCPG generated $8.1 billion in Q3 2026 revenue, down 5% year-over-year and 9% sequentially; weaker demand and increased competition cut client processor shipments 14% year-over-year, partially offset by a 9% increase in average selling prices; CCPG accounted for nearly half of Intel's revenue of around $16.5 billion. As strong server processor demand only partially offsets declining client sales, Intel guided Q4 revenue to between $15 billion and $16 billion. Intel expects the client market to deteriorate further in Q4 2026 and into 2027 as elevated memory prices and shortages of other components impact PC prices and demand, and expects the client PC market to decline 10% in 2027 compared to 2026. Intel's market capitalization has dropped nearly 50% from its mid-June peak of $673.43 billion to $338.91 billion, including $148.68 billion in one day.
client processorsserver processorsCPUsIntel, Arm and 15 totalTom's HardwarePrice Movement
Arm is expanding its business model from licensing Neoverse core IP and compute subsystem (CSS) technologies to selling finished server CPUs for artificial general intelligence (AGI) applications. Meta is a co-developer and primary customer, while OpenAI also plans to adopt the products. Customers can purchase processors directly, reducing the time and cost required to develop and deploy their own chips, while Arm can increase its chip sales revenue. This is a planned business model expansion; beyond the stated rationale, the source does not state any further cause, and no confirmed shipments or realized sales are reported.
Neoverse core IPcompute subsystem (CSS)finished server CPUsArmDigiTimes Asia
This is an interview with Dr. Sean Wei, CEO of Easy-Logic Technology, and serves as company background. The company was founded in Hong Kong in 2013, grew out of the EDA laboratory at The Chinese University of Hong Kong, and specializes in Functional ECO solutions; its core algorithms won first place in the CAD Contest at ICCAD for three consecutive years (2012–2014). Dr. Wei has served as CEO since 2020 and previously, as CTO, led development of the core algorithms and tool architecture of the flagship product EasylogicECO. The company now serves nearly 100 paying customers, including leading semiconductor design companies in mainland China, Taiwan, the United States, and South Korea. The article states that industry data indicates over 86% of ASICs require two or more tape-outs. EasylogicECO currently supports Functional ECO, Scan Chain ECO, Low Power ECO, and Metal ECO, with Clock Tree ECO, Placement ECO, Timing ECO, and Routing ECO in development; tests show it runs several times faster than traditional tools while producing patches on average 30% better in quality, exceeding 10× in some cases. The company has also introduced the EasyAI ECO Suite, in which Auto Partition reduces patch size by more than 20% on average and up to 90% in some cases, Learning/I-Learning reduces patch size by more than 20% on average, and Smart Caching can reduce runtime by up to 80% in multi-iteration scenarios. Customers typically engage via software licensing, generally proceeding to an annual license after a successful evaluation.
This is a background article describing how AI is reshaping semiconductor design and manufacturing. It relays the message from the recent Global Semiconductor Alliance European Executive Forum that scaling AI requires more than smaller transistors, and also requires addressing power delivery, connectivity, advanced packaging, and supply chain resilience. According to a presentation by Prashanth Aprameyan, Intel Foundry supports this strategy on two fronts: advanced process and packaging technologies, and AI-enabled manufacturing that improves quality and accelerates yield learning. Specific technologies cited include Intel 18A using RibbonFET Gate-all-around technology to improve electrostatic control and support scaling beyond traditional FinFET architectures; PowerVia routing power through the back of the wafer to reduce resistance, voltage drop, and power delivery noise; UCIe providing standardized, high-bandwidth, low-latency die-to-die connections; and a packaging portfolio including EMIB-T, Foveros, and Foveros Direct. The article also states that Intel Foundry's factories collect nearly a billion images each week, generating petabytes of manufacturing data, which machine learning analyzes to detect defects, classify failure patterns, and identify yield risks. The article contains no specific price, output change, or directly tradable event.
