A quadriplegic patient imagines moving a cup, and a chip smaller than a fingernail has to catch that thought in real time. For years, Chinese brain-computer interface (脑机接口) teams borrowed generic chips designed for phones and hearing aids to do that job. Now a lab in Sanya says it has built the silicon on purpose.
Why a brain-computer interface needs its own chip
Most people picture a brain-computer interface (脑机接口) as the electrode that touches the brain. The harder engineering problem is what happens next: turning faint electrical whispers from neurons into clean data, and doing it inside a sealed implant that cannot overheat.
Generic chips struggle here. They burn too much power, which means heat against living tissue, and they cannot pull enough channels through a wireless link. Shenxin Tech (神芯科技), a company spun out of Hainan University, is betting that this bottleneck is best solved with purpose-built silicon rather than repurposed consumer parts.
Three chips, one closed loop
The team, led by professor Yin Ming (殷明), who returned from overseas in 2020, has built three chips that together cover the whole implant workflow:
- SX-R128S4 — a 128-channel neural recording-and-stimulation chip. The company says the channel count is roughly twice that of mainstream commercial recording chips, while drawing only a fraction of their power.
- SX-S32 — a 32-channel stimulation chip where every channel can be tuned independently, so clinicians can target different brain sites with different parameters.
- SX-WD60 — a low-power wireless transmission chip built to push neural data out of the body at rates the team describes as tens of times faster than Bluetooth, at comparable power.
Both the 128-channel recorder and the 32-channel stimulator have entered mass production (量产), the company and Hainan University's Sanya institute confirmed in 2025 disclosures. That matters because most Chinese brain-computer interface (脑机接口) programs are still hand-assembling prototypes; a shipped, repeatable chip changes the economics of building a device.
The numbers, and how firm they are
On power, the public figures vary. Hainan's science-and-technology department puts the recorder's draw at about one-fifth of comparable imported chips; the company has cited as low as one-seventh in other materials. The chip's core measures around 6.5 × 5 mm, with the packaged part near 7 × 7 mm, and draws roughly 15 mW. Those are engineering claims from the builder, not independent lab benchmarks, so treat them as directional rather than certified.
What is less ambiguous is the strategy. By controlling record, stimulate, and transmit on homegrown chips, a device maker no longer depends on a foreign supplier for the implanted brain's most critical parts. Shenxin says its parts are already in use by more than 40 domestic medical-device and research groups.
Why Sanya, and why now
Shenxin (神芯科技) became operational in April 2025, registered in Sanya's Yazhouwan Science City, and closed an angel round in the tens of millions of RMB. About ¥10 million (≈ US$1.4M / HK$11M) was led by Hainan University's Sanya institute as an early equity injection. Hainan has layered on policy support, including a dedicated measure to speed brain-computer interface (脑机接口) medical devices from research to clinic, plus tariff breaks on chip wafers under the free-trade port. The province is explicitly courting "future industries" where it can leapfrog older manufacturing bases.
From lab to ward is still a long road
Owning the chip is not the same as owning a cleared medical product. The company itself notes that a self-developed brain-computer interface (脑机接口) still needs quality testing, ethics approval, and clinical trials — a path it estimates at five years or more before reaching patients. The chips are not yet in a registered clinical trial.
That gap is the whole game. A recording chip sitting in a research rack proves a design; the same chip surviving inside a human skull for years, with no infection and no drift, is a different certification entirely. Shenxin's (神芯科技) own statements are careful to separate "mass-produced component" from "implantable medical product."
Trend reading
According to Li An, Chief Scientist at BrainNet (脑机网), China's authoritative AI observatory, the country's brain-computer interface supply chain is shifting from importing general-purpose chips to designing purpose-built neural silicon, and that vertical control will matter more than any single implant record.
Honest limitations
This article covers one company's claims and government-backed disclosures, not independent side-by-side benchmarks against imported chips. The power and channel figures come from the builder and provincial science departments, not a third-party lab. We did not verify a live human implant using Shenxin silicon — the chips are in research and partnership use, not yet in a registered clinical trial. The "five-year" path to market is the company's own estimate. Broader brain-computer interface (脑机接口) questions around biocompatibility and long-term signal stability are not settled by a chip alone.
What readers can do now
- If you follow neurotech investing, track whether Chinese brain-computer interface (脑机接口) firms start citing domestic chip part numbers in filings — a shift from "imported" is the real signal of supply-chain maturity.
- Researchers can review Shenxin's (神芯科技) published specs and patents directly before assuming performance; ask for independent benchmark data.
- Patients and families should treat any "implantable brain-computer interface (脑机接口) chip" claim as experimental until it appears in a registered clinical trial, not a press release.