DEEPX plans to begin mass production of its DX-M2 AI processor in the fourth quarter of 2027 after completing tape-out, the final design stage before a chip is sent to manufacturing. Built using Samsung Foundry’s 2nm Gate-All-Around process, the chip is aimed at robots and industrial equipment that need to run AI workloads locally.
The company estimates power consumption at about 5W. DX-M2 is designed for on-device execution of large language models and vision-language models, allowing systems to analyze data and make certain decisions without a constant connection to cloud servers or the internet.
Local processing and low power draw could be particularly significant for autonomous robots and industrial deployments, where latency, connectivity and energy use can constrain AI hardware. DX-M2 has received a CES 2027 Innovation Award in the artificial intelligence category.
DEEPX expects to expand DX-M2 production in 2028. For Samsung, the design win adds another customer project for its 2nm manufacturing technology, though large-scale production performance and the chip’s commercial uptake remain to be seen.
