image credit- shutterstock
On March 13, 2026, China’s National Medical Products Administration (NMPA) granted Class III registration to NEO, an implantable hand motor function compensation system made by Shanghai-based Neuracle Technology with a team at Tsinghua University led by Professor Hong Bo. It was widely reported as the first invasive brain-computer interface (BCI) cleared for routine commercial medical use anywhere in the world.
The approval made the headlines. What happened next is the more instructive part. Within two days, the National Healthcare Security Administration (NHSA), the agency that runs China’s basic medical insurance system, had engaged the company. By March 22, the agency had published a national medical insurance consumable code for the device. On March 23, Shanghai’s municipal insurance authority added it to its consumables catalogue. On March 24, NEO completed procurement listing, which is the step that allows a public hospital to actually buy it.
Eleven days, from a regulator saying a product is safe enough to sell to a hospital being able to place an order. Industry observers in China took to calling the insurance step “light-speed coding”. For comparison, a novel medical device in the United States typically waits around five years for a national Medicare coverage determination after the FDA has already cleared it.
It would be easy to read this as a story about China pulling ahead of Neuralink. That reading is mostly wrong, and this article will explain why. The more durable reading is about governance. China did not simply approve a brain implant quickly. It pre-built the pricing categories a year before any product existed to fill them, published a technical vocabulary for the field, drafted classification rules, opened a second regulatory track for early clinical research, and has now issued a data standard for the artificial intelligence that decodes brain signals. Each piece arrived in sequence, and each one removed a reason for the next institution in the chain to wait.
What was actually approved
Precision matters here, because the gap between what NEO is and what it has been called is where most of the confusion starts.
NEO is a coin-sized titanium implant placed epidurally, meaning under the skull but above the dura mater, the membrane covering the brain. It does not penetrate brain tissue. It records signals from the motor cortex, digitises them, and transmits them wirelessly through the skin, with power delivered the same way. Those decoded signals drive a pneumatic glove that closes a patient’s hand. The approved indication is narrow: adults aged 18 to 60 with C2 to C6 cervical spinal cord injury who retain some upper arm movement but cannot grasp.
The evidence package supporting the approval comprised 36 implant procedures, four feasibility cases and 32 in a multicentre confirmatory study, followed for 18 months. According to the trial report, every patient achieved home-based, brain-controlled grasp assistance, and no serious adverse events were attributed to the device.
That is a real clinical result. It is also a modest one by the standards of the field’s frontier. American programmes led by Neuralink, Paradromics and the BrainGate academic consortium use intracortical electrodes that record from individual neurons, and have demonstrated cursor control, typing and speech synthesis in patients with complete paralysis. An epidural system trades signal fidelity for surgical safety. That is a defensible engineering choice, particularly for a first commercial product, but it means NEO restores a grasp rather than a hand.
Nor was NEO the first BCI-class product China cleared. A landscape analysis of the Chinese BCI clinical pipeline counted five NMPA-approved BCI products as of June 2026, including a non-invasive EEG rehabilitation system for stroke and a closed-loop deep brain stimulator for Parkinson’s disease with real-time sensing, alongside 134 registered clinical trials. NEO is the first implanted, decode-and-act device in that list. The same analysis describes a three-tier translation pipeline running from investigator-initiated trials through registered trials to approval, which is precisely the structure the new regulations are designed to accelerate.
The groundwork was laid a year early
The eleven days only make sense against what came before them.
On March 12, 2025, almost exactly a year before NEO’s approval, the NHSA issued its pricing guideline for neurological medical services and, for the first time, listed brain-computer interfaces as an independent service category. The guideline created separate billing items for implanting an invasive BCI, for removing one, and for fitting a non-invasive system. There was, at that point, no commercially approved invasive BCI in China to bill for.
Provinces followed. Hubei published what was described as the country’s first BCI pricing standard. Zhejiang and Jiangsu set their own prices. Beijing classified invasive BCI procedures as Class A insurance items at four designated hospitals. Reporting by Caixin put the government-guided price for an invasive BCI insertion procedure at 6,000 to 7,000 yuan, and noted that the NHSA has now issued 39 batches of pricing guidance covering around 180 new technologies, with a “pre-listing” mechanism announced in April 2026 that offers pricing guidance while products are still in clinical trials.
Strategy documents set the direction. The 14th Five-Year Plan (2021 to 2025) listed brain science and brain-inspired intelligence as a national frontier priority. Beijing and Shanghai each published BCI action plans for 2025 to 2030 in January 2025, Beijing targeting three to five unicorn or near-unicorn companies by 2027. In August 2025, seven central departments led by the Ministry of Industry and Information Technology jointly issued Implementation Opinions targeting key technology breakthroughs by 2027 and two to three globally influential BCI companies by 2030. The 15th Five-Year Plan, published in March 2026, named BCI among the country’s future industries alongside quantum technology and humanoid robotics. Sichuan has set itself a target of 3,000 invasive BCI procedures a year by 2030.
