Hey Neurohacker,

I paused seven weeks ago. You told me you used this for orientation, not news. So: longer gaps, denser issues, less filler.

Seven weeks of neurotech below. Nothing exploded. Everything got more durable. That's the shape of it.

Neurotechnology news | May 24 – July 12, 2026

Across the whole window: a brain-computer interface (BCI) at home for 3,800 hours without a researcher in the room, the first invasive BCI ever cleared for commercial sale filing to go public, the fourth US high-density implant going into a patient, a $650M peripheral nerve stim acquisition by the incumbent, an FDA Breakthrough for auricular vagus nerve stimulation on a real clinical endpoint, and two more states putting neural data behind opt-in consent. Every single item is a durability signal. None of them are demos.

Bottom line: the field stopped optimizing for what it can show and started optimizing for what it can keep working, ship, defend in front of a regulator, and sell. That is a much better market to build in than the one we were in six months ago.

A brain-computer interface just proved it can run someone's life for two years.

Published: June 2026, Nature Medicine
Read more ➡️ Nature Medicine paper / UC Davis release / MIT Technology Review

The UC Davis / Brown / Mass General Brigham team published two years of at-home data from Casey Harrell, an amyotrophic lateral sclerosis (ALS) patient with a 256-channel intracortical implant from the BrainGate2 trial. The numbers: 3,800 hours of independent use, more than 183,000 sentences, close to two million words, 56 words per minute average speed, 99% accuracy in-lab and 92% in his living room. He runs the system himself, without researcher setup, every day.

This is the paper that ends the "does it work?" phase of intracortical BCI and starts the "how do we ship this?" phase. Every prior speech-BCI result was a session number, hand-tuned in a lab. This one is longitudinal, unsupervised, and durable across a real signal environment: naps, background noise, sensor drift, mood, illness. If you build BCIs, the interesting engineering problem is no longer "can the decoder read speech." It is "can the decoder stay calibrated for a year with no one holding its hand." That is a different research program, and it is now the one that matters.

💡 Big Picture: Durability is the new state of the art. The demo era is over.

Published: June 17, 2026
Read more ➡️ Michigan Medicine / ClinicalTrials.gov NCT07357428 / MassDevice

Neurosurgeons at University of Michigan Health implanted the first Connexus BCI in a woman with motor neuron disease as part of the Connect-One Early Feasibility Study. Architecture worth reading closely: high-density microelectrode array in cortex, wired to a subcutaneous transceiver in the chest, transcutaneous radio-frequency (RF) link to an external receiver on the skin. The chest-transceiver split is a deliberate bet that thermal budget and long-term skin integrity are the real reliability risks, not electrode count.

The US now has four implantable BCI companies in active clinical trials (Neuralink, Synchron, Precision, Paradromics), each pursuing a distinct signal-chain architecture. That is a field, not a race.

💡 Big Picture: A heterogeneous hardware landscape is now permanent, which means whichever decoder software layer generalizes across architectures wins more of the value than any single implant vendor does.

Published: IPO filing June 11, 2026 (NMPA approval March 13, 2026)
Read more ➡️ Scientific American / MIT Technology Review / Neuracle STAR Market filing coverage

China's National Medical Products Administration (NMPA) cleared Shanghai-based Neuracle's NEO system for tetraplegia caused by cervical spinal cord injury in March. On June 11, Neuracle filed for an initial public offering (IPO) on Shanghai's STAR Market to raise about $345M. NEO is not intracortical. It uses an epidural (extra-dural) electrode array paired with a pneumatic robotic glove, keeping the electrodes above the dura and off the cortex, with a 100% grasping success rate reported across 32 clinical trial patients.

The point that hurts if you were still betting on channel-count as a moat: the first invasive BCI cleared for commercial sale anywhere is a low-density, extra-dural system with a narrow motor effector. Regulators picked the safer footprint over the more capable one, and the capital markets are validating it in-window.

💡 Big Picture: If you are prototyping in the West, the fastest regulatory path is probably not the highest-capability architecture. Design for the surgical-risk profile the regulator will credential, then earn capability increments from there.

Published: May 20, 2026
Read more ➡️ Medtronic press release / MassDevice

Medtronic announced its intent to acquire SPR Therapeutics for approximately $650M. SPR makes SPRINT, an FDA-cleared temporary percutaneous peripheral nerve stimulator for chronic pain that a patient wears for 60 days and then removes. No implanted pulse generator. No long-term lead. A drug-like course of neuromodulation therapy.

This continues the theme from Weekly Signal #58: capital is picking the smaller footprint. In this case the incumbent is doing the picking, not a startup. That is the same signal from the other side of the market.

💡 Big Picture: The center of gravity in neuromodulation is moving from permanent implants to temporary interventions, and the incumbents are buying their way in rather than building. If you are a startup with a temporary-form-factor neuromod device, the acquisition comp just repriced.

Published: June 30, 2026
Read more ➡️ Aurenar press release / NeuroNews International / WashU Medicine

The FDA granted Breakthrough Device Designation to Aurenar's V-Link System, a non-invasive auricular vagus nerve stimulator, for reducing cerebral vasospasm in adult patients with aneurysmal subarachnoid hemorrhage (aSAH). The randomized clinical trial that supported it (27 patients) cut moderate-to-severe vasospasm by more than 40% with no associated adverse events.

What's interesting is the indication, not the modality. Auricular vagus nerve stimulation has been a wellness-market punching bag for a decade. Here it is going after a hard ICU endpoint with a small but positive randomized trial, and the FDA is expediting it. The lesson for builders using non-invasive stim: pick an acute inpatient indication with a measurable endpoint and a defined care setting. The neurowellness framing is a dead end.

💡 Big Picture: The path back to credibility for non-invasive neuromodulation is inpatient acute care with hard endpoints, not consumer wellness with vibes.

Effective: July 1, 2026
Read more ➡️ TechTimes on the Connecticut rule / Cooley on the state patchwork

Connecticut's Data Privacy Act amendment now classifies neural data as sensitive data, requiring explicit opt-in consent before collection and separate opt-in before any third-party sale. It is the fifth US state (after California, Colorado, Montana, Vermont) to put neural data behind an opt-in wall. Vermont is running its own definitional refinement in parallel.

The important design problem: the state laws do not agree on what "neural data" is. Some cover raw signal only. Some cover derived features. Some sweep in inferences drawn from neural data. If you ship a consumer electroencephalography (EEG) or functional near-infrared spectroscopy (fNIRS) product, your consent flow now needs per-state branching, and your third-party data pipeline is directly in scope in Connecticut and Colorado.

💡 Big Picture: Neural-data compliance is now a systems-design problem, not a legal footnote. Design your data pipeline for the strictest state, or fork it. There is no third option.

The rest of the seven weeks, one line each.

Help improve The NeuroCircuit

Thanks for still being here.

Ricardo