People who have lost the ability to move or speak may soon have a new option: surgically implanted devices that link the brain to a computer.
More than two decades after researchers first demonstrated that a person could move a computer cursor with their thoughts, several firms are poised to take the brain-computer interface (BCI) from experimental curiosity to commercial product.
"We know it works, we know the enabling technologies are now ready," says Michael Mager, the CEO of Precision Neuroscience. "It's time to turn this academic work into a thriving industry that can make a big impact on people's lives."
Already, experimental brain-computer interfaces have been implanted in dozens of people. The latest devices go under the skin and can communicate wirelessly with a smartphone or tablet.
Elon Musk's Neuralink is the most visible player in the BCI field. But the first product to reach the market may well come from competitors including Precision, Blackrock Neurotech, Paradromics, or Synchron.
Some of these companies, like Blackrock, have much more experience than Neuralink. Others use less invasive, and potentially safer, technology that may make it easier to get approval from the Food and Drug Administration.
The first BCI customers are likely to be people living with paralysis from a spinal injury or amyotrophic lateral sclerosis (ALS). Early products will allow them to control a computer cursor, or generate artificial speech.
Neuralink's 'telepathy'
Implanted BCIs work by detecting and decoding signals coming from areas of the brain that control movement or speech. These signals indicate when a person is trying to move a limb or speak a word.
A BCI system typically includes sensors that detect brain activity, an interface that processes the signals, and an external device that turns thought into action. The result: A cursor moves, a prosthetic hand reaches, a synthetic voice speaks the words a person is trying to articulate.
"Imagine the joy of connecting with your loved ones, browsing the web, or even playing games using only your thoughts," says the narrator of a promotional video from Neuralink.
The company, which did not respond to requests for an interview, calls this capability "telepathy."
Neuralink pushed BCIs into the public imagination in early 2024, thanks to a charismatic and resilient man with paralysis.
A diving accident left Noland Arbaugh unable to move from the shoulders down. At 29, he became the first person to get Neuralink's device.
A robot threaded more than a thousand electrodes into his brain's motor cortex at the Barrow Neurological Institute in Phoenix. Then, human surgeons there installed a wireless interface about the size of a quarter in his skull.
A few weeks later, Arbaugh was on stage at Neuralink's headquarters in Fremont, California, describing his experience controlling a computer cursor.
"It's freakin' wild," he said. "When I first moved it just by thinking, it blew my mind for like a day. I just could not wrap my head around it."
A video featuring Arbaugh's remarks has attracted more than 25 million views on Musk's social media platform, X.
But the success was tempered by Neuralink's announcement a few weeks later that some of the threads of electrodes in Arbaugh's brain had "retracted," making the device less sensitive.
Since then, Neuralink has reported implanting its BCI in at least six other people. But details about those experiments remain scant.
A new technology, decades old
While Neuralink's surgical robots and wireless electronics are new, using thoughts to move a cursor isn't.
Dr. Leigh Hochberg — who holds positions at Brown University and Massachusetts General Hospital — was part of a team that pioneered the approach in 2004.
Their subject was Matt Nagle, a man who was living with paralysis after being stabbed in the neck. Hochberg's team linked Nagle's brain to a computer using old-fashioned wires that passed through his skull.
A research video from 2004 shows Nagle using his thoughts to open an email.
"It was exactly what was supposed to happen," Hochberg says. "And even for all of us that were expecting it — there was a little bit of magic there."
Nagle died in 2007 of an infection unrelated to the experiment.
BrainGate evolved into an academic consortium directed by Hochberg. And in June 2025, a team at the University of California, Davis reported that a BrainGate 2 BCI allowed a man with ALS to speak through a computer.
"I. Am. Good," the synthesized voice says in a video accompanying the study. The speech is slightly halting, spoken one word a time. But the voice sounds human — it was constructed from old audio of the man speaking.
Experiments like that one show how computer interfaces have improved, Hochberg says.
Instead of monitoring a few dozen neurons, they may listen to thousands. Instead of sending information out through wires, they use wireless protocols. And instead of interfacing with a wall of computers, the signals may go to a single laptop or tablet.
Another big change is that scientists keep finding ways to decode brain activity "more accurately, more consistently and more reliably," Hochberg says.
In the past few years, that has meant employing artificial intelligence to recognize the neural activity patterns that reveal a person's intention to speak, or pick up a bar of chocolate.
The field has also become specialized, Hochberg says, with some groups focused on decoding speech while others work on improving control of robotic limbs.
There are even groups "focused on putting information back into the brain," Hochberg says, which can add a sense of touch to a robotic arm or hand.
Fingers that feel
The University of Pittsburgh is among the leaders in providing sensory feedback through a brain-computer interface.
"You can't have fine and dextrous motor control with visual feedback alone," says Jennifer Collinger, a professor at the University of Pittsburgh. You need that sense of touch to be able to respond in a natural way."
With touch feedback, the user can tell when an artificial finger makes contact with an object or when an artificial hand is holding a cup tightly enough to keep it from falling.
So Collinger and her colleagues have been working with Blackrock Neurotech, whose brain interface technology has been used experimentally in dozens of people.
One of those people is Nathan Copeland, who was paralyzed in a car accident. In 2016, Copeland famously used a robotic arm to bump fists with President Barack Obama.
In 2021, Copeland was part of a study that showed how a sense of touch improved his ability to grasp and manipulate objects with his prosthetic hand.
"With sensation, I could feel that the hand had made contact," Copeland said in a 2021 interview with NPR. "I could also tell if I had a firm grip on it or not."
But advanced features like sensory feedback aren't going to appear in the first implanted devices on the market, Collinger says. Instead, they are likely to offer control of a computer cursor, much like BrainGate did in those lab experiments more than 20 years ago.
"There's been enough consistent success that now companies are saying, 'Okay we can offer a first-generation device to people that will offer some kind of benefit to them,'" Collinger says.
One of those companies is Precision Neuroscience, which was cofounded by Ben Rapoport, a neurosurgeon and engineer who had previously helped start Neuralink.
Precision's other co-founder and CEO, Michael Mager, says the company's short-term goal is a wireless device that allows a person with paralysis to operate a smartphone or computer.
"We think about accessing news and entertainment, we think about productivity software like Microsoft Office, Word, Powerpoint, Excel," Mager says. "If you can operate those programs as well as someone who is able-bodied, it's quality-of-life enhancing — and it's also potentially enabling for people to go back to work."
Precision's device differs from Neuralink's because it doesn't insert its electrodes into the brain.
"We have a very, very thin film that is designed to sit on the surface of the brain without penetrating into or damaging the brain," Mager says.
That makes the implant safer and less invasive, Mager says, which could make it easier to get approval from the FDA.
Synchron avoids opening the skull entirely. Its electrodes are delivered through blood vessels using technology designed to place stents in blocked arteries.
All of these devices face some common challenges, Mager says.
"We're sampling from thousands of electrodes, thousands of times a second, and the amount of data that comes off of these systems is just enormous," he says.
It is far too much data to transmit through existing wireless links. So companies are working on ways to reduce or compress the data.
Another obstacle is the cost to conduct the sort of clinical trials required by the FDA. That will probably be hundreds of millions of dollars, Mager says.
Even so, Mager thinks his company and several others, including Neuralink, have the resources and expertise to turn the brain interface concept into a marketable product.
That won't take another 20 years, he says. Perhaps another two or three.
Copyright 2025 NPR
