Compute-in-memory returns to the spotlight with edge AI's power, memory squeeze

The rise of edge AI is reviving the compute-in-memory paradigm, fundamentally reshaping chip architecture and intensifying demands on DRAM, NAND, and advanced process nodes. Upstream memory vendors face capacity and tech pressure, while midstream designers must accelerate integration of compute-memory co-design. Geopolitical tensions, especially U.S. export controls on advanced processes and AI chips, are heightening supply chain risks, pushing companies toward localization strategies. In competitive dynamics, global players like NVIDIA and AMD are racing to dominate edge AI platforms, while local firms focus on low-power, high-integration solutions. Over the next 12–24 months, if edge AI adoption accelerates, memory computing will gain traction, creating new tech dependencies and forcing global semiconductor supply chains to restructure.