When Daniel Palanker first sketched out the concept for a wireless retinal implant in 2005, the idea seemed almost too elegant: use light itself to power a chip that would replace damaged light-sensing cells. No batteries. No cables threading through the delicate tissue of the eye. Just a tiny photovoltaic square, smaller than a pencil eraser, that could restore reading vision to people who had lost it forever.

Twenty years later, that sketch has become the PRIMA device, and in October 2025, results published in the New England Journal of Medicine confirmed what Palanker and his Stanford Medicine team had worked toward: 84% of trial participants could read again after a year with the implant. Some achieved 20/42 vision with digital enhancements—not perfect, but functional. One person improved by 12 lines on a standard eye chart.

How Light Powers Vision Twice

The human retina normally relies on roughly 126 million photoreceptors—rods and cones that convert light into electrical signals. But in geographic atrophy, the advanced form of age-related macular degeneration, these cells die off in the center of the retina. Over 5 million people worldwide live with this condition, which destroys central vision while often leaving peripheral sight intact.

Here's where retinal implants diverge from intuition. Rather than trying to repair or replace photoreceptors, they bypass them entirely. The PRIMA chip sits beneath the retina, directly stimulating the neurons that would normally receive signals from healthy photoreceptors. These inner retinal cells—bipolar and ganglion cells—remain viable even after photoreceptors die.

The device measures 2-by-2 millimeters and contains 378 pixels, each 100 microns wide. A pair of glasses worn by the patient projects infrared light onto the chip. Because the light is infrared, it's invisible to any remaining peripheral photoreceptors, which means patients can use both their prosthetic central vision and their natural peripheral vision simultaneously. The chip converts that infrared light into electrical pulses that stimulate the retinal neurons, creating the perception of shapes and patterns.

No external power source threads out of the eyeball. No battery pack. The photovoltaic design means the implant operates entirely on the light projected from the glasses.

What Reading Again Actually Means

The 32 participants in PRIMA's clinical trial were all over 60 years old with vision worse than 20/320—legally blind by any measure. After one year, 27 could read. On average, participants improved by five lines on standard eye charts. Two-thirds reported medium to high satisfaction with the device.

These numbers matter because they represent a shift from theoretical possibility to practical function. Earlier retinal implants could help people detect light or perceive motion, but reading requires resolving fine detail. The ability to read food labels, subway signs, or books changes what daily life looks like for someone with severe vision loss.

That said, the improvement takes work. Participants needed several months of training to reach top performance—similar to the adjustment period for cochlear implants. The brain has to learn how to interpret signals from an artificial source. This isn't instant vision restoration; it's a learned skill.

The Implant That Came Before

PRIMA isn't the first retinal prosthesis to reach patients. The Argus II, developed by Second Sight Medical Products, received FDA approval in February 2013 after more than two decades of development. It takes a different approach: an epiretinal device, meaning it sits on top of the retina rather than beneath it. The Argus II has 60 platinum electrodes and is approved for patients with end-stage retinitis pigmentosa, a different condition that also destroys photoreceptors.

Over 500 patients globally have received retinal prostheses in the past 15 years. But the Argus II and similar devices provide vision below 20/200, which still meets the criteria for legal blindness. Functional improvements occur—better mobility, orientation, and the ability to navigate environments—but reading remained out of reach for most users.

Germany's Alpha AMS subretinal implant pushed the technology further with 1,600 stimulation units on a chip measuring just 4.0 by 3.2 millimeters. In trials, five out of six participants showed improved visual performance, and the device continued functioning for up to 24 months. Each stimulation unit measures 70 by 70 microns and includes its own photodiode, amplifier, and electrode.

The trade-off between these approaches involves placement, resolution, and power. Epiretinal implants showed more frequent adverse events—conjunctival erosion, retinal detachment, ocular hypertension—though most resolved within two months. Subretinal implants like PRIMA and Alpha AMS sit closer to the neurons they're stimulating, potentially allowing for finer resolution.

The 20-Micron Horizon

Palanker's team is already developing next-generation PRIMA chips with pixels as small as 20 microns wide—five times smaller than the current version—and 10,000 pixels per chip instead of 378. The math suggests this could provide 20/80 vision, or close to 20/20 with electronic zoom.

That projection raises a question about what counts as restored sight. Even with these advances, prosthetic vision won't match biological vision. The normal human retina is an exquisitely complex structure with multiple cell types processing information in parallel. Current implants stimulate neurons in a relatively crude way, creating perception but not the full richness of natural vision.

Yet "crude" undersells what these devices accomplish. For someone who lost the ability to see faces, read text, or recognize objects, regaining those capabilities—even imperfectly—represents a meaningful change. Quality of life improvements in mobility and daily tasks show up consistently across trials.

When Blindness Isn't Binary

The success of PRIMA clarifies something about vision loss: it's rarely all-or-nothing. Geographic atrophy destroys central vision while leaving peripheral photoreceptors intact. Retinitis pigmentosa typically works the opposite way, eroding peripheral vision first. Retinal implants work best when some neural architecture remains functional—when the inner retinal cells that relay information haven't died along with the photoreceptors.

This selectivity means retinal implants aren't a universal solution for blindness. They address specific forms of vision loss where the signal pathway remains partially intact. For conditions that damage the optic nerve or visual cortex, different approaches are needed.

But for the 5 million people with geographic atrophy, and the millions more with retinitis pigmentosa, the technology has moved from experimental to practical. The 20-year gap between Palanker's initial sketch and published clinical success reflects the difficulty of engineering devices that must function reliably inside the human eye for years. The next-generation chips with 10,000 pixels may come faster, building on what's now established: that light can power vision in more ways than one.