On April 15, 2026, surgeons at Beijing Tiantan Hospital implanted the domestically developed Beinao No.1 device in a paralyzed patient — and broadcast the entire procedure live in 4K. Frontierbeat covered the surgery the same day. Weeks earlier, Noland Arbaugh — the first person to receive Neuralink’s brain chip, implanted in January 2024 — announced he is now enrolled in college, starting a business, and preparing to speak publicly about life with a computer in his brain.
Two years ago, brain-computer interfaces were a speculative technology covered mainly by science journalists. Today they are in clinical use in multiple countries, with competing national programs, competing companies, and a rapidly expanding research base. Understanding what BCIs actually are — how they work, what they can and cannot do, and where the technology is going — requires separating signal from hype.
What a Brain-Computer Interface Is
A brain-computer interface is a system that captures signals from the brain and translates them into commands that external devices can interpret and act on. The direction of information can go both ways: from brain to device, and in advanced systems, from device back to brain.
The core components of any BCI are:
Electrodes that detect the electrical activity of neurons. Different electrode designs trade off signal quality against invasiveness — fine-resolution invasive implants capture individual neuron activity, while non-invasive external sensors pick up broader electrical patterns through the skull.
Signal processing hardware that amplifies and digitizes the raw neural signal, which is extremely small (on the order of microvolts) and needs to be separated from noise.
Decoding algorithms — typically machine learning models — that learn the mapping between specific patterns of neural activity and the user’s intended actions. These models improve over time as they accumulate more data from a given patient.
Output interface that translates the decoded intent into actual device commands: moving a cursor, typing text, activating a robotic limb, or triggering stimulation.
According to the National Institutes of Health, the fundamental goal of motor BCIs is consistent: record enough brain signals to interpret what someone wants to do, then convert that intent into a digital command. In theory, someone just needs to think about reaching for something and the system translates that thought into an action — bypassing the damaged neural connections that would normally carry that signal to muscles.
Invasive vs. Non-Invasive
The most significant technical divide in BCI design is between invasive implants that place electrodes directly in or on the brain, and non-invasive systems that read signals from outside the skull.
Invasive implants provide far higher signal quality and precision. They can record individual neuron firing rates, which enables fine-grained motor control — moving a cursor across a screen, typing, controlling a robotic arm with multiple degrees of freedom. The tradeoff is the surgery required to implant them, the long-term biocompatibility questions around electrode materials in brain tissue, and the regulatory pathway that follows from being a Class III medical device.
Non-invasive approaches (primarily EEG-based systems) record broader electrical fields through the scalp. They are safe, accessible, and require no surgery. But the signal is diffuse — you can detect someone’s intent to move their left hand versus their right hand, but not individual finger movements. This limits the bandwidth of control available to the user.
Most of the clinically significant work happening in 2026 involves invasive systems. Neuralink, Synchron, and Neuracle are the three companies with active human trials, as MIT Technology Review reported in April 2025.
Neuralink: The Most Visible Player
Neuralink was founded by Elon Musk in 2016 and received FDA approval for its first human trial (the PRIME Study) in May 2023. In January 2024, the company completed its first human implantation. By early 2025, at least five people had received the device, and Neuralink planned to expand to 20–30 more by year’s end.
A June 2025 funding round secured an additional $650 million, bringing Neuralink’s total raised to over $1 billion.
Neuralink’s distinguishing technical features are its electrode density and its surgical approach. The device uses ultrathin flexible threads — each with multiple electrodes — inserted into the motor cortex by a specialized robotic system that resembles a sewing machine. The robot is designed to avoid blood vessels and implant threads with sub-millimeter precision, reducing trauma. The chip is wireless and rechargeable, with no external wires.
The PRIME Study, which Noland Arbaugh is part of, focuses on motor control for individuals with paralysis from spinal cord injuries or ALS. Arbaugh — paralyzed from the shoulders down — has demonstrated the ability to control computers, play chess, and learn to code using only his neural activity.
Beyond motor control, Neuralink has received FDA Breakthrough Device Designations for a speech restoration device (May 2025) and Blindsight, a vision restoration system that would stimulate the visual cortex directly to restore some form of perception in people who are blind (September 2024).
Synchron: The Less Invasive Path
Synchron takes a different approach. Its device, the Stentrode, is delivered through blood vessels — not direct brain surgery. A catheter threads the device into a vein near the motor cortex, where it expands and anchors itself against the vessel wall. No craniotomy required.
The tradeoff is signal fidelity. The Stentrode collects a limited range of signals compared to a direct cortical implant, giving users a basic on/off toggle rather than continuous multi-dimensional control. MIT Technology Review noted that Synchron CEO Tom Oxley believes this simpler signal profile makes the device scalable to a far larger patient population — those who need reliable but limited control, rather than high-bandwidth communication.
Synchron has implanted its Stentrode in ten volunteers — six in the US and four in Australia.
China’s Beinao and the National Dimension
The live-streamed surgery in Beijing on April 15 was not a publicity stunt. It was a deliberate signal. China has invested heavily in neurotechnology as a strategic priority, and Beinao No.1 — developed domestically — represents its first fully home-grown cortical BCI device.
The procedure involved implanting the chip in a patient with paralysis at Beijing Tiantan Hospital, one of China’s leading neurosurgery centers. Broadcasting it live in 4K was a statement about the maturity of the technology and the country’s confidence in the procedure.
The competitive dynamic between US and Chinese BCI programs mirrors the dynamic in AI, semiconductors, and robotics: parallel development tracks, national security dimensions, and an implicit race to define the clinical and regulatory standards that will govern the technology globally.
What BCIs Can Do Today — and What They Cannot
The honest picture of current capabilities is narrower than the public imagination of BCIs tends to be.
What invasive BCIs can do today, for paralyzed patients in clinical trials: control a computer cursor with reasonable precision, type text at speeds competitive with eye-tracking systems, control a robotic arm for basic object manipulation, and — in some early work — decode intended speech sounds from neural activity.
What they cannot do: provide natural, high-bandwidth bidirectional communication with the richness of normal motor control. Restore sensation in the same way that motor control can be enabled (sensory restoration is harder than motor decoding). Be implanted without neurosurgical risk. Maintain signal quality indefinitely — electrode-brain interfaces tend to degrade over years as tissue reacts to the implant.
The most significant near-term advance will likely come from speech restoration, where several research groups including those at UC Davis and Massachusetts General Hospital have demonstrated the ability to decode intended words from neural activity at speeds that approach conversational rate, for patients who have lost the ability to speak.
The Questions That Follow
The ethical and regulatory dimensions of BCIs are not minor. Implanting a wireless digital device in a human brain that communicates via Bluetooth raises questions about data security (who owns neural data?), privacy (can the signals be intercepted?), autonomy (what happens when the company that made your implant goes bankrupt?), and identity (what does it mean to have your cognitive intent mediated by a commercial device?).
These questions are not hypothetical. Neuralink’s first patient is already living with them. The Frontiers in Human Dynamics paper published in March 2025 argued that “the potential for using BCIs in enhancing healthy individuals, as stated by the company itself, will emerge” — and that oversight will need to be considerably more rigorous than it has been for purely medical applications.
For now, the technology is in the hands of a small number of research volunteers and clinical trial participants. What it does at scale — across thousands of users, across companies with different incentives, across jurisdictions with different regulatory frameworks — is the defining question of the decade ahead.
See also: Noland Arbaugh, Two Years After Neuralink | China’s Live-Streamed BCI Surgery | China Deploys Humanoid Robots on Factory Lines