Science Corporation
Full-stack neural-engineering company; creator of the PRIMA retinal BCI — the first CE-marked brain-computer interface for form vision restoration. Founded 2021 in Alameda, CA by max-hodak (ex-Neuralink) and Alan Mardinly (ex-Neuralink biology director).
Timeline
- 2021: founded; raised >$47M seed.
- Mar 2023: FDA Breakthrough Device designation for PRIMA.
- Apr 25, 2024: acquired pixium-vision’s PRIMA IP + 3 ongoing clinical trials (incl. 38-patient pivotal study NCT04676854) after Paris commercial court approval; licensing/co-development with Prof. Daniel Palanker (Stanford, PRIMA inventor).1
- 2025: $104M convertible note (Khosla-led, April); NEJM publication Oct 20, 2025 (Holz et al. NEJM 2026;394:232-242) 2; acquired MEMS facility in North Carolina (in-house chip/MEMS manufacturing).
- Mar 2026: 489M raised; reported $1.5B valuation.
- Jul 22, 2026: CE mark (EU MDR, DEKRA) + European commercial launch of PRIMA across 30 countries; first commercial implant expected in Germany; reimbursement applications underway; 2× FDA Humanitarian Use Device designations (July 2026), FDA marketing review ongoing.3
PRIMA (the product)
- Subretinal photovoltaic implant: 2 × 2 mm, 30 µm thin, 378 pixels/electrodes; implanted under the atrophic macula; glasses w/ camera + near-infrared projector deliver power and data (“miniature solar panel”); zoom-in for reading.
- Target: geographic atrophy due to AMD (>5M patients worldwide); expanding to Stargardt disease and retinitis pigmentosa; not for wet AMD (inner-retina heterogeneity).
- Trial results (38 patients, 17 sites, 5 EU countries; PRIMAvera NCT04676854):
- 81.3% achieved ≥0.2 logMAR (≥10 ETDRS letter) improvement at 12 months (p<0.001); multiple imputation estimate 79.9% for all 38 enrolled.
- Mean improvement +25.5 letters (logMAR 0.51); best individual gain: 59 letters.
- 84.4% read letters/numbers/words at home; 68% medium-to-high user satisfaction.
- Prosthetic VA (direct stimulation, no camera): logMAR 1.32 ± 0.16 (20/417) — exactly at the theoretical 100 µm pixel sampling limit.
- No change in natural peripheral VA (logMAR 0.00 change).
- 26 SAEs in 19 participants; 81% within 2 months, 95% resolved within 2 months; no life-threatening events; DSMB: benefits outweigh risks.4
- Upgrade pathway: Wireless design allows in-situ chip replacement with higher-resolution devices (Bhuckory et al., Nat Commun, 2025; DOI: 10.1038/s41467-025-58084-y) or tiling multiple modules via minimal incision.
PRIMA system architecture
The chip. 2 × 2 mm, 30 µm-thick crystalline silicon; 378 hexagonal pixels of 100 µm (~7° of visual angle). Each pixel carries two series photodiodes between a central active electrode and a hexagonal circumferential return-electrode mesh that confines the electric field; absorbed 880-nm pulses become charge-balanced biphasic current pulses that stimulate bipolar cells. Pixel engineering (Pixium IP, now Science Corp): a reflective buried-oxide/metal layer plus hermetic titanium/ceramic backside seal turns the array into a light trap so a ≤30-µm-thick implant still generates stimulation-grade photocurrents (US12201827B2; original grant US11197993B2), and a dimensioned shunt resistor sets the rapid-discharge vs charge-loss trade-off for safe charge balancing between pulses (US10980997B2). The 30-µm thickness is itself NIR-driven design: 905-nm light penetrates only ~35 µm into silicon, so the array is thinned to just absorb the illumination pulse.5 6
The glasses (projector). A frame-mounted camera feeds a pocket processor; the processed image is projected onto the implant with pulsed near-infrared 880-nm light at a 30-Hz frame rate (above flicker fusion), with percept brightness set by pulse duration (0.7–9.8 ms, 0.7-ms increments, peak irradiance 3.5 mW/mm²). The projection engine is a DMD (digital micromirror device): a modulation micromirror array divides the pulsed beam into the pixel pattern with individual pulse-width modulation per micromirror, synchronised to the light-source pulses and duty-cycle-limited (≤ 0.5) for ocular and implant safety — the subject of Pixium’s projector patent US12061332.7 8 4
- PRIMA-1 (VR, 2017–2019): opaque video glasses (“digital mirror display”); 5.1 mm (17.5°) projected field, ~10.5 µm optical resolution, 3 mW/mm² peak.
