AMD’s Lisa Su Visits South Korea, Reportedly Discusses AI Collaboration With Samsung, SK Hynix

newsable.asianetnews.com Oct 8, 2026

AMD’s Lisa Su Visits South Korea, Reportedly Discusses AI Collaboration With Samsung, SK Hynix

Curated by Shubham · TechCapsules · Oct 8, 2026

AI-summarised brief · reviewed before publication

AMD chairperson and CEO Lisa Su traveled to South Korea on Wednesday, meeting senior executives from Samsung Electronics’ Device Solutions division and SK Hynix to deepen supply‑chain ties and expand high‑bandwidth memory (HBM) collaborations for AI accelerators. She also convened 13 local AI startups, urging integration of their hardware and software into AMD’s open‑source ROCm platform, which aims to challenge Nvidia’s CUDA dominance. During a media briefing Su announced a memorandum of understanding with the Korean science ministry to establish an AI Center of Excellence in the country, promising to hire hundreds of researchers. The visit underscores AMD’s strategy to secure critical components, broaden its AI hardware and software ecosystem, and strengthen its foothold in the fast‑growing AI market.

💡 Why It Matters

· By locking in HBM partners and cultivating a local AI talent pipeline, AMD positions itself to reduce reliance on external suppliers and accelerate its challenge to Nvidia’s AI‑accelerator lead.

Read full article on newsable.asianetnews.com

HPE Unveils The First Of The Next Generation of ProLiants for the…

Hewlett Packard Enterprise launched the first generation of ProLiant Gen 13 servers, designed for AI‑centric workloads. The new line features AMD’s 6th‑generation Epyc “Venice” CPUs, up to 256 cores per chip, and supports eight double‑wide GPUs in a 10U DL585a chassis. A 1U DL525 model offers a single 256‑core CPU for high‑throughput inference and analytics. HPE emphasizes enhanced security, performance, and power handling to meet the growing demands of enterprise AI and agentic workloads.

💡 The rollout signals a shift toward CPU‑heavy AI architectures, reflecting the industry’s move to balance GPU and CPU resources as agentic models expand. This focus on dense, secure compute could redefine how mid‑tier providers scale AI services.

digitimes.com · Oct 8

27 Samsung, SK Hynix veterans joined China’s CXMT, study shows

An investigation by the office of a South Korean lawmaker found that 27 former employees of Samsung Electronics and SK Hynix have joined Chinese DRAM producer ChangXin Memory Technologies (CXMT). The veterans, recruited over several years, held positions across research, design, process engineering and mass‑production, covering critical phases of semiconductor development. Their move reflects CXMT’s strategy to accelerate technology transfer and scale its manufacturing capabilities amid a global push for advanced memory chips. The report, based on employment records and interviews, indicates that the hires occurred despite heightened scrutiny of talent flows between South Korea and China. CXMT, a relatively new player, aims to compete with established memory makers by leveraging the expertise of these seasoned engineers.

💡 CXMT’s acquisition of senior South Korean talent fast‑tracks its bid to close the technology gap with industry leaders, reshaping the competitive landscape of global DRAM supply.

nextplatform.com · Oct 8

Memory Is The Lynchpin Of The IT Industry, And Doubly So For…

The IT industry has shifted from a compute‑centric focus—dominated by CPUs for five decades and GPUs for the last fifteen years—to a memory‑centric model driven by the rise of generative and agentic AI. Gartner’s latest chip revenue data shows memory (including DRAM, HBM and NAND flash) now outpacing non‑memory sales, with overall semiconductor revenue nearly doubling between 2025 and 2026 before a projected slowdown in 2027. Micron’s addressable market forecasts indicate that traditional DRAM categories will soon eclipse high‑bandwidth memory (HBM) sales. This surge is fueled by a massive upgrade cycle as aging server fleets are replaced, and by AI workloads demanding maximum core density, which pushes demand for high‑capacity DRAM and flash. Tight DRAM supplies, as manufacturers pivot to HBM for AI accelerators, have driven server‑grade DDR4 prices up 2‑3 × and DDR5 9‑13 × since 2022, while enterprise SSD costs have risen 6‑7 ×.

💡 AI’s appetite for memory is forcing a wholesale redesign of data‑center economics, turning memory scarcity into the primary cost driver for modern computing infrastructure.

teslarati.com · Oct 8

Elon Musk shuts down talk of TSMC taking over Terafab

Elon Musk clarified that Tesla and SpaceX will build and operate the Terafab chip factory in Texas, rejecting speculation that Taiwan Semiconductor Manufacturing Company (TSMC) would take control. Musk stated the companies would retain full ownership and operation, allowing only a possible sublease of part of the facility. The announcement follows earlier reports of TSMC exploring involvement and underscores Musk’s intent to secure in‑house chip production for future AI, robotics, and SpaceX data‑center needs. The project, valued at up to $119 billion, aims to deliver 1 terawatt of computing power annually.

💡 By keeping Terafab under its own management, Musk eliminates reliance on external foundries, ensuring priority access to critical chips for Tesla’s autonomous systems and SpaceX’s orbital infrastructure. This move could reshape the semiconductor supply chain, positioning Tesla and SpaceX as independent players in a market dominated by a few contract manufacturers.

taipeitimes.com · Oct 8

TSMC supplier Gudeng launches operations at Arizona plant

Gudeng Precision Industrial Co. inaugurated its Arizona plant on Tuesday, marking the company’s first production base in the United States and expanding its footprint beyond Taiwan, China, and South Korea. The ceremony, led by chairman Bill Chiu, highlighted Arizona’s emerging semiconductor‑aerospace cluster, bolstered by TSMC’s advanced wafer fab and planned packaging facilities. Gudeng will launch an initial parts‑processing line to serve U.S. customers and grow alongside its partners, including TSMC. The move follows Gudeng’s 2023 ground‑breaking of a plant in Japan’s Fukuoka Prefecture and reflects a broader trend of Taiwanese tech firms establishing operations in Arizona, such as United Integrated Services and Foxlink International. Chiu cited the state’s talent pool and long‑term government support for advanced manufacturing as key incentives.

💡 Gudeng’s Arizona launch deepens the U.S. semiconductor supply chain, tying local production directly to TSMC’s ecosystem and reducing reliance on overseas logistics.

semiconductor-digest.com · Oct 8

MintNeuro Raises $5M to Scale Purpose-Built Chip Platform for Neural Devices and…

MintNeuro, a UK‑based semiconductor firm, secured $5 million in funding to accelerate the commercial rollout of its purpose‑built neural chip platform. The round, led by Odyssey Ventures and joined by 33East, JSK Investments, Alumni Ventures, Empirical Ventures, Jumpspace Ventures and early backers, will expand engineering, grow international sales, and deliver pre‑release silicon to partners. The company has already supplied working chips to Motif Neurotech and Amber Therapeutics, and its modular architecture promises smaller, lower‑power neural devices for BCIs, neuromodulation, and sensory prostheses.

💡 The investment signals a tipping point for neurotechnology, enabling faster, safer device development and potentially shortening the path from research to clinical application. By providing a standardized, high‑performance chip stack, MintNeuro could reduce the time and cost that have historically stalled the adoption of brain‑computer interfaces and related therapies.