DIGITIMES: News and Insight of the Global Supply Chain

JCET lifts 2026 capex to CNY10 billion as AI packaging demand outruns forecasts

DIGITIMES — Tue, 1 Sep 2026 02:55:00 GMT

JCET Group is sharply increasing spending as demand for artificial intelligence strains advanced packaging capacity worldwide. The Chinese chip tester and assembler said the shift could influence supply chains beyond China, as more AI and automotive chips rely on the back-end processes it is now prioritizing.

XCENA teams up with Samsung on rack-scale CXL memory; demos on Intel Xeon 6

DIGITIMES — Tue, 1 Sep 2026 02:41:34 GMT

South Korean chip startup XCENA is pushing CXL beyond memory expansion and toward a new layer of server computing, unveiling measured results for its MX1 computational memory device at Hot Chips 2026 while expanding its work with Samsung Electronics on a rack-scale system.

Samsung locks 70% of memory output as HBM spot prices soar

DIGITIMES — Tue, 1 Sep 2026 02:34:14 GMT

Samsung Electronics has reportedly locked up as much as 70% of its memory production capacity under long-term supply agreements (LTAs), as relentless AI infrastructure demand for high bandwidth memory (HBM) tightens the broader chip shortage. In some cases, spot prices for memory products are now as much as 5 times the LTA contract price.

Cisco: CPO is ready for mass deployment amid surging AI data-center demand

Jingyue Hsiao — Tue, 1 Sep 2026 02:29:28 GMT

For global cloud operators and network vendors, Cisco said co-packaged optics (CPO) is moving from lab promise to commercial necessity as AI data centers strain power budgets and bandwidth limits. The company said pluggable modules, near-packaged optics, and co-packaged optics will coexist, but the shift could reshape how hyperscale infrastructure is built worldwide.

Huawei's revenue climbs 9.6%, but AI, chips and HarmonyOS exact heavier R&D toll

DIGITIMES — Tue, 1 Sep 2026 02:26:09 GMT

Huawei Technologies increased first-half revenue in 2026 while profit fell sharply, highlighting the growing cost of its effort to build a self-sufficient technology stack spanning AI chips, smartphones, operating systems and computing infrastructure.

South Korea-India joint venture opens semiconductor materials plant, targets China-free supply chain

DIGITIMES — Tue, 1 Sep 2026 02:23:48 GMT

A South Korean-Indian joint venture has completed a semiconductor materials plant in South Korea, aiming to strengthen regional supply chains for photoresist materials and reduce reliance on Chinese raw materials.

Commentary: CXMT's HBM3E leap gives China's AI chipmakers a new path to scale

DIGITIMES — Tue, 1 Sep 2026 02:20:47 GMT

China's leading DRAM maker CXMT has reportedly begun small-volume production of HBM3E, a step that could ease one of the most important hardware constraints on China's domestic AI chip industry. US export controls have restricted Chinese access to advanced high-bandwidth memory (HBM), making local supply increasingly critical to scaling homegrown AI processors.

China's humanoid robot cost cuts run into three bottlenecks

Lily Hess — Tue, 1 Sep 2026 02:20:14 GMT

Humanoid robot scaling is still constrained by price, and Chinese humanoid robots — which command the largest global market share — show a sharp gap between consumer and industrial models. Industrial versions demand high load, high precision, and strong stability, making several precision transmission parts a key battleground for Chinese players and Taiwan suppliers.

India expands chip ambitions with full Semicon 2.0 notification

DIGITIMES — Tue, 1 Sep 2026 02:16:38 GMT

India has unveiled the next phase of its semiconductor policy, a move that could reshape global chip supply chains, investment flows, and talent markets. The expanded program aims to deepen local design and manufacturing capacity, while giving multinational companies and developers a larger role in one of technology's most strategic sectors.

China reportedly moves space computing from concept towards commercial deployment

DIGITIMES — Tue, 1 Sep 2026 01:42:13 GMT

China's push to build computing infrastructure in orbit is moving beyond experimental projects, with several initiatives this year pointing towards a more coordinated effort to develop space-based AI and data-processing capabilities.

