Semtech and onsemi Shares Are Soaring, What You Need To Know - The Globe and Mail

At the ISC High Performance 2026 conference, NVIDIA unveiled its next-generation Vera Rubin platform, delivering over 7 exaflops of AI performance and up to 144 GPUs per rack, signaling a major advanc... Read original →

NVIDIA’s Vera Rubin platform isn’t just a performance leap—it’s a technical fulcrum resetting AI infrastructure economics. Its 144-GPU-per-rack design forces upgrades in power delivery, optical interconnects, and thermal management, directly benefiting Semtech (signal integrity ICs) and onsemi (high-efficiency power modules). Amid tightening U.S. export controls targeting AI chips to China, supply chain redundancy costs are rising—but hardware vendors with irreplaceable IP now command pricing power. Intel is accelerating Gaudi 4 deployments, while AMD may counter with MI400 integration into HPE/Dell systems to capture FP64 workloads. Over the next 18 months, capex will shift toward physical-layer components; memory bandwidth constraints will drive HBM3E/HBM4 adoption, creating structural order shifts for Micron and SanDisk. The market has drawn a clear line: infrastructure enablers are being re-rated, while model-layer players face regulatory overhang.