🎯 Today's Semiconductors Pulse
NVIDIA teams with SK Telecom on gigawatt‑scale AI cloud in South Korea
NVIDIA and SK Telecom announced a partnership to build a gigawatt‑scale AI cloud, with the first AI factory on NVIDIA’s DSX platform slated for 2027. The initiative will support sovereign AI, robotics, semiconductor and manufacturing workloads, expanding SK Telecom’s role beyond pure connectivity.
Also developing:
🚀 Top Semiconductors Headlines
NXP Computex Keynote 2026 Coverage
The final keynote for Computex 2026 comes from NXP, where CEO Rafael Sotomayor is talking all about what it takes to deliver AI for edge devices and robotics in the real world, and how NXP is well-positioned to accomplish this The post NXP Computex Keynote 2026 Coverage appeared first on ServeTheHome.
ServeTheHome
'Flawless on the Outside, Flipped Within': Detecting Hidden Defects in 2D Dielectrics with Light
A material may appear flawless on the surface yet fail to function properly. The cause lies in structural defects hidden within two-dimensional thin films, which are considered key materials for next-generation semiconductor devices. Recently, a Korean research team developed an optical analysis method that can identify these invisible defects using light.
Phys.org – Nanotechnology
Nvidia Partners with SK Hynix to Supply Memory for Vera AI Chips, Boosting Telecom AI Infrastructure
Nvidia announced a multiyear partnership with South Korean memory maker SK Hynix to co‑develop AI‑optimized DRAM for its new Vera AI processors. The deal, revealed on June 8, coincided with a 7.7% drop in SK Hynix shares and a modest 1.4% rise in Nvidia stock, underscoring the market’s focus on supply‑chain alignment for telecom AI workloads.
Pulse
Research Bits: June 8
Multi-tasking transistor; p-bit fabrication; tiny MoS2 nanotubes. The post Research Bits: June 8 appeared first on Semiconductor Engineering.
Semiconductor Engineering
Innodisk Showcases Full Qualcomm Dragonwing Edge AI Lineup at Computex 2026, Spanning Dragonwing IQ8 Through IQ10
TAIPEI, June 8, 2026 /PRNewswire/ -- Innodisk, a leading global AI solution provider, showcases the full breadth of its collaboration with Qualcomm Technologies, Inc. at Computex 2026. Building on the "AI on Dragonwing" computing series launched earlier this year, Innodisk has rapidly expanded the edge platform lineup to span four generations of Qualcomm Dragonwing™ processors, delivering scalable performance from 20 to 700 TOPS across a broad spectrum of edge AI workloads. Nakul Duggal, Executive Vice President at Qualcomm, greets Randy Chien, Chairman of Innodisk at Innodisk booth during Computex 2026 Architecture Meets Industrial Edge Demands The shift toward edge AI in factory automation, smart cities, and defect inspection heightens the demand for power-efficient inference. For battery-powered AMRs and AGVs, minimizing power consumption is critical to sustaining operational endurance and lowering thermal costs-positioning this high-performance, low-power architecture as a compelling choice for next-generation deployments. With Dragonwing, Qualcomm Technologies aims to deliver industry-leading power efficiency and multimodal compute to the industrial segment. Spanning the Qualcomm Dragonwing™ IQ8, IQ9, IQ10, and IQ-X Series with product longevity support through 2036 or later, this scalable processor portfolio helps enable enterprises to right-size compute infrastructure. Capitalizing on this, Innodisk unleashes this potential by providing a "AI on Dragonwing" hardware lineup of modules, starter kits, and deployment-ready systems, paired with its IQ Studio software toolkit streamlines evaluation-to-deployment with BSP, benchmarking, and model optimization. "Dragonwing processors aim to deliver industry-leading performance-per-watt with long-term supply commitment, helping to establish a solid foundation for physical AI deployment at scale," said Shyam Krishnamurthy, SVP, Product Management, Qualcomm Technologies, Inc. "Innodisk's industrial edge integration expertise makes them a critical partner in bringing the AI on Dragonwing ecosystem to market." Scalable Compute Across Four Performance Profiles The Computex lineup covers four Dragonwing™ platforms spanning distinct application profiles. The Dragonwing IQ8 Series powers the EXMP-Q801 COM-HPC Mini module, designed to deliver up to 40 TOPS (Dense) with wide-temperature operation for vision inference and sensor fusion. The Dragonwing IQ9 Series is designed to scale up to 100 TOPS (Dense) via dual NPU and powers the EXMP-Q911 module, EXEC-Q911 starter kit, and the APEX-A100 fanless edge AI system, helping to address industrial inspection, Driver Monitoring Systems (DMS), and multi-stream analytics. The Dragonwing IQ10 Series, powering the EXMR-QA01, aims to extend the portfolio to a massive 350 TOPS (Dense). Featuring an 18-core Oryon™ CPU, support for up to 20 MIPI cameras at 16 MP, and an integrated Safety Island (SAIL) module, the platform is custom-engineered to help support heavy-duty AMR navigation, humanoid robotics, and VLM and VLA edge workloads. The Dragonwing IQ-X Series stands out with native Windows compatibility. The EXDU-QX11 edge AI system, powered by the Dragonwing IQ-X processor, is designed to deliver up to 45 TOPS with native Windows compatibility. Offering dual-OS support for Windows 11 IoT Enterprise LTSC and Ubuntu 26 LTS, it is designed to streamline deployment in PLCs, HMIs, and edge controllers equipped with features such as Ethernet, MIPI CSI, and CAN. Live Demonstrations at Computex 2026 At the Innodisk booth, two live demonstrations validate deployment readiness. The first, showcasing the EXMP-Q911 module powered by the Dragonwing IQ9 processor, executes a concurrent pipeline of CNN defect inspection, DMS, and VLM-driven PPE detection within a strict 30W total system power. The system achieves an ultra-low 4.4 ms average inference latency, delivering a 2.6× speed advantage over traditional architectures while optimizing performance-per-watt. The second demo showcases vision-powered autonomous robotic navigation on the EXFC-Q911 fanless system. Engineered for AMRs and AGVs, the system manages a high-throughput 8-channel GMSL2 camera pipeline and 3D depth sensors, delivering 360-degree spatial awareness and real-time obstacle avoidance with millisecond-level responsiveness at only 32W.
The Manila Times Business
💬 Top Semiconductors Social Posts
Thread by @Bdinvestingg
Deepdive on $CRDO and why this is an opportunity- Credo makes chips that move data between servers inside data centers. As AI models get bigger, they need faster, lower-power links — CRDO builds exactly that. Revenue is expected to grow by 4x by 2029 https://open.substack.com/pub/bdinvesting/p/crdo-solving-the-connectivity-bottleneck?r=1fden7&utm_medium=ios&shareImageVariant=split
Tweet by @BenBajarin
On @intel and glass substrates, this article from @stshank from 2023 was from when a small group of us got to tour their advanced packaging rnd center and saw the glass substrate work in action. @stshank got some of the only pics allowed inside :) https://t.co/D8ks3WgTg8