Science Corporation

Science Corporation is building neurotechnology with an unusually explicit thesis: if electrodes don’t scale because they damage tissue, try scaling with biology.

They are best known publicly for two efforts:

1. PRIMA, a subretinal photovoltaic visual prosthesis program in clinical studies.
2. A biohybrid brain interface concept where living neurons are embedded in a device and then engrafted so that their axons/dendrites grow into the brain, potentially creating enormous connection density without “putting wires into a brain.”

Company background

• Website: ⁰¹
• Company overview: ⁰²
• HQ (company statement): Alameda, California.
  • Science says it has secondary offices in Research Triangle Park (NC) and Paris (France).
  • ⁰²
• Clinical posture: Science describes itself as a clinical-stage medical technology company with “dozens of patients in six countries” enrolled in ongoing clinical studies. ⁰²

• Leadership (Sep 17, 2026): Darius Shahida, previously Chief Strategy Officer, was named President. Source: ⁰³

• Pixium Vision, mentioned in older coverage, is now part of Science (per Science press materials).

The big idea: biohybrid neural interfaces (neurons as the interface)

The scaling bottleneck they’re attacking

Science’s biohybrid writeup frames a core constraint in invasive BCI as a scaling law: to get physically close to neurons, you destroy some brain tissue, and that becomes an ugly trade when you try to scale to very large numbers of “channels.”

• “Biohybrid neural interfaces” essay: ⁰⁴

What they mean by “biohybrid”

In Science’s version, the cell/device interface is created in vitro:

• Stem cell–derived neurons are engineered and embedded in an electronic device in the lab.

• The device is implanted so that only the processes (axons/dendrites) grow out into the brain.

• The cell bodies stay with the device, which they argue helps with SNR, power, and enables new types of stimulation concepts (e.g., neuron types / synaptic chemistry).

• Science’s essay (high-level concept): ⁰⁴

“Channels” vs synapses (why they think this could be different)

Science pushes a reframing: don’t count electrodes; count potential synapses.

On their biohybrid tech page, they argue (conceptually) that:

• neurons are denser than electrodes

• each neuron can form many connections

• “a million embedded neurons could form a billion synapses” with brain tissue

• Science biohybrid technology page: ⁰⁵

They cite a scaling dataset as context:

• Stevenson Lab scaling reference linked by Science: ⁰⁶

Proof-of-concept: optogenetic stimulation via an engrafted cortical “waffle” (mouse study)

Science released an early feasibility result showing that a cortical biohybrid implant can transmit information into the brain strongly enough to shape behavior.

Core design (as described in both Science’s post and IEEE Spectrum’s coverage):

• a waffle-like scaffold of microwells placed on the cortical surface
• neurons are loaded into microwells (aiming for ~one neuron per well)
• neurons survive and extend into superficial cortex
• optical stimulation (“turning on” graft neurons) can be used as a cue that mice learn to report

Primary source:

• Zappitelli et al. (bioRxiv, 2024): “Optogenetic stimulation of a cortical biohybrid implant guides goal directed behavior.”
  • DOI: ⁰⁷
  • Full text: ⁰⁸

Independent coverage with helpful engineering details:

• IEEE Spectrum (Jun 4, 2025): “Science Corp’s Biohybrid BCI Adds Neurons to the Brain.” ⁰⁹

The hard problems (they openly acknowledge)

Science’s own essay is refreshingly explicit that biohybrid BCIs are low “technology readiness” and require solving tough bioengineering:

• immune compatibility (autologous vs allogeneic, “hypoimmunogenic” lines)

• keeping graft cells alive through hypoxia, glycemic shock, and host responses

• device design that supports neuron survival (including glial support)

• Science essay (technical challenges): ⁰⁴

What else they’re building: PRIMA visual prosthesis (clinical program)

Science’s most advanced clinical program appears to be PRIMA, a subretinal implant intended to stimulate remaining retinal circuitry when photoreceptors are lost in advanced retinal degeneration.

Science describes PRIMA as:

• a light-powered subretinal implant with 378 pixels

• paired with glasses containing an infrared projector + camera that provide both power and data via patterned IR light

• being evaluated in clinical studies (Europe and US trials listed)

• PRIMA overview page: ¹⁰

The PRIMA page also links peer-reviewed and trial references (e.g., PubMed-linked items and NCT IDs).

• PRIMA in Europe (Jul 22, 2026): PRIMA received a CE mark and Science announced its European commercial launch for geographic atrophy from age-related macular degeneration; US FDA approval process is underway and PRIMA holds two FDA Humanitarian Use Device designations. Sources: ¹¹ and ¹²

• PRIMA 36-month results (Oct 5, 2026, company press release): Science reported 36-month follow-up from the PRIMAvera trial (NCT04676854), presented at the 26th EURETINA Congress in Vienna by Mahi Muqit (Moorfields Eye Hospital). Per the release: 38 patients implanted at 17 sites in five European countries; 27 evaluable for safety and 18 for efficacy at 36 months; mean gain of 29 ETDRS letters (24.5 at 12 months); 14 of 18 (78%) gained at least 15 letters; no new safety signal between months 12 and 36. The release states PRIMA is commercially available in Europe and the UK and pending FDA approval. The 36-month averages cover 18 of 38 implanted patients, and Inside BCI notes the data are not yet peer reviewed and that Science has not said which glasses setting the 36-month figure uses. Sources: ¹³ (company) and ¹⁴ (secondary)

My read (why it’s exciting)

Science’s biohybrid pitch is one of the few that feels like it’s attacking the geometry and biology of the interface, not just the electronics:

• If you can get stable, high-density connectivity via living processes, you potentially change the scaling laws.
• It also makes the interface problem less like “a bundle of wires” and more like “a grafted microcircuit,” which could open novel stimulation/learning paradigms.

But the credibility hinge is the same as always: can they make it reliable, safe, manufacturable, and ultimately clinically supportable?

first Stargardt patient

• First Stargardt patient (Oct 13, 2026, company update): Science reported that its first patient with Stargardt disease received the PRIMA implant in May 2026 in the PRIMAlia trial (NCT07266584) at Sydney Eye Hospital under Dr. Matthew Simunovic. Per the company, during his first rehabilitation session he saw a letter, and during his second session he read a word (single patient, company-reported, not peer reviewed). TechCrunch covered the announcement the same day from TechCrunch Disrupt, where CEO Max Hodak said a next-generation PRIMA implant is being worked on that he expects “in humans in the next year” (his statement). Sources: ¹⁵ (company) and ¹⁶ (secondary)