# China’s Flexible Brain Chip Holds 94% Signal After 18 Months > Source: > Published: 2026-06-19 17:54:36+00:00 Most brain implants work brilliantly on day one. By year two, many are going quiet. The culprit isn’t bad engineering — it’s biology fighting back against foreign hardware. A team from [Tsinghua University](https://interestingengineering.com/science/chinas-new-flexible-brain-computer-chip-retains-94-efficiency-after-18-months), the Chinese Academy of Sciences, and the University of Tokyo may have cracked this problem with an all-organic, hair-thin electrode array that reportedly kept firing clearly for over 18 months in animal tests. The findings, published in **PNAS**, suggest a fundamentally different approach to building brain-computer interfaces that last. Research into [frozen brain tissue](https://www.gadgetreview.com/german-scientists-revive-frozen-brain-tissue-after-week-long-deep-freeze) preservation represents another frontier in the broader effort to understand and protect neural function. ## The Problem With Putting Metal in Your Brain *Rigid electrodes trigger scar tissue that slowly chokes the signal — and no amount of software can fix a physics problem.* Conventional BCIs use stiff silicon or metal electrodes. [Brain tissue](https://www.gadgetreview.com/lab-grown-brain-tissue-just-learned-to-solve-classic-ai-problems) is roughly the consistency of soft tofu. That mechanical mismatch creates friction, inflammation, and glial scarring — the brain essentially walling off the intruder. **Utah arrays**, the industry workhorse for decades, often lose meaningful performance within one to three years. The new material, called **CHIP** (conductive hydrogel with interfacial percolation), ditches metal entirely. Here’s what the PNAS paper reports: **Conductivity:**~2,512 S/cm — exceptional for a soft hydrogel** Thickness:**~9 micrometers, thinner than a human hair** Channels:**128-channel array with 10× the density of previous hydrogel implants** Durability:**1,000 mechanical stretch cycles at 30% strain with less than 4% conductivity change The manufacturing trick that makes this possible: researchers pre-anchor the hydrogel to an ultrathin parylene substrate, then pattern electrodes via photolithography while the material is dry. Think printing fine text on crisp paper versus soggy cardboard. Once re-swelled in body fluid, the geometry holds precisely — solving the shape-instability problem that made earlier hydrogels unsuitable for high-density electrode arrays. ## 18 Months, 94% — What the Rabbit Data Actually Shows *The in vivo results are striking, though the gap between rabbit cortex and human cortex deserves honest scrutiny.* Arrays placed on rabbit cortex recorded electrocorticography signals for over **550 days** during free movement. Signal-to-noise ratio held at approximately **94%** of its day-one value at the end of the trial. Histology showed minimal inflammation and little glial scarring. The [PNAS authors describe](https://www.pnas.org/doi/10.1073/pnas.2532840123) the platform as enabling “all-organic, ultraflexible, and chronic neural interfaces” advancing “bioelectronic medicine and next-generation BCIs.” Impressive — but context matters. Rabbit cortex differs substantially from human cortex in size, motion dynamics, and required implant lifespan. [Neuralink](https://www.gadgetreview.com/elon-musks-sperm-donation-to-neuralink-executive-sparks-ethics-debate)‘s fine metal-on-polymer threads take a completely different architectural approach. Groups at MIT are 3D-printing soft polymer electrodes but haven’t matched CHIP’s conductivity figures. No human trials exist for this material. This is pre-clinical platform technology, full stop. The path forward runs through non-human primate studies, toxicology testing, and regulatory review — years of work. Innovations like the [robotic knee exoskeleton](https://www.gadgetreview.com/university-of-michigans-robotic-knee-exoskeleton-a-win-for-workplace-safety) from the University of Michigan illustrate how bioengineering breakthroughs navigate that same journey from lab to real-world human use. [China already approved](https://hellochinatech.com/p/neuracle-brain-implant-ipo) Neuracle’s NEO epidural implant using conventional flexible electrodes, so the regulatory infrastructure exists. The real significance here isn’t a product timeline. It’s that researchers may have finally broken the long-standing trade-off between softness and conductivity that has constrained [brain implants](https://www.gadgetreview.com/paralyzed-punk-rocker-creates-music-directly-from-his-brain-signals) for decades. If this material scales and its stability holds up in human-grade systems, the era of brain chips that degrade on a two-year cycle could eventually become a problem of the past.