A breakthrough from researchers in China could finally solve one of the biggest problems holding smart wearables back: the chips inside them are typically rigid and energy-hungry, which limits comfort, battery life, and real-world health tracking. A team from Tsinghua University and Peking University has introduced a new class of ultra-thin flexible chips called FLEXI, designed to bring medical-grade monitoring and on-device AI to wearables at an extremely low cost.
FLEXI stands out for two reasons: how it computes data and how it physically moves. The research, published in Nature, describes a major shift away from the usual “send data to a processor (or the cloud) and wait for results” approach. Instead, FLEXI uses a compute-in-memory architecture, meaning it processes information directly where it’s stored. By cutting down the back-and-forth energy drain between memory and a separate processor, the team says the chip can run on less than 1% of the energy consumed by standard silicon designs used in many wearable devices today. For consumers, that could translate into longer-lasting batteries and more reliable continuous tracking without constant charging.
The hardware is also built for real life. FLEXI is made using low-temperature polycrystalline silicon (LTPS) on a flexible plastic substrate, creating a chip thinner than a human hair that can bend and twist without breaking. In durability testing, the chip reportedly survived more than 40,000 bending cycles and even handled being folded down to a one-millimeter radius with no performance loss. That level of resilience opens the door to electronics that don’t just sit on your wrist, but can conform to skin, clothing, and other surfaces that traditional chips can’t handle.
In practical tests with volunteers, FLEXI showed strong performance for health and activity monitoring—two of the most important use cases for next-generation wearables. The chip detected irregular heart rhythms (arrhythmia) with 99.2% accuracy and tracked common movements such as walking and cycling with 97.4% accuracy. Results like these suggest FLEXI could support more dependable heart monitoring and fitness tracking while staying compact, lightweight, and power-efficient.
Perhaps most intriguing is the push for affordability. The researchers are targeting a manufacturing cost of under $1 per unit, which could make it realistic to embed AI-enabled health monitoring into smart textiles and connected clothing—ideas that have often been limited by cost and power requirements. While the team plans to integrate more advanced sensors over time, FLEXI already signals a compelling direction for flexible electronics: durable, low-power, and capable of handling AI tasks directly on the device.
If this technology scales as planned, it could help turn the vision of comfortable, always-on health wearables—built into fabrics, patches, and everyday garments—into something far more common and accessible.
