Indian Researchers Develop Molecular Memristor for Neuromorphic Computing

Researchers at the Centre for Nano Science and Engineering (CeNSE) at the Indian Institute of Science (IISc) in Bangalore, led by Professor Navakanta Bhat and Associate Professor Sreetosh Goswami, hav... Read original →

The IISc team’s ruthenium-based molecular memristor—delivering 14-bit analog precision and 4.1 TOPS/W—threatens to upend edge AI hardware by enabling deterministic, defect-free switching unattainable with conventional ReRAM. Its compatibility with 22nm CMOS undermines the necessity of migrating to 3nm for certain neuromorphic workloads, pressuring EDA vendors to accelerate analog-in-memory design flows. Geopolitically, an Indian startup commercializing this could sidestep U.S. AI chip export controls, though reliance on non-EUV nodes limits scaling. TSMC may benefit indirectly as demand revives for mature-node specialty platforms. Expect NVIDIA and Qualcomm to respond via strategic IP licensing or M&A to close their analog AI acceleration gap. Within 18 months, automotive-grade validation could capture >5% of industrial edge inference, catalyzing RISC-V + in-memory compute SoCs as the new architectural norm.