onsemi Sees $6 Billion Physical AI Opportunity by 2030 - thelec.net

onsemi's $6B Physical AI thesis is not a product roadmap—it is a declaration that AI's value anchor is migrating from FLOPS to joules. When Qualcomm locks every Snapdragon 8 Elite Gen 6 die to TSMC in Taiwan, China, when Samsung embeds secure elements inside phone batteries, and when Lattice bakes post-quantum crypto into FPGA silicon, the industry's center of gravity has shifted. Compute is no longer the scarce resource; power density, thermal management, and physical-layer security in the sense-decide-actuate loop are the new bottlenecks. Solidigm weighing a US NAND fab and Chinese panel makers raising LCD prices on supply anxiety confirm that supply-chain regionalization has graduated from strategic option to a P&L line item. Qualcomm and Google co-designing a PC architecture around Agentic AI directly threatens Intel and AMD's x86 moat, while LG's deepening ties to Microsoft and NVIDIA are ecosystem land-grabs in AI terminals. The 18-month inflection: the real competitive moat is no longer the GPU. It is advanced packaging (Kools' bonding for Rubin Ultra), power-semiconductor topology, and the forced compliance wave around post-quantum security. Whoever productizes physical-layer security as a commodity component first will own the next pricing cycle.