Nvidia M10 Material Upgrade Program for AI Server Systems
Nvidia has started its M10 material upgrade program to develop advanced AI server systems. The AI server supply chain experiences a major transformation because Nvidia and Shul Electronics initiated M10 Copper Clad Laminate testing. The testing program started in 1Q26 to establish PCB architectures needed for future Nvidia Kyber racks and the Rubin Ultra and Feynman platforms. The development introduces a new period for high performance computing materials through material development.
The M10 testing phase has reached its first quarter of 2026 while the team works on sampling and prototyping. The testing process needs to be finalized by 2Q26 because that is when the final testing results will be obtained. The M10 based CCL and PCB components will enter mass production during the second half of 2027 once the qualification process achieves performance requirements.
The M10 material uses quartz cloth as its current base material although manufacturers might switch to Low Dk 2 materials for future mass production runs to boost their production and commercialization capabilities. The upcoming generation of AI workloads requires these materials to maintain their signal integrity.
The M10 system supports the installation of orthogonal backplanes which combine midplane and orthogonal backplane functions. The design replaces the previous Cartridge architecture with a new system. The new material will be utilized in
- Switch Blade Motherboards Specifically designed for the Rubin Ultra and Feynman hardware platforms.
- Backplane Infrastructure The system provides high speed interconnects which support the Kyber rack architecture.
Nvidia is significantly expanding its CCL supply base with the M10 generation. The previous M9 material required Elite Advanced Materials certification while the current M10 testing program evaluates three dedicated suppliers.
- Elite Advanced Materials Maintaining its role as a key incumbent partner.
- New Chinese Manufacturer Broadening regional supply chain depth.
- Taiwan Based Manufacturer The firm supplies multiple operations which help global logistics and production networks.
Shul Electronics establishes itself as Nvidia's main provider of ultra low latency LPU and LPX racks through its M10 testing operations. The AI inference systems use 52 layer PCBs for their operation. The mass production of high layer count boards will start between 4Q26 and 1Q27. The company can produce at 52 layer density which gives them a leading position for architectural updates needed to support Nvidia's upcoming advanced AI server designs.