Taiwan clears Lam Research's NT$9.8 billion subsidiary injection as etch leader scales R&D worldwide

DIGITIMES — Tue, 1 Sep 2026 01:41:47 GMT

Taiwan's Ministry of Economic Affairs has approved a NT$9.835 billion (approx. US$310 million) capital injection by Lam Research into its Taiwanese subsidiary, Lam Research Taiwan, with the funds earmarked for research, development, and design work on critical equipment used in advanced semiconductor manufacturing processes. The Department of Investment Review cleared the case on August 31 as one of six major investments, according to the Taipei Times, citing a ministry statement. The session approved two inbound foreign investments totaling NT$12.835 billion and four outbound investments worth US$358.36 million.

Taiwan's G2C+ Alliance members target global markets

Eifeh Strom — Tue, 1 Sep 2026 01:28:18 GMT

Artificial intelligence (AI) chips are driving a new wave of integration competition among Taiwan semiconductor equipment makers. Taiwan's G2C+ strategic alliance, comprised of C Sun, Gallant Precision Machining (GPM), Gallant Micro Machining (GMM), and Contrel Technology (CT), focuses on advanced packaging, AI smart manufacturing, and global markets.

SEMICON Taiwan 2026 to spotlight AI-driven semiconductor supply chain shifts

DIGITIMES — Tue, 1 Sep 2026 01:16:21 GMT

SEMICON Taiwan 2026 is set to open soon as Semiconductor Equipment and Materials International said AI is pushing a new wave of technology upgrades across the global chip industry. At an August 31 pre-event briefing, the group said the changes are accelerating work in AI computing, advanced process technology, heterogeneous integration, and quantum computing.

Nvidia's MediaTek investment signals a broader push to shape AI infrastructure

Jingyue Hsiao — Tue, 1 Sep 2026 00:54:20 GMT

Nvidia's US$3.5 billion investment in MediaTek's overseas convertible bonds marks more than a financial bet. It underscores how the chip leader is trying to anchor its AI infrastructure strategy as cloud giants and model developers expand their use of custom silicon and rack-scale systems.

Nvidia and MediaTek link AI technologies in bid to accelerate custom data center chip adoption

Jingyue Hsiao — Tue, 1 Sep 2026 00:44:03 GMT

Nvidia and MediaTek are widening a global AI partnership that could reshape data center, edge, and PC hardware markets. MediaTek's US$3.9 billion overseas convertible bond sale, largely bought by Nvidia, signals a deeper strategic tie that may affect supply chains, customer choices, and competition worldwide.

China's Xenomi LLM brings quantum computing into AI agent decision-making

DIGITIMES — Mon, 31 Aug 2026 23:42:40 GMT

China is testing a new route for combining quantum computing with generative AI, moving beyond model training toward the decision layer used by AI agents. Beijing Guoguang Liangchao Technology has introduced Xenomi, a family of large language models that uses quantum annealing for selected routing, tool-selection, and task-decision workloads while leaving language understanding to conventional AI models.

Buima returns to profit in the second quarter on battery module demand

DIGITIMES — Mon, 31 Aug 2026 23:42:04 GMT

Buima swung back to profitability in the second quarter of 2026 as a recovery in its battery module shipments and gains from asset disposal lifted results. The company said demand tied to low Earth orbit satellites, wearable devices, and other applications helped improve profitability, while margin measures and earnings per share reached quarterly records.

Commentary: Why America fell behind on Open models — and how Nvidia plans to close the gap

Sherri Wang — Mon, 31 Aug 2026 23:41:27 GMT

When DeepSeek burst onto the scene in early 2025, it wiped nearly US$600 billion from Nvidia's market value in a single day. By the summer of 2026, Moonshot AI's Kimi K3 had triggered another wave of anxiety, sharpening divisions in Silicon Valley over the growing strength of China's open models.

Nvidia targets HBM race with custom NVHBM, expands NVLink Fusion

Yunnie Cheng — Mon, 31 Aug 2026 23:40:51 GMT

As trillion-parameter large language models (LLMs) and agentic AI become mainstream, the AI infrastructure bottleneck is shifting from raw compute power toward system-level integration across compute, memory, interconnects, and thermal management. To bypass traditional hardware constraints, Nvidia announced NVHBM, its next-generation custom HBM technology, and integrated it into the NVLink Fusion ecosystem.

