Neuralink Has Its First Human Patient. The Real Race for Brain-Computer Interfaces Is Just Starting.

On January 29, 2024, Neuralink announced that its first human patient had received an implant from its N1 chip — a 1,024-electrode array placed on the surface of the motor cortex. The patient, 29-year-old Noland Arbaugh, had been paralyzed below the shoulders following a diving accident. Eight days post-implant, he was using his thoughts to move a computer cursor and play chess. He later streamed himself playing Civilization VI for over eight hours using only neural commands.

This was not the first human BCI demonstration — that distinction belongs to researchers who implanted electrode arrays in patients decades earlier under academic studies — but it was the most visible, backed by Elon Musk's company and its $363 million in venture funding, and it brought brain-computer interface technology into mainstream technology coverage in a way academic studies had not.

In the months since, BCI moved from proof-of-concept to a crowded competitive space with multiple companies demonstrating implanted devices in human patients. The technical question — can a chip in a brain let a paralyzed person control a computer? — has been answered affirmatively. The frontier has shifted to harder problems: electrode durability, signal fidelity over time, surgical risk, and eventually, the question of what this technology should actually be used for.

How They Work

Invasive BCIs — the kind that require surgery — consist of electrode arrays placed in or on the brain's cortex that record electrical signals from neurons. When neurons fire, they produce tiny electrical pulses; the electrodes detect these pulses, and signal processing software interprets the patterns as intended movements, cursor directions, or other outputs.

Neuralink's N1 chip uses 1,024 electrodes across 64 flexible threads, each thinner than a human hair. The surgical robot that places the threads — the R1 — inserts them with precision designed to avoid blood vessels, which reduces the bleeding and inflammation that have historically degraded electrode signal quality over time. The chip transmits data wirelessly; there are no wires through the skull, removing a major historical infection risk.

Synchron takes a different approach. Its Stentrode device is implanted via a catheter through the jugular vein into the superior sagittal sinus — a blood vessel running along the top of the brain. No brain surgery required. The electrodes sit inside the vessel wall, close enough to the motor cortex to detect neural signals. The signal fidelity is lower than a direct cortical implant, but the safety profile is dramatically better: endovascular implantation is a well-established medical procedure, while open cranial surgery is not.

Precision Neuroscience, founded by a former Neuralink co-founder, uses a different insertion method: a thin flexible array placed on top of the cortex through a small slot in the skull, without penetrating brain tissue. The Layer 7 Cortical Interface, as Precision calls it, has been placed in surgical patients during planned cranial procedures since 2023, accumulating safety data on how the brain responds to surface electrode placement without the risks of a dedicated implantation surgery.

The Electrode Durability Problem

One of Neuralink's most significant early disclosures was that some of Arbaugh's electrode threads had retracted from the cortex in the weeks after implantation — a phenomenon called "pullback" — reducing the number of effective recording electrodes from 1,024 to several hundred. Neuralink's software team adapted the decoding algorithms to compensate, and Arbaugh's cursor control actually improved after the retraction as algorithms were refined, but the episode illustrated the fundamental challenge of keeping precision recording devices stable inside living tissue.

The brain sits in cerebrospinal fluid and moves slightly with every heartbeat and breath. Rigid implants — earlier-generation electrode arrays that sit in place and don't flex — experience micromotion relative to surrounding tissue, causing scarring that degrades signal quality over months to years. Neuralink's flexible threads were designed to move with the brain, reducing this scarring. The retraction events suggest the biocompatibility problem is not fully solved.

Synchron's endovascular approach sidesteps some of these issues — the device sits in a blood vessel rather than brain tissue — but the signal quality tradeoff is real. Synchron patient Timothy Dick, who received the Stentrode in 2021 in Australia, has demonstrated typing and tablet control through neural commands, but at a speed considerably slower than Neuralink's demonstrations.

What They Can Actually Do Right Now

Current approved BCIs can do a meaningful but limited set of things. Paralyzed patients can move computer cursors, type using gaze-and-neural-selection systems, and control tablet interfaces at speeds comparable to a slow typist (roughly 20-30 words per minute for the best current systems). BrainGate consortium patients at academic hospitals demonstrated cursor control and even limited robotic arm movement in prior trials. Neuralink has demonstrated cursor control at speeds that make computer use genuinely practical for daily activities.

Decoding more complex signals — speech, fine motor control, emotion — remains largely experimental. A Stanford/Neuralink collaboration published results in 2024 showing speech decoding at 62 words per minute for an ALS patient, substantially faster than previous systems. The patient could communicate in a way that resembled natural conversation speed for short exchanges.

The Non-Invasive Track

Alongside implanted devices, a parallel track of non-invasive BCIs has seen significant investment. Neurosity, Meta's research division, and several startups are developing EEG-based headsets that read neural signals through the skull. The signal quality is much lower — the skull attenuates electrical signals substantially — but the safety profile is trivially better than surgery.

Meta's 2023 paper demonstrating decoding of imagined handwriting from surface EEG attracted significant attention, though the system required considerable training data and worked best in constrained environments. The practical applications for non-invasive BCIs are currently limited to simple control interfaces and potential mental health monitoring, not the high-throughput communication that implanted devices approach.

The Ethical Landscape

As BCIs approach practical clinical use, the ethical questions are becoming concrete rather than hypothetical. Who owns the neural data recorded by a BCI? What are the liability implications if a hacked BCI is used to manipulate motor control? How are consented research patients protected when a company like Neuralink is simultaneously a for-profit entity with investors expecting returns?

The FDA's regulatory framework for BCIs as medical devices requires safety and efficacy demonstrations for specific indications — currently focused on paralysis and motor restoration. Whether and how these devices might eventually be approved for cognitive enhancement, communication augmentation in non-disabled users, or integration with consumer devices remains deeply uncertain — both technically and regulatorily.

The first-generation results are genuinely impressive for patients who had no other options. Arbaugh has described the ability to control his computer as transformative for his independence and quality of life. At that level — restoring lost function to people with severe paralysis — the technology's value is clear. Everything beyond that is still a very long road.