Eight layers, one direction
Pull the instruments apart and a pattern emerges. China has built something close to a complete governance stack for a single device category, and it has done so in roughly three years.
The stack starts with strategy, which in China’s system binds ministries, provinces, state capital and universities to a shared set of targets. Beneath it sits ethics: national BCI research ethics guidelines issued in February 2024 set out principles including beneficence, autonomy, privacy and minimum harm. Then vocabulary. The terminology standard YY/T 1987-2025, issued by the NMPA on September 16, 2025 and in force from January 1, 2026, defines what counts as a BCI medical device, categorises technical approaches, and standardises terms for signal acquisition, decoding and clinical application. The NMPA’s deputy director-general for device registration said at the time that inconsistent terminology had been hindering industrial development.
Classification followed. Draft guidance was published for comment in March 2026 and final guidance on classification and generic naming was issued in June 2026. It holds that a product qualifies as a BCI medical device only if it acquires signals from the central nervous system and decodes them in real time, among other core characteristics. That matters more than it sounds. It tells developers which rulebook they are playing by before they spend money on a pivotal trial.
Then pricing, approval and reimbursement, which is where the eleven days happened. And finally, data. On September 14, 2026 the NMPA issued what it described as the world’s first product standard for BCI medical devices that use AI to process electroencephalogram data. It sets requirements across data acquisition, processing, annotation, storage and access, and takes effect on September 1, 2027. It is China’s third BCI-specific device standard, following the terminology standard and a 2025 standard on test methods for closed-loop implantable neural stimulators.
The order of issue is itself the point. Pricing came before approval. Vocabulary came before classification. The data standard arrived after the first product but more than a year before it takes force, giving developers a known target. At no stage did a downstream institution have to wait for an upstream one to finish.
A second door into the clinic
Seven weeks after NEO’s approval, the architecture gained a new wing. The Regulations on the Clinical Research and Clinical Translational Application of Biomedical New Technologies came into force on May 1, 2026. They are administered by the National Health Commission (NHC), not the NMPA, and they cover a category of “biomedical new technologies” defined broadly enough to include both invasive and non-invasive BCIs.
Under this track, early-stage clinical research can proceed after internal academic and ethics committee review at the host institution, followed by registration with the NHC within five working days. No prior government approval is required.
Read quickly, that sounds like a shortcut around device regulation. Law firm Bird & Bird, in an analysis of the new framework, cautions that it is not. The NHC track is a route into the clinic for research and translational use, not a route into the market. Once a programme produces a technology with a defined product form, the medical device track and its Class III registration requirements apply. Developers who blur that boundary risk regulatory uncertainty later, precisely when they can least afford it.
The strategic logic is clear enough. China wants a large volume of human data flowing from hospitals running early BCI research, and it wants that research to feed a pipeline that eventually passes through the device regulator. Separating the two tracks lets the first move fast without weakening the second.
Approved, coded, listed and still unsold
For all the institutional speed, the commercial picture is sobering, and the most revealing evidence comes from the company itself.
Neuracle filed for an initial public offering on the Shanghai Stock Exchange’s STAR Market on June 11, 2026, seeking to raise Rmb 2.5 billion. Of that, Rmb 1.54 billion is earmarked for BCI research, Rmb 550 million for working capital and Rmb 410 million for manufacturing capacity. The prospectus reports 2025 revenue of Rmb 108 million, up 64 per cent year on year, and all of it came from non-invasive products, chiefly EEG systems sold to hospitals and research institutions. The implant that made global headlines generated no revenue in the reporting period. By the end of 2025 the company had implanted 32 patients.
Cost is the obvious constraint. Estimates cited by US developer Paradromics put the total cost of a NEO procedure at roughly 300,000 to 500,000 yuan. The government-guided service fee covers the surgical procedure, a small fraction of that total, and the device itself is not yet fully reimbursable under national insurance. The first procedure carried out under the guided pricing framework, at Wuhan’s Tongji Hospital in December 2025, was funded by research grants. The prospectus risk disclosures list uncertain clinical adoption, uneven provincial reimbursement and limited pricing benchmarks.
None of this undermines the governance story. It sharpens it. A billing code is necessary for a device to reach patients at scale; it is not sufficient. China has removed the administrative reasons a hospital might decline to adopt a BCI. It has not yet removed the financial ones. The next test is whether national insurance extends to the consumable itself, and on what evidence.
