NEURALINK: The Brain Chip That Is Actually Working

A 34-year-old who could no longer speak or type is now composing thoughts at 120 words per minute. Twenty-six people worldwide are living proof. The question is no longer whether this works.

Image created by Author via ChatGPT / Prompted by Abel Arroyo

A man who lost the ability to speak and type eighteen months before receiving a surgical implant in his motor cortex described the experience in a paper published in Nature Neuroscience with seven words: it's like having a keyboard in my head. He is identified only as Patient 2 in the published literature. He is 34 years old. He has ALS. He typed that sentence using only his thoughts, at a speed of 120 words per minute, faster than the average person types on a physical keyboard. The implant in his skull is a quarter-sized disc called the Link, produced by Elon Musk's company Neuralink, containing 4,096 ultra-thin electrode threads inserted directly into his motor cortex by a fully automated surgical robot. As of late June 2026, 26 people worldwide are living with a Neuralink device. None of them are science fiction. All of them are patients who had run out of other options. The question that has occupied the scientific community for two years is no longer whether brain-computer interfaces work. It is who controls them, what they will eventually be used for, and whether the company that currently leads this field can be trusted to answer that question honestly.

THE ORIGIN STORY

Elon Musk's path to founding Neuralink began not with neuroscience but with fiction. His stated interest in brain-computer interfaces stemmed from a science fiction concept called the neural lace, introduced in a series of ten novels called The Culture, written by Scottish author Iain M. Banks between 1987 and 2012. In Banks' fictional universe, the neural lace is a digital layer of intelligence woven into the human brain at birth, giving its host instantaneous access to vast computational networks without any external device. Musk defined his own version publicly as a digital layer above the cortex, and described the long-term goal simply: symbiosis with artificial intelligence. He believed, and has stated consistently, that artificial intelligence poses an existential threat to humanity if it surpasses human cognitive capability without any mechanism for humans to participate in or direct that superintelligence. His solution was not to slow AI development. His solution was to upgrade human cognition fast enough to stay relevant.

Neuralink was founded in 2016 alongside a team of eight scientists and engineers including Max Hodak, Benjamin Rapoport, Dongjin Seo, Paul Merolla, Philip Sabes, Tim Gardner, Tim Hanson, and Vanessa Tolosa, drawing from neuroscience, biochemistry, and robotics. In April 2017, the company publicly announced its short-term goal of making devices to treat serious brain diseases, with the eventual goal of human enhancement, what is sometimes called transhumanism. Musk said he believed the device could eventually be something analogous to a video game, like a saved game situation, where you are able to resume and upload your last state, a description of human consciousness as a data file that can be backed up, transferred, and restored.

The founding team did not stay intact. Max Hodak, who had been one of the company's most prominent early voices, left in May 2021, tweeting afterward that he was glad to no longer be working for a company with aims as absurd as Jurassic Park, a pointed public departure that signaled internal disagreements about the company's direction and ambitions. He was not the last senior departure. The company that emerged from its early years was built increasingly around Musk's personal vision rather than the collective scientific judgment of the neuroscience team that had originally given it credibility.

What Neuralink actually needed to solve first was a problem that had existed since the earliest brain-computer interface research began. The Utah Array, a device developed by Blackrock Neurotech and used in academic research since the early 2000s, had demonstrated definitively that a paralyzed person could control a cursor on a screen using neural signals. The BrainGate consortium, a collaboration between Brown University, Harvard Medical School, Massachusetts General Hospital, and the Providence VA, had treated patients successfully with this approach for nearly two decades. But the Utah Array used 100 electrodes. Reading the motor cortex with 100 electrodes is like trying to understand a symphony by listening to a single instrument. The signals were there. The resolution was not. Neuralink's core engineering contribution was solving the density problem at scale.

THE TURNING POINT

The first human implant happened on January 29, 2024. The patient was Noland Arbaugh, a 29-year-old from Arizona who had been paralyzed from the shoulders down since a diving accident years earlier. Arbaugh told journalists he had complete faith in Neuralink going in, knowing there were risks and that it might not work. Within weeks of the procedure he was controlling a computer cursor, browsing the internet, playing online chess, and posting on social media using thought alone. He spent an entire night playing Civilization VI, a complex strategy game, for hours without fatigue. The engineers at Neuralink were monitoring his neural signals from California. Musk himself stated publicly that the patient seems to have made a full recovery with no ill effects, and that the patient was able to move a mouse around the screen by just thinking.

The early weeks did produce a real technical setback. A number of the ultra-thin electrode threads retracted slightly from the brain tissue after implantation, reducing the effective number of functional electrodes. The engineering team adjusted the signal processing algorithms to compensate, and Arbaugh's performance recovered and continued improving as his brain adapted to the device. This retraction issue became the primary engineering problem the N2 chip was redesigned to address.