Micron announced that about 700 eligible employees in Taiwan will receive five years' worth of salary in bonuses as three bonuses are paid in a single payout; according to Digitimes this covers roughly 17% of the more than 4,300 production-line employees at Micron Taiwan, while the remaining staff will also receive three bonuses in October. The lump-sum payout consists of an annual performance bonus estimated at 7.5 to 12.5 months of salary for the majority of workers, a mass-production bonus of 20% of annual salary, and a one-time $31,000 (NT$1,000,000) appreciation bonus prorated for employees with less than a full year of service. Micron workers across the globe also expect a new quarterly profit bonus paid 50% in cash and 50% in stock, starting January 2027, with the first one paid 100% in cash. The source states this is a direct effect of the AI-driven memory chip boom, which brought Micron, Samsung and SK hynix massive orders from AI hyperscalers and record profits; Samsung workers voted to accept a $340,000 average bonus in May 2026, while SK hynix workers are set to receive $477,000 in bonuses this year, potentially growing to almost $900,000 in 2027. Meanwhile, the Micron Taoyuan union remains in a pay dispute with management: some officials have launched a one-day strike demanding clearer profit-sharing arrangements and bonus terms, nearly 70% of Taoyuan members turned down management's latest offer, and the two sides are tentatively set to meet again on November 6; the Micron Taichung union has already agreed to a 10% employer's pension contribution and a tripling of the day-shift allowance. The article notes that reliance on performance and profit bonuses means these incentives could fall sharply if AI demand declines and the demand cycle normalizes.
An explosion in mid-October at Intel's Fab 28 near Kiryat Gat, Israel killed two people, seriously injured one, and forced the entire facility to shut down for about two weeks. Intel CFO Dave Zinsner said on the company's earnings call that Intel expects the outage to cost approximately $600 million in lost revenue over two quarters, with roughly $300 million revenue reduction in Q4 2026 and a similar impact anticipated in Q1 2027, plus an approximately 2% reduction in Q4 non-GAAP gross margin. The disruption affects server processors made on Intel's 10nm-class process technologies; Zinsner said 'almost all of the $300 impact from Fab 28 is related to 10nm server products,' and shipments will be reduced in both Q4 2026 and Q1 2027. The cause was the explosion at Fab 28. Wafers in fabrication when the accident occurred are being assessed for recovery, with most wafer write-offs expected in Q4 2026, explaining the margin impact concentration in that quarter. Production is set to restart by the end of this week (this Friday), but Intel has not specified when Fab 28 returns to full production capacity, has not identified the affected Xeon processor families, and has not disclosed whether shipments of client CPUs made on Intel 7 (e.g., Raptor Lake) will also be disrupted. Because Intel is struggling to satisfy strong server processor demand and other fabs may lack product-specific qualifications or face high transfer costs, Intel cannot fully offset the lost output at other facilities.
10nm-class server processorsIntel 7 (10nm ESF) manufactured server processorsXeon 6 I/O dies (Intel 7)IntelTom's HardwareDisaster/Disruption
During the Q3 2026 earnings call, Intel CEO Lip-Bu Tan confirmed that the company's upcoming flagship client CPU, codenamed "Nova Lake," will be produced by Intel Foundry using its 18A-P node. Initially, Intel had planned to outsource the main compute tile production for "Nova Lake" to TSMC using its 2 nm N2 node, with TSMC expected to handle about 60-70% of the production and the remainder produced in-house by Intel Foundry on the 18A node, meaning a dual-sourcing arrangement for the next-generation chip. Later supply chain information suggested the company intended to bring the majority of "Nova Lake" compute tile production in-house, with these compute tiles made on the Intel 18A-P node. The CEO stated: "Nova Lake," our first Intel 18A-P client product, is tracking ahead of schedule, with design wins expected to be twice those of "Panther Lake" at launch, and more details will be shared at CES in January. The source does not state the cause for the shift of production in-house. This production arrangement remains planned; no completed production is reported.
Nova Lake client CPU (18A-P compute tile)Intel, Intel Foundry and 3 totalTechPowerUpCapacity Change