- PRIMA-2 (AR, since Oct 2019): transparent AR design → simultaneous peripheral natural and central prosthetic vision; 5.3 × 4.3 mm (18.5°) field, 6.7 µm resolution, 3.5 mW/mm²; projection adapted to the patient’s refraction; electronic zoom (up to ×12); improved beam homogeneity and alignment; optional tinted lens (~65% white-light attenuation) for bright environments.8 5
Next generation (Science Corp filings, 2024–2026). The successor projector adds a spatial light modulator + steering module + two-stage eye tracking: coarse steering from external eye features (pupil, corneal reflection) and fine pattern correction from retinal features (individual cells, vasculature, or the implant itself) — aiming at cell-resolution mapping of the projected frame onto retinal cells, with multi-wavelength sensing (US20250249280A1; later filing US2026/0166320 continues the line for retinal implants — see visual-prosthesis-patents).
Portfolio beyond PRIMA
- Optogenetics / Lightsheet — RGC-targeted gene therapy approach (PRIMA targets bipolar cells); Hodak: “too early to know which approach… will work best.”
- Biohybrid interfaces — living neurons connected to electronics.
- Vessel — advanced perfusion circuits for organ transplantation/critical care.
Strategy & framing
Disease-first BCI: pick a specific disease with a clear patient population (GA/AMD >5M), controllable surgical risk (subretinal), and simple-but-real structure (the retina as an extension of the brain) — the fastest path to regulatory approval and revenue, versus the “full brain interface” ambition of neuralink. See max-hodak for the founder’s account. PRIMA was also “the first really definitive demonstration” of form vision in humans — earlier devices produced only phosphenes (Hodak, Apr 2024).1
Related pages
- max-hodak — founder/CEO story
- pixium-vision — PRIMA’s EU originator, acquired 2024
- retinal-prosthesis, retinal-prosthesis-companies — product category and schema
- brain-computer-interface, brain-interface — category context
- neuralink — the “full brain interface” alternative
References
- Holz FG, Le Mer Y, Muqit MMK, et al. Vision Restoration with the PRIMA System in Geographic Atrophy Due to AMD. N Engl J Med. 2026;394:232–242. DOI: 10.1056/NEJMoa2501396. PMID: 41124203. PMC7618305 (open access).
- Science Corp. press releases 2024–2026 (PRIMA NEJM, acquisition, CE mark).
- Crunchbase News (2026-03): 489M total, $1.5B valuation.
- BrightFocus Foundation (Jul 2026); NEJM 2026;394:232-242.
- Bhuckory MB, et al. Enhancing Prosthetic Vision by Upgrade of a Subretinal Photovoltaic Implant in situ. Nat Commun 2025. DOI: 10.1038/s41467-025-58084-y.
- Palanker Lab (Stanford). Photovoltaic Retinal Prosthesis for Restoration of Sight in Retinal Degeneration. https://web.stanford.edu/~palanker/lab/retinalpros.html 7
- Palanker D, Le Mer Y, Mohand-Said S, et al. Simultaneous perception of prosthetic and natural vision in AMD patients. Nat Commun 2022;13:513. DOI: 10.1038/s41467-022-28125-x. 8
- Muqit MMK, Le Mer Y, Olmos de Koo L, et al. Prosthetic visual acuity with the PRIMA system at 4 years follow-up. Ophthalmol Sci 2024 (medRxiv preprint, PMC10680875). 5
- Wang B-Y, Chen ZC, Bhuckory M, et al. Electronic photoreceptors enable prosthetic visual acuity matching the natural resolution in rats. Nat Commun 2022;13:6627. 9
- Mathieson K, Loudin J, Goetz G, et al. Photovoltaic retinal prosthesis with high pixel density. Nat Photonics 2012;6:391–397. 6
- PRIMA PATENTS (projector & chip): US12061332B2, US12201827B2, US10980997B2, US20250249280A1 — see visual-prosthesis-patents.