Interview

Neuralink co-founder DJ Seo: 7 human participants, next stop is speech, vision, and inner thoughts

Jun 27, 2025 with DJ Seo

Key Points

  • Neuralink has implanted its device in seven human participants and is scaling toward commercial production, with two users already controlling dual joysticks in first-person shooters simultaneously.
  • The company is expanding beyond motor control into vision restoration through Blindsight, targeting the first patient next year, and eventually decoding inner thoughts.
  • Neuralink frames itself as an engineering company, not neuroscience, hiring across RF design, robotics, and clinical surgery with a selection filter favoring low ego-to-ability ratios.
Neuralink co-founder DJ Seo: 7 human participants, next stop is speech, vision, and inner thoughts

Summary

Neuralink has implanted its device in seven human participants and is declaring the technology reliable enough to begin scaling. DJ Seo, co-founder and president, confirmed the company's near-term roadmap now extends well beyond motor control, targeting speech decoding, restored vision, and eventually capturing the speed of inner thought.

The Device and the Procedure

The implant, branded Telepathy, targets the motor cortex and is designed for patients with traumatic spinal cord injury or ALS. Electrode threads roughly 1/20th the width of a human hair are inserted by a proprietary surgical robot that maps the brain surface in real time to avoid blood vessels. The skull opening created during the craniotomy is sealed by the implant base itself, which houses the battery and onboard computer. The entire procedure currently takes approximately three hours, and the finished implant is fully wireless and invisible beneath the skin.

The robot is not an ancillary tool. Seo describes it as central to the company's thesis from day one: scaling to millions, and eventually billions, of users is incompatible with manual neurosurgery.

Performance and AI-Assisted Decoding

Two participants have now used Neuralink to play a first-person shooter, controlling dual joysticks and multiple button inputs simultaneously. Seo notes the company is already observing early signs of superhuman performance capabilities, though he stops short of specifics. The neural decoding pipeline currently uses relatively simple machine learning, but benefits from a co-adaptation loop where both the silicon model and the user's own brain are continuously learning. With seven participants now generating data, Neuralink is identifying cross-patient neural pattern similarities that could meaningfully reduce individual calibration time.

Signal quality remains a hardware-first problem. Seo is direct that AI denoising has limited applicability at the electrode level given tight signal-to-noise constraints; the company's edge comes from proprietary low-power, low-noise amplifier designs rather than software correction.

Next Products

Blindsight, Neuralink's vision restoration product, works by converting camera input from glasses into electrical stimulation of the visual cortex. Seo confirmed the first Blindsight patient is planned for next year. The product represents a meaningful architectural shift from output-only (motor intent) to input-driven stimulation.

Looking further out, Seo frames Neuralink as a platform analogous to the iPhone at launch, expecting a third-party application layer to develop on top of the device. He explicitly envisions a Neuralink app store providing a bidirectional conduit between human intent and external machines, including computers, phones, and prosthetics.

Scale and Hiring

Neuralink currently employs 300 people across Austin and Fremont. Seo characterises it emphatically as an engineering company, not a science one, and says neuroscience background is not a prerequisite. The hiring filter the company applies is an ego-to-ability ratio significantly below 1.0, a heuristic designed to sustain interdisciplinary collaboration across RF chip design, robotics, neurosurgery, animal care, regulatory affairs, and clinical operations simultaneously. Open roles span UIUX, firmware, ASIC design, machining, and clinical surgery.

Foundational Risk

Seo acknowledges the field of brain-computer interface was built on 20 to 30 years of Department of Defense and academic funding, work he says no venture capitalist would have underwritten at the time. He expresses direct concern about what reduced NIH and public research funding means for the next generation of BCI indications, framing it as a systemic risk to the pipeline that eventually feeds companies like Neuralink.