A high-stakes technology race is playing out in the human brain. Brain-computer interfaces are already letting people with paralysis control computers and communicate their needs, and will soon enable them to manipulate prosthetic limbs without moving a muscle.

The year ahead is pivotal for the companies behind this technology.

Fewer than 100 people to date have had brain-computer interfaces permanently installed. In the next 12 months, that number will more than double, provided the companies with new FDA experimental-use approval meet their goals in clinical trials. Apple this week announced its intention to allow these implants to control iPhones and other products.

There are dozens of so-called “neurotech” startups. Four lead the field of implants: Paradromics, Synchron, Precision Neuroscience and Elon Musk’s Neuralink, which in some ways is the most ambitious of the four. All but Paradromics have reached the point at which they are putting tech inside people’s heads. Each has its own approach, and all offer reasons they believe their product will come out ahead.

All four are betting they’ll eventually become a standard part of care for tens of thousands, perhaps even millions, of us. The prize they’re after: Morgan Stanley projects a $1 billion-a-year brain-computer implant market by 2041.

Other than perhaps the quest for human-level artificial intelligence, or colonization of other planets—not coincidentally two other areas where Musk is a big proponent—few fields exhibit such a wide gulf between a technology’s potential and its near-term prospects.

“There is a vision that this is going to be a mass-consumer thing, which is a vision that you can sell,” says Dr. Iahn Cajigas, a neurosurgeon at the University of Pennsylvania who has done pioneering research on brain implants, and has installed them in a handful of patients. “As a clinician, I find that kind of a dangerous way to talk.”

These are medical products, he emphasizes, with all the risks that attend brain surgery, including infection. “To take the risk of a brain implant, if you’re a young person with no medical problems, because you’re at the mall and you want a better interface with your phone, I don’t know how reasonable that is in the current world we live in,” Cajigas added.

For the leading companies in the brain-computer interface market, it’s generally accepted that the more bandwidth required, the more invasive the implant must be. Future breakthroughs in signal processing aside, implants have to go deeper into our brain tissue to get the best performance.

Unknowns about safety, performance and cost are why the trials that happen in the coming year could make or break these four contenders.

The blood-vessel route: Synchron

Synchron, the first to collaborate with Apple, is among the least invasive. Its implant, a tubular mesh of electrodes, is run through a major blood vessel in the brain, like a stent. It can be installed without opening the patient’s skull, so more physicians could be trained to perform the operation, says Kurt Haggstrom, the company’s chief commercial officer.

The downside: The brain-activity readings from the electrodes tend to be less precise. In the Apple scenario, patients must wear Apple’s Vision Pro goggles for now. They move a cursor via eye tracking, not mind control, then “click” an item by thinking about a large movement of one of their limbs.

By the end of 2025, Synchron is to begin final FDA trials of its implantable brain-computer interface. Those trials will take about two years, says Haggstrom.

A slit in the skull: Precision Neuroscience

Precision Neuroscience aims to put a small, flat array of electrodes onto the surface of people’s brains. While the current system is wired, Precision is developing one that is completely wireless, where nothing protrudes through skin and it communicates and recharges wirelessly.

With 1,024 electrodes spread across 1.5 square centimeters, the system can potentially do more than Synchron’s. For example, it might be able to translate thought to speech.

A key challenge: Neuralink and others benefit from decades of deep-brain recordings in primates. Precision records neural activity differently, and researchers are only beginning to map the signals, says Cajigas, who has tested it in 11 patients so far. (He’s not a paid Precision collaborator.)

“In the next year, I think this could be a viable solution for patients who are amputees to control a robotic hand,” he adds.

With its new FDA permissions, Precision can install its system in a person’s head for up to 30 days. The company will be putting its devices in somewhere between dozens and a hundred patients in the next 12 months, says CEO Michael Mager.  If those trials are successful, the company will test more permanent implants.

The cortex as pin cushion: Paradromics

Paradromics’ brain-computer interface looks like a coin with Velcro on one side, with 421 tiny electrodes that push 1.5 millimeters into the brain. Installing several of these electrode arrays could allow for an especially fast connection, like the difference between a bad Wi-Fi signal and a great one. It can record from individual neurons, like Neuralink’s system, says Chief Executive Matt Angle.

The company’s electrodes are so small, they could in theory go unnoticed by the patient’s brain, preventing the kind of scarring and other issues that bedeviled early systems in university labs, he adds. The company hasn’t installed one in a human yet, but two have been inside the brains of sheep for three years, and both maintained a strong connection to the brain throughout that time.

Paradromics is part of an FDA program designed to accelerate the approval of breakthrough medical devices, and plans to start its first clinical trial in humans later this year.

Deep brain dive: ​​Neuralink

Neuralink has implanted devices in three patients , Musk, its founder, has said. The second patient has shown off capabilities previously demonstrated only in research labs, where wires went deep into participants’ brains and ran directly to external computers. With electrodes implanted seven millimeters into the brain, that Neuralink patient could design software, play videogames and more.

This kind of implant comes with potential trade-offs, says Cajigas. There’s the question of whether, over time, the brain will respond to these electrodes in ways that make them unusable. And then there’s the matter of upgradeability: Once you’ve put electrodes deep into your cortex, it’s not clear how easily you’ll be able to take them out and put in a new model. Neuralink didn’t respond to requests for comment.

The future is wide open

Getting a brain implant might one day become as routine as, say, getting a cochlear implant, which by 2022 had reached a million hearing-impaired patients. If so, the ability to directly interface with our brains could be one of the most transformative medical, and potentially consumer, technologies in history.

Experts I interviewed described various potential uses for brain-computer interfaces: figuring out which medication works best for our particular brain chemistry; using just thoughts to control vehicles, limbs and exoskeletons; and generating speech directly from thought.

Getting there requires vaulting over one other hurdle that has nothing to do with science: These startups have to become real businesses, says Justin Sanchez, former head of brain-implant research at the Pentagon’s R&D arm, the Defense Advanced Research Projects Agency.

It’s possible one could one day become a medical-device giant in its own right. But most of these companies are likely to run out of money or get acquired by big medical-technology companies, first. Whatever happens, brain-computer interfaces have advanced far enough that experts agree they can already give doctors new ways to improve patients’ lives, and are likely to show up in many more of our heads in the future.

Write to Christopher Mims at [email protected]