The second patient, identified as Alex in company updates, avoided the thread retraction problem entirely, benefiting from revised electrode geometry informed by the first patient's data. Alex is using his implant to learn computer-aided design software, designing three-dimensional objects on a screen using thought alone. A third patient named Brad used his implant to play Mario Kart with his children, a moment Neuralink co-founder DJ Seo described in a company update as incredible. These are not laboratory results presented in abstract form. They are human beings doing things they could not do without a machine in their skull, described in their own words and documented in peer-reviewed literature and company updates.

By September 2025, twelve people worldwide had received Neuralink implants. By January 2026, twenty-one. By late June 2026, twenty-six, with trials now running not just in the United States but in Canada, the United Kingdom, and the United Arab Emirates, where seven patients are enrolled in the GB PRIME study in Great Britain alone. The company is running two parallel clinical studies. The PRIME Study targets people with quadriplegia from spinal cord injury or ALS who want to control computers and robotic arms through thought. The VOICE trial focuses specifically on decoding intended speech from neural signals for people who have lost the ability to speak. Imagine if Stephen Hawking could communicate faster than a speed typist or auctioneer, Musk said publicly. That is the goal.

THE BODY OF WORK

The Link implant is a 23 millimeter disc, roughly the size of a quarter, that replaces a small circular section of skull during surgery. The R1 surgical robot, which Neuralink developed specifically for this procedure, inserts up to 192 electrode threads per minute using tungsten-rhenium needles 25 micrometers in diameter, thinner than a human hair, completing the insertion process in 20 to 40 minutes. The robot's precision matters because the electrode threads must be placed to avoid blood vessels while achieving sufficient depth in the cortical tissue to record neural signals reliably. The N1 chip, used in the first patient, contained 1,024 electrodes. The N2 chip, now being implanted in current participants, contains 4,096, four times the density, capturing neural signals from a substantially larger area of the motor cortex and producing the communication bandwidth that enabled the 120-words-per-minute result. The device transmits data wirelessly through the skull to an external receiver, which connects to a computer and translates the neural signals into cursor movement, keyboard input, or robotic arm control through machine learning algorithms trained on each individual patient's neural patterns. You can go to hospital in the morning and leave by afternoon, Musk has said. And it can be done without general anesthesia.

The competing company Synchron has taken a fundamentally different approach. Their Stentrode device is a small mesh stent inserted through the jugular vein and navigated through the vascular system to a blood vessel near the motor cortex, where it sits inside the vessel wall and reads neural signals through the surrounding tissue. The procedure resembles a cardiac stent implantation. It carries substantially lower surgical risk than Neuralink's approach but produces lower signal resolution. Reading neural activity through a blood vessel wall gives you less information than electrodes placed directly in brain tissue. Both approaches are active in human trials in 2026. Both are producing real results. They represent two different philosophies about how much risk is acceptable in exchange for how much capability.

Precision Neuroscience, founded by former Neuralink employees, is developing a flexible cortical surface array called the Layer 7 Cortical Interface that sits on the brain's surface rather than penetrating it, another middle path between the two existing approaches. Paradromics is developing Connexus, a high-bandwidth BCI designed for information transfer speeds that exceed any current system. Blackrock Neurotech's Utah Array, which established the field's proof of concept two decades ago, remains in use in academic research settings. The competitive landscape in 2026 is the most active in the field's history, driven by a combination of genuine medical need, massive private capital, and the looming question of what these devices will eventually be capable of beyond their current therapeutic applications.

Neuralink is also developing a second device called Blindsight, which represents a fundamentally different direction from Telepathy, the motor control application currently in human trials. Where Telepathy reads signals coming out of the brain, Blindsight works in the opposite direction, writing signals into the brain's visual cortex to create the perception of sight. The pipeline is a camera mounted on a pair of glasses that captures visual data, sends it wirelessly to an implanted chip in the visual cortex, which then stimulates neurons to produce phosphenes, specific perceived points of light that the brain gradually learns to organize into meaningful images. Blindsight received FDA Breakthrough Device Designation in September 2024. As of January 2026, Musk stated that the device was ready for its first human implant and awaiting final FDA regulatory approval. The initial visual experience, he acknowledged, would be low resolution, like Atari graphics. His longer-term claims, that Blindsight could eventually exceed natural human vision and allow users to perceive infrared, ultraviolet, and radar wavelengths similar to Geordi La Forge from Star Trek, have been met with scientific skepticism. Ione Fine of the University of Washington published a computational model showing that even a 45,000-electrode cortical implant would produce only a blurry, barely recognizable image, because each individual neuron has its own neural code and there is no current method to identify those codes accurately enough for high-resolution artificial vision.

