A paralysed man in Shanghai picked up a cup using only his thoughts, then later wrote his own name without the device at all. A paralysed man in Texas moved a cursor with his mind and played chess online.
Same dream, two very different regulatory endings.
The two programs, side by side
Neuralink put its first implant in a person in January 2024. By 30 June 2026 the company had roughly 26 participants enrolled globally in its trials, up from 21 at the start of the year, according to disclosures reflected in a US SEC filing and corroborated by multiple outlets. The implant, called N1, is about 23 mm across and carries 1,024 electrodes on 64 ultrathin threads placed in the motor cortex by a surgical robot. Trials run in the US, the UK (seven patients via GB-PRIME at University College London Hospital), Canada and the UAE.
China's NEO (Neural Electronic Opportunity) was approved by the National Medical Products Administration (NMPA) on 13 March 2026 as the world's first approved implantable brain-computer interface (脑机接口) Class III medical device, confirmed by Xinhua and People's Daily. It was developed by Neuracle Technology (Shanghai) Co., Ltd. (博睿康) and tested under a multicentre trial led by Huashan Hospital (Fudan University) across 11 hospitals with 32 cases, where hand-grasp function improved in 100% of patients.
The gap that actually matters
The headline gap is regulatory. Neuralink remains an investigational device in the United States with no FDA premarket approval as of 2026. China granted NEO a marketing licence and, by July 2026, performed its first commercial implant surgery at Huashan. Shanghai has since brought the system onto medical insurance — meaning a Chinese patient can, in principle, be treated and reimbursed today, while an American patient can only join a study.
That is not a verdict on which technology is better. It is a verdict on which system moved a device from lab to licence first.
The technical divergence nobody mentions
The two implants are not the same kind of invasive.
- Neuralink uses intracortical threads pushed into the brain tissue — higher resolution, higher surgical risk.
- NEO uses an epidural approach, sitting on the brain's surface, trading some signal richness for lower tissue damage.
Different trade-offs. Neuralink's 1,000-plus electrodes suggest finer control; NEO's epidural placement is explicitly designed to avoid direct brain injury. Neither has published long-term decay data — both need five to ten years of follow-up before anyone can claim durability.
Three cold showers
- Approved is not cured. NEO is a "hand motor function compensation system." It routes around the injury; it does not repair the severed nerve. Some patients showed signs of neural remodelling after training, but the mechanism is not yet understood.
- The populations are narrow. Both target specific, severe conditions — quadriplegia from cervical spinal cord injury, or ALS. Tens of patients, not tens of thousands.
- It is still surgery. Infection, stability and signal decay over time are open questions for every implantable brain-computer interface (脑机接口).
What "approved" actually unlocks
An approval number sounds bureaucratic, but it changes three things a trial cannot. First, a doctor can prescribe the device rather than recruit a subject — the difference between a treatment and an experiment. Second, hospitals can budget for it, and in Shanghai's case insurers can reimburse it, opening a path most Western BCI patients do not yet have. Third, the manufacturer can sell, service and improve it as a product, not a research line item. None of this means NEO is finished; only that it crossed the line Neuralink has not, from study to shelf.
What this means if you or a family member is affected
- Check approval and reimbursement before hope. "Approved" and "covered" are different doors; China's NEO has walked through both, Neuralink through neither.
- Match the device to the diagnosis. These are not general-purpose brain upgrades; each has strict indications.
- Be wary of consumer "BCI" gadgets. Headbands and focus trainers are not the same as a regulated implantable brain-computer interface (脑机接口).
The race is not about who implanted first. It is about who turned an implant into a product a doctor can prescribe.
Honest limitations
Neuralink's count (about 26 as of 30 June 2026) is drawn from the company's own disclosures as reflected in a US SEC filing and secondary reporting; we could not independently audit the enrolment and present it as company-reported. NEO's approval date, 32-case trial and 100% grasp-improvement rate come from Xinhua, People's Daily and Fudan University's Huashan Hospital — primary or near-primary Chinese sources; we did not access the NMPA registration dossier directly. "World's first approved implantable BCI" reflects China's NMPA claim and Xinhua wording; no comparable FDA approval existed as of 2026, but other jurisdictions may use different device definitions. No public RMB price for NEO was available; Shanghai insurance coverage is reported but the reimbursement amount is not stated here, so no currency conversion is given.