Neuralink: revision 2 to 3 — Future Human Wiki

Revision 2

Revision 3

horizon: "late 2020s"

categories: ["organizations", "cybernetics"]

tags: ["bci", "neurotechnology", "implants", "elon musk", "clinical trials", "paralysis"]

summary: "American neurotechnology company founded in 2016 developing a wireless implanted brain–computer interface with flexible electrode threads inserted by a surgical robot."

summary: "American neurotechnology company founded in 2016 developing a wireless implanted brain-computer interface with flexible electrode threads inserted by a surgical robot."

updated: "2026-07-27"

updated: "2026-10-29"

issues: ["Series E size and valuation are attributed to the company with no citation.", "Blindsight and CONVOY paragraphs rest on company statements not cited as footnotes."]

issues: ["Series E size and valuation are attributed to the company with no citation.", "Blindsight and CONVOY paragraphs rest on company statements not cited as footnotes."]

---

The company began implanting the device in humans in January 2024 under an FDA investigational

device exemption.

Neuralink's technical contribution is largely in engineering and surgery rather than neuroscience:

Neuralink's technical contribution lies largely in engineering and surgery rather than

flexible electrode threads that move with the brain, automated insertion, high channel counts in a

neuroscience: flexible electrode threads that move with the brain, automated insertion, high

sealed wireless package, and a hermetic implant with no percutaneous connector. Its scientific

channel counts in a sealed wireless package, and a hermetic implant with no percutaneous connector.

results to date have been communicated mainly through livestreams and company blog posts rather

Its scientific results to date have been communicated mainly through livestreams and company blog

than peer-reviewed publication, which is the most common criticism levelled at it by academic BCI

posts rather than peer-reviewed publication, the criticism academic BCI researchers most often

researchers.

level at it.

## Technology

The N1 implant houses a battery, wireless radio, and signal-processing electronics in a package

that sits in a recess milled into the skull, flush with the surrounding bone. From it extend 64

polymer threads bearing 16 electrodes each — 1,024 channels in total — each thread narrower than a

polymer threads bearing 16 electrodes each, 1,024 channels in total. Each thread is narrower than

human hair and far more compliant than the rigid silicon shanks of a [[utah-array]]. The design

a human hair and far more compliant than the rigid silicon shanks of a [[utah-array]]. The design

intent is to reduce the micromotion-driven tissue damage that drives glial encapsulation and signal

is meant to reduce the micromotion-driven tissue damage that causes glial encapsulation and signal

loss in stiff arrays.

Because the threads are too floppy to insert by hand, the R1 robot uses a needle and optical

guidance to place each one individually, steering around surface vasculature. The company described

the architecture in a 2019 white paper based on rodent work.[^musk2019] Recording, spike detection,

and compression happen on the implant; decoded output is sent over Bluetooth to an external device.

Users control a cursor by attempting or imagining movement, with the mapping learned by a decoder

Users control a cursor by attempting or imagining movement, and a periodically recalibrated

that is periodically recalibrated — the same basic paradigm of [[neural-decoding]] established in

decoder learns the mapping. This is the same basic paradigm of [[neural-decoding]] established in

academic laboratories two decades earlier.[^hochberg2006][^gilja2012]

The device is read-only in its current form. It records but does not stimulate, which places it in

The device is read-only in its current form. It records but does not stimulate, which puts it in a

a different category from the implanted neurotechnologies already in routine clinical use, such as

different category from the implanted neurotechnologies already in routine clinical use, such as

the [[cochlear-implant]] and [[deep-brain-stimulation]], which write signals in rather than

reading them out. Adding a stimulation channel is a stated goal and is required for the visual

programme, but the safety profile of chronic cortical stimulation is less well characterized than

  { "year": "2021", "title": "Macaque cursor control", "text": "A monkey plays a cursor-based video game using the implant, replicating a capability shown in primate laboratories years earlier." },

  { "year": "2021", "title": "Macaque cursor control", "text": "A monkey plays a cursor-based video game using the implant, replicating a capability shown in primate laboratories years earlier." },

  { "year": "2023", "title": "Human trial cleared", "text": "The FDA grants an investigational device exemption; the PRIME study opens recruitment for people with quadriplegia." },

  { "year": "2023", "title": "Human trial cleared", "text": "The FDA grants an investigational device exemption; the PRIME study opens recruitment for people with quadriplegia." },

  { "year": "2024", "title": "First human implants", "text": "The first participant receives the device in January; the company later discloses that many threads retracted from the cortex, and modifies its technique for subsequent surgeries." },

  { "year": "2024", "title": "First human implants", "text": "The first participant receives the device in January; the company later discloses that many threads retracted from the cortex, and modifies its technique for subsequent surgeries." },

