Ting-Hao Lin, R&D vice president at Kinsus Interconnect Technology, said at the 2026 TPCA Show that surging demand for AI computing is driving advanced packaging toward larger form factors. To support this trend, substrate materials must deliver low warpage and exceptional dimensional stability while overcoming challenges in thermal expansion mismatches and structural stress. The source states no specific capacity, pricing or expansion plans.
advanced packaging substratesKinsus Interconnect TechnologyDigiTimes Asia
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
US AI startup Anthropic has formally filed for an initial public offering (IPO) with the US Securities and Exchange Commission (SEC), identifying competitive pressure from Chinese AI companies, model distillation, and cross-border technology restrictions among its operational risks. The source does not state the cause beyond these listed risk factors, and provides no specific date, offering amount, or details of any chip export control measures.
Tesla and SpaceX are building Terafab in Texas, with Musk aiming to establish a facility capable of processing 1 million wafers per month within 10 years. Intel has joined as a technology partner, and TSMC has confirmed discussions on potential cooperation. Procurement officials from Tesla and SpaceX visited Korean suppliers at SEMICON West. Whether Terafab will materialize as planned remains uncertain.
LG Uplus said on Oct. 28 that it signed a memorandum of understanding (MOU) with Ecolab Korea, the Korean subsidiary of global water treatment solutions provider Ecolab, at its Yongsan headquarters in Seoul on Oct. 27 to optimize AI data center cooling system operations, and the companies will jointly develop liquid cooling operating standards for next-generation AI data centers, aiming to establish an integrated operating framework covering cooling fluid management and equipment anomaly detection. The companies plan to combine LG Uplus' data center building and operating experience with Ecolab Korea's cooling fluid management technology to conduct joint technical validation, and will first validate the operational stability of the Technology Cooling System (TCS) Loop, a cooling fluid circulation system connecting AI server racks with coolant distribution units. LG Uplus will provide a large-scale liquid cooling test environment under construction at its Pyeongchon 2 data center, designed to validate the cooling performance of AI infrastructure based on next-generation GPU architectures; Ecolab Korea will establish a cooling fluid quality management framework using its proprietary digital water quality monitoring system. The companies also plan to develop an AI-based predictive maintenance system. Based on the validation results, LG Uplus plans to establish an integrated operating framework covering cooling equipment installation, fluid quality management and real-time monitoring. The company did not disclose when the operating standards would be finalized or implemented at actual data centers. The event remains at the planned/agreement stage, not an implemented standard.
liquid cooling systems for AI data centerscooling fluidTechnology Cooling System (TCS) LoopLG Uplus, Ecolab KoreaThe ElecPartnership/M&A
Global Technology said on Oct. 28 that it signed a memorandum of understanding (MOU) with Austria's TriLite Technologies to develop, manufacture and commercialize automotive laser beam scanning (LBS) projection solutions. The companies will first conduct a proof of concept (PoC) for a system that projects images or information from vehicle doors onto the ground, aiming to verify its suitability for automotive applications by the first half of next year. The technology uses light sources installed in vehicle doors or side mirrors to display images and guidance information on the ground, with applications including welcome lighting, exit warnings and pedestrian guidance; minimizing optical module size is essential because the technology must fit into limited spaces inside vehicles. TriLite will supply an ultra-compact LBS optical engine that uses a single MEMS mirror to steer light from RGB lasers and can project onto curved surfaces or the ground without a separate projection lens, while Global Technology will handle overall system design, including drive electronics, module integration, optics, thermal management structures and vehicle interfaces, with plans to validate brightness, operational reliability and safety under actual vehicle conditions. After completing the PoC, the companies will consider expanding the partnership to mass production, prioritizing discussions on domestic manufacturing of LBS optical engine modules and joint development of automotive SoCs; under the proposed arrangement, TriLite would be responsible for algorithms and technical specifications, while Global Technology would handle semiconductor design implementation.
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.
