Samsung reportedly starts Tesla chip trial production at Taylor fab

digitimes.com Sep 17, 2026

Samsung reportedly starts Tesla chip trial production at Taylor fab

Curated by Shubham · TechCapsules · Sep 17, 2026

AI-summarised brief · reviewed before publication

Samsung Electronics has reportedly begun trial production of Tesla’s next‑generation AI5 processor at its Taylor foundry in Texas. The move marks the first 2 nm wafer production at the U.S. facility and includes yield verification before full‑scale manufacturing. The trial is part of Tesla’s strategy to secure advanced AI chips for its vehicles and data centers, while Samsung expands its high‑performance semiconductor portfolio. The partnership underscores the growing collaboration between automotive and semiconductor giants in the AI hardware race.

💡 Why It Matters

· The trial signals a pivotal shift toward domestic 2 nm chip production, reducing reliance on overseas fabs and accelerating Tesla’s AI capabilities.

Read full article on digitimes.com

Ireland’s Emergence as a Full-Stack Semiconductor Powerhouse

Ireland has developed a comprehensive semiconductor ecosystem over five decades, now highlighted by a 2040 national strategy that consolidates EDA/IP, design, fabrication, and advanced packaging within a single cluster anchored by Analog Devices. The country participates in the European Chips Act, co‑investing in three pan‑European pilot lines for sub‑2‑nanometer chips, fully depleted silicon‑on‑insulator, photonics, and advanced packaging. A new coordinating research institute, Tyndall National Institute, and a €100 million government fund aim to support future manufacturing infrastructure. Talent acquisition remains the critical challenge, with AI chip firms already present and a university pipeline being mapped to meet future needs.

💡 Ireland’s integrated semiconductor strategy positions it as a pivotal player in Europe’s high‑tech supply chain, potentially reshaping regional manufacturing and innovation dynamics.

international.astroawani.com · Sep 17

China’s Huawei sets 2027 launch for new AI chips as it targets…

Huawei Technologies announced it will introduce two new artificial‑intelligence processors in 2027, aiming to compete with U.S. chipmaker Nvidia. The 960DT is slated for launch in the first quarter, followed by the Ascend 960PR in the third quarter, said rotating chairman David Wang at a Shanghai event. Huawei also unveiled its UnifiedBus interconnect, which it says will enable massive AI chip clusters, linking up to one million processors in “superclusters” and supporting more than 1,000 smaller “supernodes” already shipped to over 370 customers. The company highlighted 11 semiconductors built on UnifiedBus and stressed that chip‑to‑chip communication is critical as U.S. export controls limit China’s access to advanced foreign silicon. Huawei expects the chips to power data‑center workloads and AI services.

💡 Linking millions of domestic AI cores gives China a high‑performance computing route that bypasses U.S. chip restrictions.

9to5mac.com · Sep 17

Some memory for Apple devices may be made in the US for…

Apple may see its memory chips produced in the United States for the first time if a deal between supplier SK Hynix and Intel materializes. SK Hynix, which supplies both DRAM and NAND flash to Apple, is reportedly negotiating to manufacture chips at Intel’s planned Ohio plant. The arrangement would allow Apple to source memory domestically, a shift from its current reliance on overseas suppliers. The proposal faces potential hurdles from South Korea, which may oppose the export of chip‑making technology.

💡 Domestic production could reduce Apple’s supply‑chain exposure to geopolitical tensions and strengthen U.S. semiconductor manufacturing capacity.

semiconductor-digest.com · Sep 17

Micron Advances Memory Innovation With the World’s First Ultra-Dense Module for Next-Generation…

Micron Technology unveiled the world’s first 512‑GB DDR5 RDIMM, demonstrated on multiple server platforms. The module, built with vertical interconnect packaging and through‑silicon vias, delivers up to 9,200 MT/s and enables 12 TB of DDR5 in a 24‑slot dual‑socket server. AMD and Intel are validating the design, while Micron highlights power savings of over 60 % versus four 128‑GB modules and performance gains for AI, analytics, and in‑memory databases. Production is targeted for the second half of 2027.

💡 The new module lets data centers keep larger datasets in main memory, cutting reliance on slower storage and boosting throughput for AI and database workloads. It also offers significant power efficiency, helping operators reduce operating costs while scaling compute density.

semiconductor-digest.com · Sep 17

Arteris Accelerates Industry Transition to Multi-Die Chiplet-Based Architecture

Arteris, Inc. announced an expansion of its multi‑die portfolio aimed at easing semiconductor firms’ shift from monolithic system‑on‑chip (SoC) designs to chiplet‑based architectures for AI, high‑performance computing (HPC) and consumer electronics. The new FlexGen Multi‑Die network‑on‑chip (NoC) IP extends connectivity across die‑to‑die links using a single UCIe PHY, cutting PHY area, power and I/O needs by up to 50 % and adding virtual‑channel link technology and quality‑of‑service controls for non‑coherent AI and HPC workloads. It complements the existing Ncore Multi‑Die solution for cache‑coherent systems and integrates with Arteris’ Magillem automation and Cycuity security tools. Together, the suite lets engineering teams reuse designs, simplify integration and maintain performance as they partition systems across multiple silicon dies.

💡 By preserving NoC intelligence across chip boundaries, Arteris removes a major bottleneck that has slowed the adoption of larger, heterogeneous chiplet ecosystems. This accelerates time‑to‑market for next‑generation AI and HPC products.

businessworld.in · Sep 17

ASML Looks To Expand India Supply Chain, Engineering Capabilities

ASML executive vice‑president Wayne Allen told attendees at Semicon India 2026 that the Dutch lithography equipment maker is evaluating ways to expand its supply‑chain and engineering footprint in India. He said the country’s large market, strong talent pool, established chip‑design firms and existing research base create a foundation for a broader semiconductor ecosystem. ASML is mapping local suppliers and partners to support various stages of the value chain, aiming to boost quality, efficiency and scale for current manufacturers. The company also announced a collaboration with Tata Electronics to assist Tata’s planned 300 mm fab in Dholera, Gujarat, providing lithography expertise. Allen emphasized that long‑term cooperation among government, industry, academia and research institutes will be essential for a sustainable Indian semiconductor sector.

💡 By cultivating home‑grown suppliers, India can lower its dependence on traditional East‑Asian chip supply chains, accelerating the country's move toward self‑sufficiency in advanced manufacturing.