A Short History of Neuralink — Ending With “I Love You”

Neuralink R1 surgical robot on stage beside presenter at a 2020 Neuralink event
Photo: Steve Jurvetson / Flickr (CC BY 2.0). Neuralink R1 surgical robot at a 2020 company event.

Brain–computer interfaces used to live mostly in labs and science fiction. Over the last decade, Neuralink has pushed one version of that idea into public demos, animal studies, and then human clinical trials. The company moves fast, talks loudly, and publishes updates that mix engineering progress with aspirational product names. That mix makes it easy to confuse a livestream with a peer-reviewed result, or an FDA pathway with a green light to sell a device.

This timeline is a chronological record of what Neuralink has actually done in public, from its 2016 founding through September 2026 — twelve milestones with primary sources and the company’s own video where it exists. It is not a product review, not investment advice, and not a claim that the technology is ready for general medical use.

A few ground rules up front. FDA Breakthrough Device designation speeds regulatory dialogue; it is not approval that a product is safe or effective. Company demos and blog posts are useful primary sources, but many performance numbers (bits per second, hours of use, “world records”) are company-reported and not independently replicated in peer-reviewed journals. The N1 implant and related trials remain investigational. Patient names that appear here — Noland Arbaugh, Alex, Brad, Nick, Kenneth Shock, and others — are already public in Neuralink or press materials; the tone is respectful, not promotional.

Milestone index (2016–2026)

Quick map — not a comparison table.

  • 2016–2017: Neuralink is founded and steps into public view
  • 2019: First public reveal of the Link platform and surgical robot
  • 2020: FDA Breakthrough Device tag and the live pig demo
  • 2021: Monkey MindPong — wireless BCI gameplay
  • 2023: FDA clears Neuralink’s first-in-human clinical study
  • Jan 2024: First human implant — Noland Arbaugh joins PRIME
  • May 2024: 100 days in — records, daily use, and thread retraction
  • Jul–Aug 2024: Second participant Alex — Counter-Strike and CAD by thought
  • Sep–Nov 2024: Blindsight Breakthrough, Brad with ALS, Canada and CONVOY
  • 2025: Speech Breakthrough, Series E, and trials across countries
  • Jan 2026: Two years of Telepathy — 21 participants worldwide
  • Sep 2026: VOICE trial — Kenneth Shock’s “I love you” speech demo

2016–2017: Neuralink is founded and steps into public view

Neuralink Corp. was founded in 2016 as a California medical-research company aimed at implantable brain–computer interfaces. Exact day-level filings often get rounded in secondary summaries; the widely reported founding date is June 21, 2016, with medium confidence on the day itself. What is solid is the public debut: in late March 2017, outlets including the Wall Street Journal and The Verge reported that Elon Musk and a founding team of scientists and engineers were building a company to connect brains and computers.

On April 20, 2017, Tim Urban’s long explainer on Wait But Why became the early vision document many people still cite. Near-term framing focused on treating serious neurological disease; longer-term messaging floated human–AI symbiosis and “neural lace” language that was aspirational, not a clinical result. There was no official Neuralink YouTube channel demo from that founding window — contemporary news coverage is the record we have.

2019: First public reveal of the Link platform and surgical robot

July 16, 2019 was Neuralink’s first major hardware coming-out. At a launch event, the company described a high-channel flexible-thread brain–machine interface, a neurosurgical robot designed to insert those threads with vessel avoidance, and a path toward a wireless, fully implanted device. Those architectural choices — thin threads, robot insertion, wireless implant — are the roots of later clinical systems branded around the N1 implant and R1 robot.

A technical preprint followed on bioRxiv, with a peer-reviewed version later in JMIR. Early paper channel counts should not be frozen as “the” final design: clinical-generation N1 systems are typically described around roughly 1,024 electrodes on flexible threads, and the company has continued to iterate electrode density, retention, and surgical approach. The official “Introducing Neuralink” video remains the cleanest public snapshot of that 2019 reveal.

2020: FDA Breakthrough Device tag and the live pig demo

In mid-2020 — reported as July 2020 at month-level precision, without a day-stamped public FDA press letter in open sources — Neuralink’s implant system received FDA Breakthrough Device designation. That pathway is meant to speed interaction with the agency for devices that may offer more effective treatment for life-threatening or irreversibly debilitating conditions. It is not clearance to begin human trials, and it is not commercial marketing approval.

On August 28, 2020, Neuralink livestreamed a progress update featuring pigs with wireless Links, including Gertrude, with real-time cortical spike playback audible and visible to viewers. Coverage from The Guardian and CNBC captured both the spectacle and the caveats. Later reporting, including a Reuters special report, placed the Breakthrough step in a longer FDA back-and-forth that would include a rejected human-trial bid before clearance arrived in 2023. The company’s own Summer 2020 update video is the primary demo record.