Taiwan workplace AI use reaches 96% as employers push skills upgrades

DIGITIMES — Mon, 31 Aug 2026 23:40:34 GMT

Ninety-six percent of professionals in Taiwan use AI in the workplace, topping the Asia-wide average of 93%, according to a new survey released by Robert Walters. The report, which covered nearly 5,000 professionals and employers across 10 Asian markets, showed that Taiwanese companies are already using AI far beyond basic tasks and are preparing for broader changes in job design and skills.

Samsung weighs HBM4 bonuses as memory, foundry teams align

Lily Hess — Mon, 31 Aug 2026 23:40:20 GMT

Samsung Electronics is using a cross-division "One Team" model to develop its sixth-generation high-bandwidth memory, HBM4, but the approach is creating a new challenge for the newly established Device Solutions (DS) special management performance bonus system. Labor and management are now negotiating detailed payout rules to clarify how One Team participants and some contract workers will be rewarded and to prevent employees from being disadvantaged because they belong to different units.

AI server cooling specialist Nidec expands Vietnam capacity, forms liquid-cooling team

Levi Li — Mon, 31 Aug 2026 23:39:42 GMT

Nidec Chaun-Choung Technology plans to establish a dedicated liquid-cooling team and deepen cooperation with major cloud service providers, citing rising thermal-management demand from AI servers. The company also plans to expand cooling capacity beyond its existing operations in Kunshan, China, to Hanoi, Vietnam, with mass production and shipments expected to begin in 2028.

Humanoid robots gain a sense of touch, lifting tactile sensor adoption above 60%

Levi Li — Mon, 31 Aug 2026 23:38:26 GMT

If vision serves as a robot's eyes, tactile sensors function as its skin and peripheral nerves. In China, leading dexterous-hand makers have increasingly begun treating tactile sensing as a standard feature. The shift was particularly visible at the 2026 World Robot Conference, where queues formed throughout the day at tactile-glove demonstration areas as visitors took turns grasping objects and watching real-time changes in touch data. According to Chinese market research estimates, tactile-sensor penetration in dexterous robotic hands has risen from about 20% in early 2025 to more than 60% in 2026.

Taiwan manufacturing moves into the prosperity zone as US export growth slows

DIGITIMES — Mon, 31 Aug 2026 23:38:02 GMT

Taiwan’s manufacturing business-climate indicator moved into the index’s red “prosperity” category in July, with the Taiwan Institute of Economic Research (TIER) saying on August 31 that a sharper rise in optimism among local manufacturers lifted its operating environment index. At the same time, year-over-year export growth to the US fell to 28.0% from 267.6% in January, the lowest level since February 2025.

Acme sees years of SiC growth as data center demand builds

Sherri Wang — Mon, 31 Aug 2026 23:37:42 GMT

Capex by the world's nine largest cloud service providers (CSPs) is expected to rise about 90% year over year in 2026, while AI servers are growing faster than conventional servers, increasing demand for power and energy-management solutions.

OpenAI compresses Jalapeno design to 9 months

Charlene Chen — Mon, 31 Aug 2026 23:37:17 GMT

OpenAI unveiled performance results for its in-house AI chip Jalapeno at Hot Chips 2026 ahead of Nvidia's latest earnings report, underscoring not whether it can beat GB200 or GB300, but how unusually fast it moved from design to tape-out.

Luxnet eyes orders through 2028, DCI revenue set to double in 2027

Lily Hess — Mon, 31 Aug 2026 23:11:22 GMT

AI-driven demand for optical transceiver modules is rising rapidly, and LuxNet said its business in the second half of 2026 will grow sharply from the first half. With capacity coming online in 2027, revenue from data center interconnect (DCI) products is expected to multiply, driving a leap in overall operations, while 1.6T optical modules are slated to enter mass production in the second half of 2027.

Kaimei sees 2H26 momentum strengthen as AI demand lifts orders and resistor price hikes take effect

Scarlett Yu — Mon, 31 Aug 2026 23:10:57 GMT

Kaimei Electronics, a Yageo Group company, said order visibility remains strong for the second half of 2026 as AI server demand continues to ramp and automotive and industrial-control markets recover. Across its three major product lines, the average book-to-bill (B/B) ratio reached 1.2 in June and July, while pricing pressure has eased significantly, supporting a more optimistic operating outlook.

