In 19 surgeries on rats, the startup successfully placed the threads 87% of the time, the paper said. Musk and his team said the technique has promise to capture more neural data more safely than existing approaches.
Neuralink will face a long list of formidable obstacles as it aims to move from rats to the clinic. Decoding the mysteries of the brain is a steeper challenge than manufacturing cars or drilling tunnels or even building rockets. Even peer-reviewed and published data in rats usually disappoints when it gets translated to human patients — and Neuralink doesn’t even have that yet. And there are plenty of reasons that neuroscientists who have worked on brain-machine interfaces for years have yet to make nearly any commercial traction.
Musk, whose eccentricities and outlandish Twitter presence have helped him build a cult following, put that offbeat personality on full display on Tuesday night. Donning a black jacket, a white button-up shirt, and no tie, he mused at one point about his technology giving people the option of merging with AI. “We are a brain in a vat,” he remarked at another point. Later, he spoke dreamily about communicating telepathically.
Asked about Neuralink’s animal research by a member of the audience, Musk turned earnest for a moment in speaking about the startup’s seriousness in caring for the rats it uses in its research — before making a joke that didn’t quite land about “karmic payback” for the Black Plague.
Neuralink is also conducting research in monkeys at the University of California, Davis, Musk said. Then he spilled the goods: “A monkey has been able to control the computer with its brain,” Musk said. (At no point has the company provided evidence to support that assertion.)
Musk had clearly gone off-script. Max Hodak, the company’s president up on stage beside him, seemed a bit rattled. “I didn’t realize we were running that result today, but there it goes,” Hodak said.
As for its planned clinical study, Neuralink said it would focus on enrolling patients with paralysis of all four limbs due to a spinal cord injury. No details were shared about the intended size of study, nor the specific endpoints it would aim to evaluate. Musk did, however, say offhandedly that one of the company’s goals is to allow patients who are paraplegic to use their thoughts to type at a rate of 40 words per minute.
Asked about the company’s FDA pathway, Hodak said Neuralink would attempt to pursue an early feasibility study, under what’s known as an investigational device exemption.
Ryan Stellar, vice president of product management at Enzyme, a startup working on software to help life sciences companies with the regulatory process, told STAT that he expects technology as ambitious as Neuralink’s will demand rigorous, long-term study.
“Unless the FDA is blinded by Elon’s star power, a premarket trial of significant size” — 100-1,000 people — “will probably be needed, with a minimum observational period of two years, but perhaps as much as seven,” Stellar speculated.
At Tuesday night’s event, Musk trotted out several key members of the team of scientists and executives he’s assembled to carry out the company’s ambitions. Among them were the company’s senior scientist, Philip Sabes, who was previously a full-time professor at the University of California, San Francisco, researching how the brain processes movement. Another was the company’s head neurosurgeon, Dr. Matthew MacDougall, who delivered his part of the presentation in blue scrubs.
Musk said the event’s primary purpose was to recruit talent to the San Francisco-based startup, which has a headcount near 100. Neuralink’s website lists a handful of job postings, including for an accountant and a software engineer who can build robots.
Neuralink, which was incorporated in 2016, has brought in $158 million in funding, $100 million of it from Musk himself. In a filing with the Securities and Exchange Commission this past May, the company said that it had brought in $39 million of an anticipated $51 million funding round.
Neuralink’s big reveal, which was also live-streamed online, had an atmosphere unlike that of a typical corporate update. Attendees sipped red wine and posed for selfies. After the presentation, many of them crowded around a display case filled with several prototypes of Neuralink’s technology, jostling to take a perfectly framed smartphone photo. The scene was reminiscent of tourists in a museum shoving their way to the best position to snap a photo of a precious treasure from antiquity.
Neuralink’s event was invitation-only — interested attendees had to fill out an online form making a case for why they should get a golden ticket — and those who received an emailed invitation were asked, “for security reasons,” to refrain from publicly sharing the location of the event and to avoid bringing “any bags larger than a purse.”
Until Tuesday night, Neuralink had disclosed little about its research, despite big promises from Musk. In a September 2018 appearance on the comedian Joe Rogan’s podcast, Musk smoked pot, drank whiskey, and posited that brain-machine interface technology “will enable anyone who wants to have superhuman cognition.”
In a paper posted to a preprint server in March, a team of researchers affiliated with Neuralink described a technique, which they likened to a “sewing machine,” similar to what was unveiled on Tuesday night.
Neuralink has the highest profile among a number of startups working on brain-machine interfaces. A few of the buzziest include Kernel, founded by the 41-year-old tech entrepreneur and venture capitalist Bryan Johnson, and Paradromics, which is working on an implantable chip designed to record and stimulate electrical activity in the brain. DARPA, the U.S. Department of Defense agency focused on futuristic research projects, has its own initiative that’s funded research with an eye toward the development of brain-machine interfaces that could be used by members of the military.
Despite the excitement around brain-machine interfaces, the field has made little commercial progress since the development of the first prototypes more than a decade ago.
The first applications of the technology focused on movement. They aimed to read electrical signals in the motor cortex corresponding to the intention to move — and then used software to try to translate those signals into instructions to operate a computer cursor or robotic arm.
Meanwhile, decoding brain signals into speech has become the next frontier in the field.