| Mobile Web

Search results for Ethernet

Industry

AMD launches exascale AI infrastructure Helios

AMD has unveiled its Helios rackscale solution, an AI infrastructure platform integrating Instinct MI455X GPUs, sixth-generation EPYC server CPUs, ROCm software and Pensando networking. The company said Helios delivers up to 2.9 exaflops peak FP4 per rack, 31 terabytes of HBM4 and 1.7 petabytes per second of memory bandwidth. AMD said it outperforms rival flagship systems in FP4 performance, HBM capacity and bandwidth, and scale-out bandwidth.

Industry

Electronica Shanghai 2026: Electrification and AI reshape electronic components landscape

At Electronica Shanghai 2026, exhibitors highlighted a shift in intelligent driving from maximising raw computing power to sustaining performance under limited power. Texas Instruments showcased scalable automotive SoCs and radar, while suppliers promoted sensors and high-voltage power components for faster charging and tougher thermal conditions. Connectors drew attention as higher voltages and growing in-vehicle data traffic raised demands for reliability. Discussions also focused on power delivery losses in data centres.

AI & Enterprise

OpenAI unveils in-house AI inference server chip to go live by year-end

OpenAI and Broadcom jointly developed an AI inference server chip called Jalapeno, OpenAI said on June 24. The move is part of OpenAI\'s strategy to reduce its reliance on Nvidia chips and secure its own hardware. OpenAI plans to bring its first Jalapeno servers online by year-end and then expand use. The chip is designed only for inference, and OpenAI said early tests showed higher performance per watt than current state-of-the-art products.