Science Corporation Implants PRIMA Chip in First Stargardt Patient, Advancing Retinal BCI

Science Corporation, founded by former Neuralink president Max Hodak, has implanted its PRIMA subretinal chip in the first patient diagnosed with Stargardt disease, an inherited condition that progressively destroys central vision. The procedure took place in May at Sydney Eye Hospital as part of an ongoing clinical trial. During initial rehabilitation, the patient reported seeing a single letter, then reading a full word — early evidence that the implant can bypass degenerated photoreceptors and stimulate the intact inner retina.

The PRIMA system consists of a 2 mm × 2 mm photovoltaic chip placed beneath the retina and a pair of glasses equipped with a camera that wirelessly transmits visual data to the implant. The device converts light patterns into electrical pulses, effectively functioning as artificial photoreceptors. Current output is monochromatic, high-contrast, and limited to a narrow field of view — what Hodak describes as "looking through a straw." Yet the acuity within that window is sharp enough for character recognition.

Why it matters for GPU / AI infrastructure

Retinal prostheses like PRIMA generate data streams that demand real-time, low-latency processing — computer vision pipelines, signal encoding, and adaptive stimulation algorithms all run on accelerated hardware. As these devices scale toward wider fields of view and higher electrode densities, the computational load per patient will grow, creating demand for edge GPUs and cloud inference capacity. AiGpu's infrastructure is positioned to support the training and deployment of the vision models that turn raw camera feeds into meaningful neural stimulation patterns.

Hodak outlined a roadmap: a next-generation implant with smaller electrodes and a larger footprint is slated for human trials within a year, with a five-year target of approaching native visual acuity. If achieved, the addressable population expands beyond the ~150,000 Stargardt patients worldwide to millions with geographic atrophy and other outer-retina degenerations. Each implanted system becomes a persistent edge node in a distributed neurotechnology network — a use case that aligns directly with high-performance, low-latency GPU cloud services.