Silergy data center business surges as new products scale up

Lily Hess — Mon, 31 Aug 2026 23:10:36 GMT

Silergy said at a recent earnings call that most of its applications, excluding its computing product line, posted quarter-over-quarter growth of more than 20% in the second quarter of 2026. Automotive revenue rose about 50% year-over-year, industrial revenue increased 49.7%, and data center-related applications grew nearly 90% year-over-year.

LGD launches PFAS exit roadmap, targets full phaseout by 2035

DIGITIMES — Mon, 31 Aug 2026 23:10:19 GMT

LG Display (LGD) has introduced a new "PFAS-free" component verification process for suppliers as it prepares for expected European Union (EU) restrictions on PFAS, aiming to eliminate the so-called "forever chemicals" across all products by 2035.

Samsung is said to develop 8-layer HBM4E for Nvidia's NVHBM push

DIGITIMES — Mon, 31 Aug 2026 23:10:12 GMT

Samsung Electronics is reportedly developing an 8-layer version of seventh-generation high-bandwidth memory for Nvidia, matching the chipmaker's requirements for its customized NVHBM platform. According to Seoul Economic Daily, the shift could give Samsung an early position in the next-generation artificial intelligence memory market if development moves ahead smoothly.

Chinese OSAT earnings surge on AI packaging demand, but recurring profit tells a narrower story

DIGITIMES — Mon, 31 Aug 2026 23:09:45 GMT

Five listed Chinese outsourced semiconductor assembly and test providers reported first-half 2026 results in the last two weeks of August, showing net profit growth of up to 317%, yet the figures that strip out one-off items reveal a far more uneven recovery — one in which two companies are carrying most of the operating improvement, one is still losing money on its core business, and the sector's capital commitments now run well ahead of what it currently earns.

AI Macs, Part 2: Mac mini, Mac Studio gain enterprise AI ground; Apple still lacks the platform

DIGITIMES — Mon, 31 Aug 2026 23:09:35 GMT

Apple's AI hardware momentum is forcing an enterprise question it did not plan for. Mac mini and Mac Studio are increasingly being used as local AI infrastructure, but Apple's business remains built primarily around consumer devices and services.

China specialty foundry XMC plans fourfold capacity growth around 3D integration, silicon photonics

DIGITIMES — Mon, 31 Aug 2026 23:09:25 GMT

Wuhan Xinxin Semiconductor Manufacturing Company (XMC) has laid out a 10-year expansion strategy centered on 3D integration, silicon photonics (SiPh), and specialty processes, targeting a fourfold increase in production capacity while building a broader semiconductor ecosystem around Wuhan.

SK Hynix explores Intel as second source for HBM base dies

DIGITIMES — Mon, 31 Aug 2026 23:09:10 GMT

SK Hynix is reportedly considering adding Intel Foundry as a second manufacturing source for high-bandwidth memory (HBM) base dies, a move that could reduce its reliance on TSMC as the logic layer beneath HBM becomes increasingly important to cost, performance and accelerator design.

John Ternus takes over an Apple with one growth engine, no obvious second

DIGITIMES — Mon, 31 Aug 2026 23:08:57 GMT

John Ternus becomes Apple's chief executive on September 1 with comfortable near-term numbers and an uncomfortable medium-term question. Apple's June quarter was a record. Its largest business has not sold materially more units in a decade, its newest hardware category is shrinking, and the technology reordering the industry runs, for now, on someone else's model.

Analysis: Tim Cook hands Ternus a slower-growing but far richer Apple than the one Steve Jobs left him

DIGITIMES — Mon, 31 Aug 2026 23:08:46 GMT

When John Ternus becomes chief executive on September 1, the company he inherits will be measured on a scorecard Tim Cook wrote, not the one Steve Jobs left behind. Fifteen years of accounts show two different machines wearing the same logo: Jobs ran a company that accelerated, Cook ran one that compounded. The distinction matters for Ternus, because the growth rates Cook normalized are the ones he now has to beat.

Nvidia takes US$3.5 billion of MediaTek's record US$3.9 billion convertible bond

Joseph Chen — Mon, 31 Aug 2026 14:25:54 GMT

MediaTek has completed pricing on US$3.9 billion of overseas convertible bonds, more than NT$120 billion and what the company says is the largest overseas CB ever issued in Taiwan's capital market. Nvidia subscribed for US$3.5 billion of it — roughly 90% of the offering.