2021: Monkey MindPong — wireless BCI gameplay

On April 8, 2021, Neuralink released “Monkey MindPong.” A macaque named Pager, with Links in both motor cortices, played Pong using decoded neural activity after an earlier period of joystick training. The practical claim was less “monkeys can control cursors” — academic BCIs had shown related behaviors for years — and more the wireless, fully implanted sensing package as a form factor aimed at future human cursor control.

BBC News and TechCrunch covered the release the same week; independent experts quoted in that coverage noted that peer-reviewed quantitative comparison to prior BCI literature was limited at the time. The official video is still the best primary artifact for what the company claimed that day.

2023: FDA clears Neuralink’s first-in-human clinical study

On May 25, 2023, Neuralink announced that the FDA had cleared it to begin a first-in-human clinical study of its implant and surgical robot. Headlines often said “approval”; more precise language is FDA clearance to begin an investigational human study under the IDE pathway — still not a commercial marketing approval. Reuters and CNBC reported the announcement that day.

Context matters: Reuters had previously reported that the FDA rejected an earlier human-trial request in 2022, citing numerous safety questions. On May 25, 2023 the company said it was not yet recruiting; the PRIME Study opened recruitment later in September 2023 after hospital and IRB approvals. The trial is listed on ClinicalTrials.gov (NCT06429735). Neuralink’s clinical-trial explainer video frames the company’s own pitch for what PRIME was meant to test.

January 2024: First human implant — Noland Arbaugh joins PRIME

On January 28, 2024, at Barrow Neurological Institute in Phoenix, Noland Arbaugh became the first PRIME Study participant to receive the wireless Link — framed publicly under the product name Telepathy. Arbaugh has quadriplegia after a spinal cord injury. Surgery was reported as successful; he was discharged the next day. Early public statements cited promising neuron spike detection.

Reuters covered Musk’s January 29 statement. Later official narratives appear in Neuralink’s 100-day user-experience update and A Year of Telepathy, with additional patient interview context in WIRED. The company’s “P1 Plays Chess” video became an early public face of what that first implant looked like in practice.

May 2024: 100 days in — records, daily use, and thread retraction

By spring 2024, Arbaugh was using Telepathy for independent computer control — chess, strategy games, Mario Kart, browsing, livestreaming. Neuralink’s May 8, 2024 user-experience update described first-session BCI cursor rates of 4.6 bits per second, later 8.0 BPS, framed against prior published BCI literature the company cited. Those figures are company-reported; independent replication in peer-reviewed journals remained limited.

The same update disclosed a hard engineering fact: some electrode threads had retracted, reducing effective channel count. Software, decoder, and interface changes restored and then exceeded prior performance by the company’s account. Independent use often ran many hours a day. That combination — real capability, real hardware degradation, and software recovery — is more informative than a highlight reel alone. Recruitment and assistive-technology framing continued in company videos from that period.

July–August 2024: Second participant Alex — Counter-Strike and CAD by thought

Alex, the second PRIME participant (also living with spinal cord injury / quadriplegia), received a Link in July 2024 at Barrow and was discharged the next day. Neuralink’s August 21, 2024 second-participant update described out-of-box cursor control within minutes, Counter-Strike played with Link plus a Quadstick, and Fusion 360 CAD work by day two — including a custom charger mount that was 3D-printed. Reuters reported the second implant in early August.

Importantly for the thread-retraction story, Neuralink reported surgical mitigations (including reducing brain-motion gap) and said Alex showed no thread retraction. Exact calendar day of surgery is stated only as “last month” relative to the August 21 blog — month-level precision is what the primary source supports. The official Fusion 360 design clip is the clearest public demo of that second-participant beat.

September–November 2024: Blindsight Breakthrough, Brad with ALS, Canada and CONVOY

Fall 2024 was less a single demo and more a cluster of expansions. On September 17, 2024, Neuralink announced FDA Breakthrough Device designation for Blindsight, a vision prosthesis concept. Neuralink’s post and Reuters covered the designation; IEEE Spectrum reminded readers that Breakthrough is not proof of effectiveness. As of late September 2026 research for this article, Neuralink’s primary updates index did not confirm a completed Blindsight human implant — treat human efficacy as prospective, not demonstrated.

Around November 2024, Brad — the first ALS participant in PRIME, nonverbal — received a Link and used Telepathy for outdoor typing and communication beyond eye-tracker limits, later including AI-cloned voice demos in 2025 press materials. Health Canada’s CAN-PRIME approval was announced November 20, 2024 (CAN-PRIME launch). CONVOY, a feasibility trial extending N1 control to assistive robotic arms and cross-enrolling PRIME participants, was announced with a November 24, 2024 release date (CONVOY; NCT06710626). There is no official Neuralink Blindsight demo video from the designation itself; the best reputable long-form for the Brad/ALS beat is the Core Memory piece below, with additional Brad narrative in A Year of Telepathy.