  { "year": "2024–2026", "title": "International expansion", "text": "Regulators in Canada, the United Kingdom, and the United Arab Emirates permit studies; the company reports additional participants and a robotic-arm extension study." }

  { "year": "2024-2026", "title": "International expansion", "text": "Regulators in Canada, the United Kingdom, and the United Arab Emirates permit studies; the company reports additional participants and a robotic-arm extension study." }

]

```

several months later and reported recovering performance through changes to signal processing and

decoding rather than reimplantation.[^nlink2024] For the second participant the company said it

altered its technique, sculpting the skull surface to reduce the gap under the implant and

inserting threads deeper. Thread retraction is a specific instance of the general problem that

inserting threads deeper. Thread retraction is one instance of a general problem for every

afflicts every penetrating array: the brain moves, pulses, and reacts to foreign material.

penetrating array, which is that the brain moves, pulses, and reacts to foreign material.

A companion study, CONVOY, extends control from a cursor to an assistive robotic arm, moving the

device from computer access toward the broader ambitions of [[neuroprosthetics]]. The company has

programme, Blindsight, aims to restore a crude visual percept by stimulating visual cortex directly;

it received an FDA breakthrough device designation in 2024 but had no published human results as of

early 2026. Cortical visual prostheses have a long and largely disappointing history, and the

resolution limits that constrained earlier efforts — described under [[retinal-implant]] — apply

resolution limits that constrained earlier efforts, described under [[retinal-implant]], apply

with equal force here.

> [!caution] Company reporting, not trial reporting

Neuralink is funded by venture capital and by Musk personally. It raised a Series E in 2025 that

the company said totalled several hundred million dollars at a valuation in the billions, a figure

far above what any BCI company has justified on revenue, since none has an approved product. The

capital advantage is real: implant development, robot engineering, animal work, and multi-country

capital advantage is real. Implant development, robot engineering, animal work, and multi-country

regulatory submissions are expensive in a way academic consortia such as BrainGate cannot match.

regulatory submissions cost more than academic consortia such as BrainGate can spend.

The founding team has largely dispersed. Several co-founders left within the first few years, and

one went on to co-found Precision Neuroscience, which pursues a thin-film surface approach closer

Assessments divide along a consistent line. Engineers credit the company with solving practical

problems the field had left unsolved: a fully implanted wireless device with no skull-mounted

pedestal, an insertion robot that makes hundreds of precise placements, and a manufacturing process

for polymer electrode arrays. Those are the bottlenecks between a laboratory demonstration and a

for polymer electrode arrays. Those problems stand between a laboratory demonstration and a

product.

Neuroscientists are more sceptical of the framing. The cursor control demonstrated in 2024 was

qualitatively similar to results published from academic implants more than a decade earlier, and

the product name "Telepathy" describes typing with a decoder, not thought transmission — an

the product name "Telepathy" describes typing with a decoder rather than thought transmission, an

overstatement of the kind examined under [[brain-to-brain-interface]]. Claims that the device will

one day enable symbiosis with artificial intelligence, discussed as [[human-ai-merger]], are

untested and rest on a bandwidth argument that the low-dimensional structure of cortical activity

Neuralink has denied wrongdoing and published its own account of its animal-care programme.

A third strand of criticism concerns data. A device that streams cortical activity to a phone

creates a record of neural signals held by a private company, raising the questions taken up under

creates a record of neural signals held by a private company, which raises the questions taken up

[[mental-privacy]] — not because a cursor decoder reads thoughts, but because the underlying

under [[mental-privacy]]. A cursor decoder does not read thoughts, but the underlying recordings

recordings support inferences the user never intended to make available.

support inferences the user never intended to make available.

## Outlook

The questions that will determine whether Neuralink matters are unglamorous. Do threads stay put

The questions that will determine whether Neuralink matters are practical ones: whether threads

over five years? What fraction of participants develop infection or require explant? Can a decoder

stay put over five years, what fraction of participants develop infection or require explant,

run for weeks without a technician? Does the device work in a home, unsupervised, on a bad day? The

whether a decoder can run for weeks without a technician, and whether the device works in a home,

company's willingness to run multi-country studies suggests a push toward a pivotal trial, but no

unsupervised, on a bad day. The company's willingness to run multi-country studies suggests a push

implanted BCI from any developer has yet been approved for routine clinical use, and the regulatory

toward a pivotal trial, but no implanted BCI from any developer has yet been approved for routine

path for a permanent cortical implant in a non-fatal condition has no precedent to follow.

clinical use, and the regulatory path for a permanent cortical implant in a non-fatal condition has

no precedent to follow.

A separate question is what happens if the company stops. Implanted devices create obligations that

outlast product lines, an issue that the emerging [[neurorights]] frameworks have begun to name but

outlast product lines, an issue the emerging [[neurorights]] frameworks have begun to name but have

not yet resolved.

## See also