Imagine recognizing a letter nobody has told you the shape of. Not reading a sentence — just telling an H from an R by a pattern of light your own eye no longer sends. For one patient in China, that happened a day after a switch was turned on; two others progressed more slowly.
The device doing it sits inside the eye. The company behind it just raised enough money to try it on more people.
On November 3, Hangzhou-based Nanochap (暖芯迦) announced a RMB 500 million (≈ US$75M / HK$585M) Series C+ round led by Gaorong Ventures. It was the company's second raise this year, bringing cumulative funding to RMB 1 billion (≈ US$150M / HK$1.17B). The money goes toward clinical trials and manufacturing for a visual brain-computer interface (脑机接口) that aims to give back a version of sight.
Key takeaways
- Nanochap's retinal system, called E-BCI, uses a 320-channel electrode array to stimulate the retina; a separate visual-cortex system, V-BCI, with 1,280 channels, is still in development.
- The company says three blind patients have been implanted, and that the first could identify complex letters the day after activation. These are company-reported outcomes, not published peer-reviewed results.
- The round was led by Gaorong Ventures, with insurers and state-linked industrial funds joining — a sign that vision restoration is being treated as a serious medical market, not a science project.
- Nanochap is one of several Chinese groups now competing with US efforts such as Neuralink's Blindsight and Science Corporation's PRIMA. Most visual-interface companies pick one route; Nanochap is pursuing both.
What the implant actually does
The system splits into two pieces. A small camera mounted on a pair of glasses captures the world. Software converts the image into a pattern of electrical pulses and sends it wirelessly to an implanted array, which stimulates the retina. The retina, still connected to the optic nerve, carries the signal to the brain. What the patient perceives is not normal vision: it is points and shades of light arranged into recognizable shapes.
That is why the 320 channels matter. Each channel is an independent point of stimulation, so more channels mean a finer picture. Think of it as the difference between a coarse LED grid and a high-resolution screen. Nanochap's longer-term device, the 1,280-channel V-BCI, skips the retina entirely and stimulates the visual cortex directly — a route that could help patients whose optic nerves are damaged, but one that carries more surgical risk.
The company has been building the underlying pieces since 2014: stimulation chips, high-density electrodes, hermetic packaging and encoding algorithms. That full-stack approach is unusual. Most teams buy their chips and focus on one part of the chain. Nanochap also sells neural-interface and biosensing chips to other customers, a business that reportedly generates millions of units in annual shipments and helps fund the longer visual work.
The three patients — and what "first" means
The clinical story is told in three cases, all described by the company. In May 2026, a woman blind for nearly twenty years received the 320-channel implant at a hospital in Hangzhou. According to Nanochap, she identified the letters H and R the day after activation and, over the following two months, learned to write the characters for "China," grasp a cup and walk along a line on the floor while turning corners.
A second patient, a 60-year-old man blind for four years, was implanted in Wuhan in a semi-invasive procedure that does not penetrate the eyeball. A third, a 66-year-old woman blind for eight years, was operated on in October using a subchoroidal approach. Both reportedly progressed from recognizing simple shapes to writing characters and walking without assistance.
These are striking claims, and they need careful framing. The company describes the first as China's first high-resolution retinal-path visual brain-interface trial conducted under GCP (Good Clinical Practice) standards, and the second as a world first for a semi-invasive retinal implant. "First" here is scoped to a particular technique and a particular claimant. Other teams may define the category differently, and I have not seen the trial registry, the ethics approvals or any patient-level data.
The honest summary is narrower: a small number of patients have shown early signals that the approach may work, and the company is now raising money to test it in more of them. That is real progress. It is not yet proof that the system works reliably across many patients or over many years.
Why the money is arriving now
Vision restoration has become a quiet international race. Neuralink's Blindsight has received US FDA breakthrough-device designation and targets the visual cortex. Science Corporation's PRIMA, a retinal implant, reported results in the New England Journal of Medicine and has been reported to have won European regulatory approval. The prize is large: the company and its investors cite roughly 2.2 billion people worldwide with some form of vision impairment, including tens of millions who are fully blind.
In China, brain-computer interfaces were written into the government work report this year and folded into the country's future-industry planning. That policy signal, combined with early human data, is what convinces insurers and state-backed funds to write checks. The lead investor, Gaorong Ventures, argues that high-throughput links between machines and the human visual system are the key breakthrough — a reasonable thesis, and one that will now be tested against clinical reality.
Honest limitations
This article relies on Nanochap's own announcement, a post from lead investor Gaorong Ventures, and reporting by STAR Market Daily and an English-language finance outlet that summarized it. All describe the same November 3 financing. The clinical outcomes are company-reported; I have not seen a peer-reviewed paper, a registry entry or independent examination of any patient. The phrase "first" is the company's, scoped to its own definitions.
I have also not verified the registered names of the hospitals, the trial design, the sample-size plans or the safety record beyond what the company chose to disclose. The comparison with US programmes is directional, not head-to-head: different devices, different patient groups and different endpoints. Currency conversions use a November 2026 reference rate and will drift.
What readers can do now
- If you follow medical technology, separate "implanted" from "approved." A device that has been placed in three patients is at a very different stage from one that is sold to hospitals. Ask which stage a headline is describing.
- If you are a patient or caregiver, read the endpoint, not the adjective. "Could write characters" and "restored vision" are not the same claim. Ask how many patients, how long they were followed and what happened when the device was switched off.
- If you are watching the sector, track the channels. Channel count is the crude but useful measure of how much detail an implant can deliver. It is the number to watch as these systems mature.
Three patients learning to read a letter is a genuine milestone. The more important question is whether the next thirty do as well — and whether the answer is published, not just announced.
