Own ON Semi For AI Power? NVIDIA Is Growing Faster And Costs Less. - Trefis

This article analyzes the investment potential of NVIDIA and ON Semiconductor in the context of the AI boom. While ON Semiconductor has seen a significant stock surge (up 122% over the past three mont... Read original →

NVIDIA’s AI chip dominance is triggering a cascade effect across the tech stack: its Grace Hopper platform demands upgrades in power delivery, high-speed interconnects, and thermal solutions—favoring SiC/GaN suppliers, while ON Semiconductor remains sidelined from AI server supply chains despite its automotive SiC push. Tightening U.S. export controls on advanced computing raise compliance costs; ON’s rushed capacity shifts to Mexico and Czechia inflate capex, whereas NVIDIA’s tight integration with TSMC’s CoWoS packaging and CUDA ecosystem offers superior supply-chain resilience. With Blackwell commanding over 95% of the training market, rivals like AMD and Intel may pivot to open-architecture edge AI, indirectly pressuring ON’s industrial IoT margins. Over the next 18 months, as AI investment shifts from training to inference, NVIDIA’s full-stack advantage will sustain its premium, but ON’s valuation hinges on achieving >40% gross margins in auto/industrial segments by 2027—or face a sharp correction.