Neurotechnology news | July 15 – August 6, 2026

Three weeks produced the first decade-long safety record for writing signals into a human brain, the first CE-marked implant that restores readable vision, a first-in-human recording streamed out through the skin on a beam of light, and a three-year result where a man with complete paralysis got movement and touch back at once.

Bottom line: every one of these announcements says it works. Every one of them buries the number that tells you what it costs to keep working. The field is past proving feasibility. The only spec that still separates a demo from a product is degradation, and almost nobody puts it in the headline.

Published: July 15, 2026,
Read more ➡️ UPMC release / Science Translational Medicine

Teams at the University of Pittsburgh and the University of Chicago published the first and longest human safety study of intracortical microstimulation (ICMS), the technique of writing signals into the brain by passing current through implanted electrodes. Five participants with spinal cord injury. 168 million stimulation pulses. 27 cumulative implant-years. Zero serious adverse events.

Sensations evoked in the hand stayed in the hand and did not drift to other body parts over time. Sensations that persisted after stimulation stopped occurred roughly once per 23,000 trials, none painful, none requiring intervention. And 64% of electrodes remained functional on average, including 60% in one participant after a full decade, with decay accelerating later in the study.

The safety result is real and it matters. It clears the path for bidirectional devices people take home, and the same data underwrites microstimulation in visual and auditory cortex. But read that last sentence again. Over ten years, roughly a third of the array stopped working, and the loss rate got worse with time rather than better.

That is the number to design against. If you build anything that writes to cortex, your channel budget is not what you implant. It is what survives to year five. Nobody ships a spec sheet with a decay curve on it. This is the first honest one, and it came out of academia, not a company.

💡 Big Picture: Stimulation safety is no longer the open question. Electrode survival is. The next real milestone in this field is not more channels, it is an array that still has its channels in ten years.

Published: July 21, 2026
Read more ➡️ ABILITY Neurotech / MassDevice

Geneva-based ABILITY Neurotech ran its first human procedure at the Technical University of Munich's Klinikum rechts der Isar. The study takes up to five patients already undergoing brain tumor surgery and records 20 to 30 minutes of neural data from each, intraoperatively. The first patients were under general anesthetic, purely to validate the signal chain end to end. Later stages target decoding in conscious patients performing speech and motor tasks.

The architecture is the point. It is a fully implantable optical brain-computer interface (BCI) that streams raw, lossless neural data through a transcutaneous light link, with no implanted battery and no wires crossing the skin. The company describes the bandwidth as well above Bluetooth-class systems. ABILITY spun out of the Wyss Center and is running a dual EU/US regulatory path.

Every implantable BCI eventually hits the same two walls: getting high-bandwidth data out, and getting power in, without a wire through the skin or a battery in the body. Radio costs you thermal budget. Wires cost you infection risk. ABILITY is betting light solves both.

Keep the caveat in view. An intraoperative recording under general anesthetic is not a BCI, and there are no decoding results yet. But the constraint they chose to attack is the right one.

💡 Big Picture: The implant fight has moved from the electrode to the boundary. Whoever solves data-out and power-in cleanly gets to use anybody's array.

Published: July 27, 2026, Nature Medicine
Read more ➡️ Nature Medicine / Business Wire

The study, led by Chad Bouton during his tenure at the Feinstein Institutes, follows one participant with complete tetraplegia over three years. The system decodes movement intent from cortical implants and drives stimulation of the spinal cord, skin, and brain. Reported results: arm strength gains of 86% on the right side and 62% on the left that persisted for months after stimulation ended, decoding held above 84.6% accuracy for five months without retraining, and restored touch sensation at the wrist through a technique the team calls cortical mirroring.

Two numbers matter and neither is the one in the press release. Five months without retraining is a real recalibration interval, and recalibration burden is precisely what keeps BCIs in labs. And gains that persist after the system is switched off suggest the stimulation is driving plasticity rather than merely substituting for lost function.

Caveat it properly: this is n=1. One participant, three years, no replication. An existence proof, not a rate.

💡 Big Picture: The company built on an implanted BCI result is shipping a product with no surgery in it. Its first commercial device, NeuStim, is cleared non-invasive surface stimulation and launches this fall. When the science and the go-to-market point in opposite directions, follow the go-to-market. That is where the reimbursement is.

Published: July 22, 2026
Read more ➡️ Science Corp / FDA HUD designation

PRIMA received a CE mark under EU Medical Device Regulation from DEKRA and launched commercially across 30 European countries on July 22. The trial behind it, published in the New England Journal of Medicine, ran 38 patients across 17 sites in five countries, all with geographic atrophy caused by age-related macular degeneration (AMD).

Mean improvement was 25.5 letters on the standard ETDRS eye chart, more than five lines. 84% of patients could read letters, numbers and words. 80% gained at least 10 letters of prosthetic acuity at 12 months, with no decline in whatever natural vision remained. Science also announced FDA Humanitarian Use Device designation the same day, and the first commercial implant is expected in Germany.

Look at what the implant actually is: a 2 x 2 mm subretinal photovoltaic chip. Glasses project near-infrared light onto it, and the chip converts that light directly into stimulation current. No battery. No wires. No telemetry link. The implant does no computation and stores no power.

All the intelligence sits in the glasses, outside the body, where you can update it. That is the design lesson, and it is the second item this fortnight where the answer to the power-and-data problem was light.

💡 Big Picture: Passive beat sophisticated to market. The implanted half of your system should be the half that never needs a firmware update.

Published: July 2026
Read more ➡️ Epitel Series B / Aurenar seed

Epitel closed a $26M Series B for remote wireless electroencephalography (EEG) monitoring on July 28. Aurenar, which took an FDA Breakthrough designation for auricular vagus nerve stimulation in June, added a $5.7M seed. Just off the front edge of this window, Hemispheric came out of stealth with $52M for an EEG-based assessment platform starting with PTSD, and Gestala raised roughly $62M for an ultrasound BCI.

Implants get the coverage. Measurement gets funded steadily and unglamorously. If you are an engineer trying to enter this field without a surgical team attached to you, this is where the tractable problems and the jobs are. An EEG platform that survives contact with real clinical workflow is a harder and better-funded problem than most people building decoders assume.

💡 Big Picture: The fastest route into neurotech as a builder is still measurement, not stimulation. Lower regulatory drag, real capital behind it, and you can prototype it on a desk.

Thank you for tuning in,

Ricardo