A brain implant is not one device. It is an electrode that touches neurons, a chip that reads and writes their signals, a wireless link that carries data out, and software that turns noise into intent. Most companies borrow pieces from somewhere else. One Beijing startup decided to build every layer itself — and that decision is why it keeps showing up whenever China talks about matching Neuralink on hardware.
A Tsinghua lab becomes a company
NeuraMatrix (宁矩科技) was founded in November 2019, incubated from research at Tsinghua University (清华大学). Its founding team had been working on brain-machine interfaces since around 2010, partly at the University of Pennsylvania and the Max Planck Institute before returning to China. The company positions itself as a platform play: instead of one product, it sells the full stack of invasive brain-computer interface (脑机接口) — electrode materials, a dedicated neural-interface chip, the host system, and the decoding software.
The chip is the headline
The piece that gets attention is the company's self-designed system-on-chip for neural interfacing. NeuraMatrix says the chip supports up to about 1,000 channels of signal access, with wireless data bandwidth around 100 Mbps and compute above 1 GOP/s — figures the company cites to argue its chip rivals or beats comparable implant electronics on noise resistance and wireless transmission. It describes the part as China's first commercial bidirectional brain-computer interface (脑机接口) chip, handling both recording and stimulation.
Those specifications are company disclosures, not independently audited benchmarks, and exact channel counts have shifted across the company's own product descriptions over time. They are best read as evidence of a vertical-integration strategy rather than a head-to-head victory.
Three product lines, three markets
NeuraMatrix organises its roadmap into three tracks:
- NeXScope — research and medical-grade tools for labs and hospitals, the beachhead where invasive recording is already useful for neuroscience.
- NeXPrime — medical-grade systems aimed at clinical use, the harder and slower regulatory path.
- NeXLive — consumer-health applications, the largest but least proven market.
The logic is familiar from the broader implant field: prove the hardware in research, migrate it to therapy, then chase consumer scale. Each step raises the regulatory and trust bar.
Money followed the hardware bet
In March 2021, NeuraMatrix closed a Pre-A round of several million US dollars led by Matrix Partners China (经纬中国). In November 2021 it raised a renminbi 100 million (≈ US$14M / HK$110M) A round led by HG Capital (华盖资本) and Prosperity7 Ventures — the diversified venture fund backed by Saudi Aramco — with Matrix Partners China following on. The financing, reported by state-media outlet Guangming Online (光明网) and trade press, was earmarked for chip development, mass production, and software. At the time the company said it had already taken orders worth tens of millions of renminbi from research institutes and hospitals.
That capital story matters because invasive brain-computer interface (脑机接口) is brutally hardware-intensive: every channel you add needs cleaner silicon, better packaging, and a wireless link that does not cook tissue. A funding round tied to chip mass production is a signal that the bottleneck the field complains about — reliable, scalable, domestic implant electronics — is being attacked directly.
Why "full-stack" is the real moat
Implant makers that rely on off-the-shelf components hit a wall: consumer-grade chips are not built for the noise, power, and size limits of something sitting inside a skull. Owning the chip lets a company tune power, bandwidth, and electrode count together. NeuraMatrix's bet is that China's implant future needs domestic silicon as much as it needs clever algorithms.
The company it keeps in China's implant race
NeuraMatrix is one name in a wider domestic push into invasive brain-computer interface (脑机接口). What separates the players is how much of the stack they actually own. Some focus on the electrode — the interface that touches living tissue. Some focus on decoding software, which is where the flashy demos come from. NeuraMatrix's chosen identity is the unglamorous middle: the chip and the host system that make high channel counts usable outside a bench. That is a slower, capital-heavier road, but it is also the road that determines whether a country can build implants without leaning on foreign silicon. For readers, the tell when judging any implant startup is simple — ask what it makes versus what it buys.
Honest limitations
NeuraMatrix's channel, bandwidth, and compute figures are company statements, not results from a published independent teardown. The "first commercial bidirectional BCI chip in China" claim is the company's own framing. The consumer (NeXLive) line is described but not yet a clinically validated product, and invasive medical approval in China remains a long path that the company has not publicly completed. Funding amounts and investor names are confirmed by Guangming Online (光明网) and trade coverage, but the company's current clinical-trial status is not detailed in the public sources reviewed here.
What readers can do now
- Labs and hospitals shopping for invasive recording gear should ask whether the vendor controls its own chip and electrode supply — that determines long-term serviceability and upgrade cadence.
- Founders in neurotech (神经科技) should note the lesson: in implants, owning the silicon is often what separates a demo from a product.
- Readers tracking the field should watch NMPA (国家药监局) filings and clinical-trial registrations, not funding headlines, as the real proof that an implant platform has graduated from lab to clinic.