The comparison everyone is making
Much of the international commentary has framed NEO as China overtaking the United States. On technology, it has not. Precision Neuroscience received FDA 510(k) clearance in April 2025 for its Layer 7 cortical interface, a 1,024-electrode array cleared for temporary use. Synchron closed a $200 million Series D in November 2025. Neuralink has implanted around 20 patients across its research studies, with commercial approval not expected before 2027 at the earliest. Paradromics received FDA approval for an early feasibility study of its high-data-rate intracortical system in November 2025. None of these companies yet holds premarket approval for a chronically implanted BCI. Several are technically more capable than NEO.
Where the United States is genuinely behind is the step after approval. In the US system, FDA clearance starts the reimbursement process rather than ending it. The Centers for Medicare & Medicaid Services then conducts its own assessment, which for a novel device with national coverage takes around five years. A study of 64 novel technologies authorised between 2016 and 2019 found that fewer than half achieved any Medicare coverage at all. The industry calls it the reimbursement valley of death.
Washington is responding. In April 2026, CMS and the FDA jointly announced the RAPID coverage pathway for Class II and Class III breakthrough devices, under which CMS would issue a proposed coverage determination on the day of FDA authorisation, potentially enabling Medicare coverage within about two months. A separate bill, H.R. 5343, would grant breakthrough devices four years of automatic Medicare coverage and require payment codes within three months of approval. It has not been enacted.
The honest summary is the one a US competitor reached itself. Paradromics, in a public analysis of NEO, concluded that China is not leading on pure technology, but is arguably more progressive in its regulatory and administrative coordination, and that this is the real strategic advantage. It is unusual for a company to say that about a rival system. It is worth taking seriously.
What speed does not prove
Three cautions belong in any serious reading of the sequence.
Speed is not evidence. Thirty-six implants with 18 months of follow-up is a sound basis for approving a narrowly indicated device. It is a thin basis for population-level confidence in long-term implant stability, infection risk, device retrieval and the durability of decoding performance. The system China has built will generate real-world outcome data faster than anyone else’s. Whether it publishes that data transparently, including the failures, will determine whether the rest of the world trusts it.
Coordination has a price. The same central alignment that moved pricing, approval and reimbursement in lockstep can also produce pressure to show results against targets. Plans measured in unicorns and procedure volumes create incentives that a regulator focused only on patient safety would not. Independent scrutiny of the evidence becomes more important, not less, as policy speed increases.
Neural data is a governance question in its own right. The new AI-EEG standard addresses how brain signal data is collected, labelled and stored for training decoding models, and no regulator elsewhere has issued an equivalent. It does not settle who may access that data. Commentators in the United States have pointed to China’s data security and intelligence laws as a reason for caution about neural data held by companies operating under them. That concern is not unique to China, since neural data privacy is unresolved in most jurisdictions, but it will shape whether Chinese devices and datasets travel across borders.
The questions Asia-Pacific regulators should now answer
The most useful response to China’s sequence is not to admire it or fear it. It is to ask every other regulator in the region how long each of the same steps would take there. Japan, South Korea, Australia, Singapore and India all have capable device regulators and at least some mechanism for expedited review of innovative products. Korea, for instance, introduced an integrated review system for innovative medical technologies in 2023 that aims to cut market entry for digital therapeutics from 390 days to 80 through concurrent assessment. What none has yet shown is a BCI-specific pathway that links classification, pricing, reimbursement and data rules in advance of the first product.
These are not rhetorical questions. For device companies, the answers determine where the first patients outside China will be treated and where the first real-world evidence outside China will be generated. A region with an ageing population, a large burden of stroke and spinal cord injury, and several credible neurotechnology research bases has every reason to want those answers to be good ones.
The lesson travels further than the device
It is tempting to treat NEO as a one-off, a single company and a single approval propelled by national ambition. The instruments tell a different story. Pricing categories, terminology, classification rules, a research track and a data standard are not built for one product. They are built for a category, and once built they can be reused for the next implant, the next indication and the next technology China chooses to prioritise. The NHSA’s 39 batches of pricing guidance for 180 new technologies make clear that BCI is one application of a general method.
That method is reproducible. It does not require China’s political system to copy. It requires a regulator, a payer and a standards body to agree, before a product arrives, what the product will be called, how it will be classified, what it will cost to use and how its data will be governed. Most health systems in Asia-Pacific run those four decisions in sequence, and wait for each to finish before starting the next.
China ran them in parallel and started early. The result was eleven days. The question for every other regulator in the region is not whether they could have approved NEO. It is how many years it would have taken them to let a hospital buy one.
Ankit Kankar
