Recently, a research team led by Professor Ren Tianling from Tsinghua University’s School of Integrated Circuits published a groundbreaking study in the prestigious journal Nature, detailing a significant advancement in flexible electronics and edge AI chip technology. The team developed the FLEXI series of flexible digital in-memory computing chips, which leverage low-temperature polycrystalline silicon thin-film transistor technology. These chips offer a range of benefits, including ultra-thin form factors, flexibility, extremely low power consumption, high energy efficiency, and cost-effectiveness.

Through a comprehensive process-circuit-algorithm co-optimization strategy, the FLEXI chips ensure stable performance even under demanding conditions such as high-frequency computing, extreme mechanical stress, and accelerated aging. Notably, they have demonstrated long-term reliability, maintaining stability for over six months. The FLEXI series comprises three models: FLEXI-1 (1 kb), FLEXI-4 (4 kb), and FLEXI-32 (32 kb). Each chip integrates approximately 265,000 transistors, along with SRAM memory, computing units, and peripheral circuits, all on a single flexible substrate.

Experimental evaluations reveal that the FLEXI chips can operate reliably within a power supply voltage range of 2.5–5.5V. They withstand more than 40,000 bending cycles without significant performance loss. Additionally, the chips boast a single-unit cost of under $1 and achieve an overall production yield of 70%–92%. In real-world applications, the FLEXI chips have been successfully deployed for continuous monitoring and recognition of daily activities, highlighting their vast potential in wearable health monitoring and in-sensor computing for multimodal sensors.