2025: Speech Breakthrough, Series E, and trials across countries

2025 stretched the clinical footprint. On February 5, Neuralink’s A Year of Telepathy summarized three pioneers, claiming 670-plus implant-days and 4,900-plus Telepathy hours — company figures. On May 1, the company received FDA Breakthrough Device designation for a speech restoration program (again: designation ≠ proven efficacy). On June 2 it announced a $650 million Series E and said five people were then using Neuralink, with trial sites named across the US, Canada, and UAE — a snapshot later superseded by higher counts.

On July 31, 2025, GB-PRIME launched in Great Britain after MHRA/HRA approvals, with sites including UCLH and Newcastle. Robotic-arm work progressed alongside digital cursor control; participants such as Nick featured in company clips gesturing with an assistive arm by thought. The device remained investigational throughout. The robotic-arm video below captures the physical-autonomy thread that CONVOY-era work was aiming at.

January 2026: Two years of Telepathy — 21 participants worldwide

On January 28, 2026 — two years after the first human implant — Neuralink published Two Years of Telepathy, announcing 21 trial participants enrolled worldwide. Reuters reported the same figure that day, noting the company had previously said 12 people had chips as of September 2025. Prefer “at least 21 as of January 2026” unless a later primary Neuralink update supersedes it; secondary trackers claiming higher totals after that date were not company-confirmed on Neuralink’s public updates index through late September 2026 research.

Company highlights in that post included education (Noland back in college; Sebastian in medical school with claimed BCI use up to about 17 hours a day), art (Audrey), ten-finger imagined typing up to about 40 words per minute (Jake), robotic-arm daily tasks (Nick), and cursor control claimed to approach or exceed able-bodied mouse bits-per-second for some users — again, company-reported. Neuralink claimed zero serious device-related adverse events as of that update. A VOICE speech trial launched (NCT07224256). The roadmap language pointed toward denser implants (around 3,000 electrodes discussed) and less-invasive transdural insertion research. An implantation timeline graphic in the post sketched growth across 2024–2025 halves (1 + 2 in 2024; 4 + 13 across 2025). Hardware and robot generations continue to iterate — avoid treating any one design as final.

September 2026: VOICE trial — Kenneth Shock’s “I love you” speech demo

In mid-September 2026, Neuralink shared a short company clip captioned “A beautiful surprise” (~September 16), showing ALS participant Kenneth Shock — enrolled in the investigational VOICE speech trial (NCT07224256) — use his N1 implant so that decoded intended speech produced audible words for his wife. Shock received the implant around January 2026 at UT Southwestern Medical Center in Dallas; Neuralink has described him among early speech-restoration participants. In the widely circulated moment, a reconstructed, pre-ALS-like voice said “I love you.”

Treat that clip as a company-reported demo, not peer-reviewed efficacy data — the same caveat that applies to earlier Telepathy livestreams. Reputable coverage such as TechTimes summarized the post; Neuralink’s own materials continue to frame speech work under its May 2025 Breakthrough Device designation for speech and the broader speech-restoration program. Breakthrough still ≠ FDA approval, and the company states that devices remain investigational and not commercially available.

Looking ahead — caveats, and the next decade

A decade of public milestones is real — and none of it equals general clinical availability. Devices remain investigational; Breakthrough tags for Link, Blindsight, and speech are process tools, not proof of safety or effectiveness; Blindsight human efficacy was not confirmed in Neuralink primaries through late September 2026; and the VOICE “I love you” clip is best read as a company-reported demo. Durability, independent performance checks, less-invasive surgery without signal loss, and animal-welfare questions covered in investigative reporting all remain open. The useful habit is still: separate the livestream from the peer-reviewed claim.

If the next ~10 years follow the direction already sketched in company roadmaps and early trial work — denser implants, more refined surgical robots, and less-invasive approaches such as transdural insertion research — speech and vision programs could mature from designation and early demos into clearer human data. More participants, across more sites, would make everyday communication, cursor control, and assistive-arm use less rare, and the evidence base harder to dismiss as highlight-reel only. That path is a plausible direction, not a guarantee; timelines in this field have a habit of stretching.

What already feels less speculative is the human through-line. A reconstructed “I love you,” a student back in class, someone designing with thought alone — these moments are still early and still uncertain in how broadly they will scale. They also point toward a quieter possibility: that brain–computer interfaces might, for some people living with paralysis or progressive disease, restore speech, partial vision, and a measure of independence that felt out of reach a decade ago. That is a hope worth holding carefully — and a record worth updating as the next ten years unfold.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top
Aglena