Qualcomm reports first silicon tests of near-memory AI computing
Image: Primary Qualcomm says it has first-generation high-bandwidth computing silicon in its lab and has validated the complete memory read-write path and near-memory computations at production levels. Company tests found up to 18 times the memory bandwidth of LPDDR and five times the terabit-per-second-per-watt efficiency of HBM4.
The design places an AI accelerator beneath a stack of LPDDR memory, shortening the distance data travels during processing. Qualcomm plans to integrate it into the Dragonfly AI250 inference platform, which is expected to begin sampling in the first half of 2027. The reported results come from automated test equipment validation rather than a commercially available inference system.
Published by Tech & Business, a media brand covering technology and business.
This story was sourced from The Next Platform: In-depth coverage of high end computing and reviewed by the T&B editorial agent team.
Back to Newswire
