AI’s hidden energy tax: More than 80% of scale-up traffic is just memory moving around

Posted

Memory bottlenecks are emerging as one of the costliest problems in AI infrastructure as inference overtakes training. GPUs sit idle waiting on data, while high-bandwidth memory, or HBM, stays bolted to processors that cannot share it. Photonics is shaping up as one way out. Marvell Technology Inc. has been pairing optics with photonic memory to […]

The post AI’s hidden energy tax: More than 80% of scale-up traffic is just memory moving around appeared first on SiliconANGLE.



Continue reading at SiliconANGLE »

AI, Cube Event Coverage, NEWS, #SC26, #theCube, accelerated computing, agentic AI, AI data centers, AI inference, AI infrastructure, Andy Mills, artificial intelligence, DDR5, disaggregated memory, energy efficiency, exascale computing, GPUs, HBM, High Performance Computing, high-performance storage, high-speed networking, HPC, HPC infrastructure, HPC Unites, John Furrier, KV Cache, Marvell, memory pooling, MemoryAI Photonic Appliance, Penguin Solutions, Photonic Fabric, Preet Virk, Research Computing, SC26 2026, SC26eventpage, SciNet, sustainable computing, NASDAQ:MRVL, NASDAQ:PENG