In March 2026, NVIDIA initiated supplier qualification testing for M10, a next-generation copper-clad laminate (CCL) material designed for its Rubin Ultra and Feynman AI platforms—marking a pivotal shift toward ultra-high-speed, low-loss PCB technology.
Compared to the current M9 material, M10 reduces signal loss by 30%–40% versus traditional FR-4, enabling stable transmission at speeds beyond 1.6 Tbps while improving thermal resistance for high-power AI accelerators. The new material will be applied to orthogonal backplanes and switch blade motherboards, critical components in NVIDIA’s next-gen AI server architecture.
Mass production of M10-based PCBs is targeted for Q2 2027, with multiple suppliers qualified to mitigate single-source risks. Industry experts note that M10 will become the new performance benchmark for high-end PCBs, driving a new round of material and equipment upgrades across the supply chain.