In the face of overwhelming cloud-based AI hype, Arm has debuted its new Lumex chip platform, hoping to provide on-device AI-based processing to developers that do not, or cannot, tolerate the latency or cost of such a round-trip. Put simply, there are new CPU designs (C1-Ultra, C1-Premium, C1-Pro, and C1-Nano), a new GPU design (Mali G1-Ultra), and a raft of under-the-hood improvements that aim to boost AI performance. The most immediate of these is the introduction of SME2, the second generation of Arm’s Scalable Matrix Extension (SME), running in the new Armv9.3 architecture. Lumex is the mobile-focused Arm offering, unveiled earlier this year as part of a rebrand. Arm’s Neoverse targets the data center, while Zena is for automotive, Niva…