THE TRAGEDY

The institutional contradictions inside Neuralink are the part of this story mainstream coverage consistently underreports because the human stories from the trial are too compelling to compete with.

Musk has never resolved the tension between the three distinct and incompatible goals he has stated publicly for this company. The first is helping disabled people regain function, the mission that generated FDA approval and the human stories that dominate press coverage. The second is building a consumer product for healthy people to enhance cognitive performance, play video games at unprecedented neural bandwidth, and achieve what Musk calls human-AI symbiosis. He has described this as an eventual consumer technology that healthy people will elect to receive voluntarily. The third goal is preventing an existential threat from artificial intelligence by merging human cognition with machine systems before AI surpasses human intelligence entirely. He stated publicly that he believes AI is an existential threat to humanity if it goes unchecked, and that the device is his proposed solution to that threat.

These are not the same project operating under different timelines. A medical device that helps a paralyzed ALS patient communicate operates under one regulatory, ethical, and clinical framework. A consumer enhancement product for healthy people operates under a completely different one. And a technology designed to merge human consciousness with artificial intelligence to prevent an AI apocalypse operates under a framework that does not yet exist and that no regulatory body has yet defined. Neuralink moves fluidly between all three framings in public communications, presenting its most sympathetic face when scrutiny is highest and reverting to the transhumanist vision when addressing investors and enthusiasts.

The regulatory dimension of this tension came into sharp focus in January 2026, when Neuralink recruited a senior official directly from the FDA office that regulates the company. STAT News described the move as having surprised, impressed, and infuriated competitors in the field simultaneously, with rivals arguing that Neuralink's aggressive recruitment of federal regulators creates a revolving door dynamic that compromises the independence of the oversight process. The company's valuation has now exceeded nine billion dollars after a 650 million dollar funding round in June 2025, making the financial stakes around that regulatory relationship substantial.

The animal welfare concerns have a specific, documented history. Wired magazine published an investigation in September 2023 titled The Gruesome Story of How Neuralink's Monkeys Actually Died, reporting that monkeys used in Neuralink's preclinical testing suffered from brain hemorrhages, paralysis, and infected implant sites, and that records were improperly maintained during the transportation of biological materials from test animals. The Physicians Committee for Responsible Medicine spent years writing to lawmakers and federal agencies about the company's animal testing protocols. Musk stated publicly that no monkey had died as a result of a Neuralink implant and that the company had used terminal animals already near death for testing. The FDA's own initial rejection of Neuralink's human trial application cited concerns about lithium battery safety, potential implant migration within the skull, thread biocompatibility, long-term data security, and surgical complications including infection and hemorrhage. The company addressed those concerns through additional animal testing and design modifications before receiving approval. That history means the human trial results being reported in 2026 are built on a foundation that was genuinely contested at the regulatory level before it was cleared.

THE LEGACY AND THE VAULT

What is verified and documented as of July 2026 is this. Twenty-six people are living with a machine connected to their nervous system that allows them to interact with computers using thought alone. A man who lost his ability to speak is composing thoughts at 120 words per minute. A paralyzed patient is playing complex strategy games for hours. Another is learning to design three-dimensional objects using only his mind. A father who had not used his hands in years played a video game with his children. These are not projections. They are documented clinical outcomes from ongoing FDA-regulated trials, published in peer-reviewed journals, and confirmed by independent researchers.

What is not yet verified, and what the next decade of this technology's development will determine, is whether the company producing these results can be trusted to deploy them at consumer scale without regulatory capture, without creating a cognitive divide where wealthy people upgrade their neural bandwidth while everyone else falls further behind, and without honestly resolving the fundamental question it has never formally answered. Neuralink's own website states its ultimate goal is to create a generalized input-output platform capable of interfacing with every aspect of the human brain. That is not a medical device mission statement. That is a declaration of intent to colonize human consciousness itself.

What does it mean that the most consequential medical technology since the MRI is being developed primarily by a private company whose founder has stated in his own words that its ultimate purpose is to merge human minds with artificial intelligence to prevent an AI apocalypse, that the FDA official most recently responsible for regulating that company now works for it, and that the 26 paralyzed patients currently benefiting from the technology are simultaneously the proof of concept for a consumer product none of them were asked to beta test?

About the Author

Abel Arroyo is a cultural critic and forensic biographer dedicated to the Intellectual Resistance, the creators, whistleblowers, and truth tellers who broke the machine to save their souls. His work interrogates the intersection of power, suppressed knowledge, and human cost, providing rigorous examination of the systems and technologies that define how we understand existence. If you find value in these deep dives into the architects of modern dissent and the machinery of the modern world, consider supporting independent journalism. buymeacoffee.com/dathaze20j