Reports of internal display damage have emerged for Apple's first foldable smartphone, the iPhone Duo. According to industry sources on Oct. 28, following the device's launch on Oct. 23, devices reported in the United States, Australia and other markets showed similar symptoms, including cracks near the lower portion of the central fold and long white lines extending to the right; some users said they heard a cracking sound after folding and unfolding several times, followed by screen damage. The leading hypothesis is that metal components inside the external hinge may have broken or malfunctioned, damaging the back of the display; cracks in similar locations across multiple countries have raised concerns about a possible structural defect in the hinge itself. The external hinge consists of more than 100 precision components; if a connecting component breaks or the support plate fails to move properly, the metal structure could press against the back of the OLED panel, and damage to internal OLED wiring could cause bright lines. Apple uses 3D printing for the titanium hinge cover, and observers suggested its adoption, potentially motivated by cost reduction, may be linked to hinge durability issues. The device had hinge production yield problems during development; the primary external hinge supplier was changed from U.S.-based Amphenol to Taiwan's Shin Zu Shing, which also reportedly supplied 'golden samples' meeting Apple's standards. Foxconn's assembly yield reportedly stood at slightly above 60% in mid-September, with yield problems of key components such as the hinge and OLED display, supply delays and design changes cited as possible causes. Supply chain: Shin Zu Shing and Amphenol supply the external hinges; South Korea's Fine M-Tec reportedly supplies the internal hinge (backplate); Samsung Display is the exclusive foldable OLED panel supplier; upper UTG is supplied by Lens Technology and South Korea's UTI; lower UTG by Lens Technology and China's Biel Crystal. Observers believe machining or assembly issues in the external hinge could be resolved through process improvements, but a design change may be necessary if the problem lies in the hinge structure itself. Apple said it is aware of an issue affecting internal displays of a small number of iPhone Duo devices and that users should contact AppleCare for a replacement.
internal OLED display (foldable OLED panel)external hingecentral support plateApple, Amphenol and 9 totalThe Elec
Related manufacturers
Apple, Amphenol, Shin Zu Shing, Fine M-Tec, Samsung Display, Lens Technology, UTI, Biel Crystal, Foxconn
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.
Seagate said on Oct. 28 that its Mozaic drives have become the only shingled magnetic recording (SMR) products included on Broadcom's host bus adapter (HBA) approved vendor list, securing compatibility between its high-capacity hard disk drives and Broadcom components that connect servers and storage devices. Mozaic is Seagate's HDD platform incorporating heat-assisted magnetic recording (HAMR) and SMR technologies: HAMR increases recording density and drive capacity, while SMR records data tracks in overlapping patterns resembling roof shingles. An HBA is a component used in data centers to connect multiple HDDs and identify and manage individual drives. Seagate also announced its fiscal first-quarter 2027 results on Oct. 28: revenue for the quarter ended Oct. 2 rose 61% to $4.233 billion from $2.629 billion a year earlier, and GAAP net income reached $1.761 billion, 3.2 times the $549 million recorded a year earlier. Chairman Dave Mosley said growth in data creation, retention and reuse is expanding storage demand and that long-term visibility into customer demand supports the company's growth outlook.
high-capacity HDDhost bus adapter (HBA)Seagate, BroadcomThe Elec
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
AXT Inc of Fremont, CA, USA — which makes gallium arsenide (GaAs), indium phosphide (InP) and germanium (Ge) substrates and raw materials at plants in China — says that its majority-owned China-based wafer manufacturing subsidiary Beijing Tongmei Xtal Technology Co Ltd has entered into definitive agreements for a private equity financing. Under the agreements, certain institutional investors would make a private equity investment in Tongmei in ordinary shares of a minimum of RMB1,912,350,000 (about US$285m) and up to a maximum of RMB3,668,639,053.25 (about US$545m). The financing is intended to fund further expansion of Tongmei's manufacturing capacity for indium phosphide substrates, expansion of its principal business operations, capital expenditures, and general working capital. Definitive agreements have been signed, but the financing and the capacity expansion have not been implemented or completed.
Microsoft and Nvidia publicly released the Surface Laptop Ultra on October 7, built around the application processor originally referred to as the N1X and now renamed the GB10. The chip was co-developed by NVIDIA (which contributed the Grace Blackwell GPU subsystem) and MediaTek (which supplied the multi-core CPU cluster and reportedly also handled full SoC integration duties); it was originally intended for—and may eventually still show up in—Arm-based Windows PCs, but repeated development hurdles reportedly delayed both SoC and system shipment plans, and lingering functional bugs precluded Windows compatibility (explaining the DGX Spark's Linux OS foundation). The machine comes in two versions sharing the same NPU core: 5120-core GPU with 18-core CPU, and 6144-core GPU with 20-core CPU; system memory ranges from 24 GB to 128 GB of LPDDR5x unified memory; SSD options start at 512 GB (PCIe Gen 4) with 1 TB and 2 TB (PCIe Gen 5) configurations; pricing is $2,599 baseline up to $5,899.99 top-end, with delivery promised 'later this year'; an RTX Spark-based developer platform at $5,999.99 and up is also available for pre-order. In the same period, Amazon launched its latest-generation Alexa-branded (no-longer-Fire) tablets in 8-inch, 11-inch and 12-inch 'Pro' flavors, running 'vanilla' Android with a return to the Google Play Store, priced from $299 up to $679.99 (not including cases or the high-end model's optional pen and keyboard accessories). Apple's week began Monday with a new-generation iPad mini (now with the A20 Pro, a slightly larger 8.5-inch display switched from LCD to OLED, priced from $699 and up; the 2024 entry-level 128 GB Wi-Fi-only model was originally $499 but was bumped to $599 in June of this year), Tuesday brought a 14-inch entry-level MacBook Pro now running the M6 SoC (versus last year's M5) plus a 24-inch iMac making a two-generation jump from the M4 to the M6, and today Apple unveiled its first touchscreen-augmented MacBook Pro laptops in both 14-inch and 16-inch displays with ProLED XDR panels and both M5 Pro and Max SoC options, priced from $2,899 to $14,549.
Nvidia RTX Spark N1X (GB10) application processorApple A20 Pro SoCApple M6 SoCApple, Microsoft and 6 totalEDN
TetraMem has completed tapeout of the MLX300, a 12-nm analog in-memory computing (A-IMC) SoC based on multi-level RRAM technology. The chip was developed using Synopsys EDA software delivered through Synopsys Cloud FlexEDA, extending multi-level RRAM-based AI chip development to advanced FinFET nodes. The MLX300 tapeout follows TetraMem's MLX200/MLX200M multi-level RRAM A-IMC SoCs, which completed full-mask tapeout in 2026 on a commercial 22-nm process. The MLX300 uses TetraMem's architecture in a single-chip accelerator for audio, vision, and smaller language models, supporting up to 64 million INT4 parameters on-chip without external DRAM for model-weight storage; TetraMem says the SoC targets energy efficiency exceeding 50 TOPS/W.
This is a market roundup (paid press release) reporting that major U.S. indices pushed higher midweek on semiconductor and tech earnings: the S&P 500 rose 1.51% over the week's three sessions to close at 7,739.13; the Nasdaq Composite rose 1.98% for the week to 26,468.60; the Dow gained 0.62% for the week to 52,536.29; and the VXN, a measure of Nasdaq volatility, fell 3.51 points for the week to 24.79. The source attributes the gains to semiconductor and tech earnings. On earnings: Texas Instruments reported third-quarter revenue of $6.06 billion, up 28% from the same quarter a year ago, with net income of $2.43 billion, EPS of $2.62, and operating profit up 73% to $2.87 billion, and guided fourth-quarter revenue of $5.80 billion to $6.30 billion with EPS of $2.45 to $2.79. On Broadcom: Bank of America, Blackstone, Citi and Morgan Stanley announced a $63 billion financing structure, executed through a special purpose vehicle to purchase Broadcom XPU AI accelerators and networking solutions to be leased to Anthropic, with deliveries expected starting in calendar 2027 and Broadcom providing residual value support equivalent to the first lien portion of the financing. The article also mentions Meta, IBM and Lockheed Martin developments as additional company news.
Marvell is reportedly considering acquiring Lumentum to expand its product portfolio. Lumentum produces optical communication lasers and optical transceiver modules and currently has a market capitalization of approximately US$83 billion, compared with Marvell's US$221 billion; Lumentum has also attracted a US$2 billion investment from NVIDIA. A potential acquisition would further broaden Marvell's product portfolio, fill gaps in its optical component offerings, and strengthen its ability to compete with Broadcom. Bernstein analysts cautioned that the potential deal's size could raise concerns over financing and integration, noting Marvell may be unable to fund it entirely with cash; as of its latest quarter Marvell held approximately US$3.92 billion in cash and cash equivalents. Background: earlier this year Marvell acquired photonics company Celestial AI, gaining access to its Optical Multi-Chip Interconnect Bridge (OMIB) technology, and reached a chip agreement with Google granting Google the right to purchase up to US$12.2 billion worth of Marvell shares. The deal remains at a reported, considering stage and is not confirmed as agreed or completed.
According to an October 26 report by Changsha Evening News, Hunan Titan Future Technology has developed China's first 12-inch polycrystalline silicon carbide (SiC) substrate. The substrate offers thermal conductivity exceeding 280 W/(m·K) and is designed for applications requiring high thermal conductivity, including AI computing chips and advanced packaging, serving as a heat-conducting base within semiconductor packaging structures to help alleviate heat buildup during high-power chip operation. The product is currently undergoing sample validation and small-batch trial production and has completed validation by leading companies. Titan Future's technology builds on research by the Powder Metallurgy Research Institute at Central South University, and the company had previously completed domestic development of 6-inch and 8-inch polycrystalline SiC substrates. On September 28, a project invested in by Titan Future Semiconductor (Jiangsu) broke ground in Jinhu Economic Development Zone, Jiangsu Province, to establish an integrated manufacturing base for high-purity, high-density SiC components and polycrystalline substrates, primarily producing high-performance SiC components and TaC-coated products for semiconductor applications; once fully operational, it is expected to have annual production capacity of 200,000 units, generate CNY 600 million in annual sales, and contribute an additional CNY 30 million in tax revenue. Titan Future also plans facilities in Changsha, Xiamen and Jinhu, with Changsha as its technology development and application incubation center and the Xiamen facility planned to have annual capacity for 500,000 composite SiC substrates.
According to an exclusive Reuters report, CXMT is expected to start mass production of its first-ever HBM memory in the HBM3E form factor within the next two months, meaning it will officially start by the end of the year. CXMT began risk production of HBM3E memory in August, but the company has struggled with the yield of HBM3E memory modules, and this situation remains similar today, according to the latest report. The yield challenge stems from the immaturity of the through-silicon via (TSV) technology it employs: front-end manufacturing achieves around 30% yields, while back-end processing adds about 70% on top of the original 30%. Data suggests CXMT is attempting to achieve HBM3E with only 3,000 TSVs per layer and trying to pack 8 layers into the regular 8-high HBM3E configuration (compared with Samsung's roughly 5,000 TSVs per layer on HBM2 and SK hynix's more than 8,000 TSVs per layer on HBM3). Given these challenges, the company is unlikely to transition to a higher-capacity 12-high HBM in the near term as it addresses engineering issues. JEDEC's official data for HBM3E indicates a 1,024-bit width and speeds from 9.2 Gbps per pin to 12.4 Gbps, with most standard configurations targeting 9.6 Gbps per pin, total bandwidth around 1.2 TB/s, and capacities from 24 GB using 8-high stacks to 36 GB using 12-high stacks of 24 Gb layers. The mass production start is an expectation, not a confirmed completed event.
According to unconfirmed rumors, NVIDIA's RTX 60 series could be delayed until 2028, with similar speculation surrounding AMD's RDNA 5; if true, consumer GPU upgrade cycles could stretch to roughly three years. NVIDIA previously launched the RTX 30 series in 2020, the RTX 40 series in 2022, and the RTX 50 series in early 2025, maintaining a cycle of just over two years. The source states the cause: AI is a major factor — NVIDIA's August earnings report showed quarterly Data Center revenue of US$89 billion, approximately 92.5% of its US$96.2 billion total revenue; gaming is no longer its primary revenue driver, with R&D resources and memory supplies increasingly directed toward AI. Amid strong AI accelerator demand and rising memory prices, the RTX 50 SUPER series has reportedly been put on hold due to GDDR7 costs. Another factor is weakening upgrade incentive: TechSpot's testing found the RTX 5080 outperformed the RTX 4080 SUPER by an average of just 4% at 1440p and 11% at 4K in traditional gaming benchmarks without frame generation. Integrated graphics advances (Intel's Panther Lake Arc B390, AMD's Strix Halo Radeon 8060S, and the Blackwell GPU in RTX Spark) are also pressuring entry-level discrete GPUs. The source notes these performance results do not represent the entire RTX 50 lineup, and the delay reports remain conditional ('if true') speculation.
SK Hynix reported Q3 2026 revenue up 328% Y/Y to ₩104 trillion ($78.9 billion), surpassing ₩100 trillion for the first time, with net income up 216% to ₩39.8 trillion (c$30 billion), an operating margin of 80 percent, and a net cash position of ₩98.7 trillion ($74 billion), effectively debt-free. The company abandoned the US IPO of its Solidigm subsidiary and discontinued its review, citing careful consideration of recent financial market conditions and other factors; the IPO had been intended to raise external capital for expanding Solidigm's production facilities. SK Hynix said HBM4 is becoming its mainstream product as of this quarter and should overtake HBM3E in volume and revenue next quarter; it has 14 LTAs (Long-Term buying/supply Agreements) covering more than half of its memory capacity for 2027 and 2028, and it sees memory demand exceeding supply through to 2030. Its spending decisions include returning most of its Kioxia share gain to shareholders, with ₩100 trillion ($75.4 billion) going to shareholders this year; ramping the M15X fab's output to maximum capacity by the end of the year to increase HBM4 production; raising the proportion of its 321-layer NAND process to more than half of its domestic production capacity in the fourth quarter to increase high-performance, high-capacity eSSD supply; and rising production output in 2027 as the Yongin Phase 1 plant enters full-scale operation. Paying for a US fab for Solidigm remains unsolved, with the company still able to use about another $30 billion, per the source.