The 4,096-electrode era: where brain-computer interfaces stand today.

For most of its history, the brain-computer interface field was defined by a single device: the 100-channel Utah array, essentially unchanged since the 1990s. That era is over. The defining trend of the past two years is channel-count scaling — and the numbers now move fast enough to feel like the early days of Moore's law.

In April 2024, surgeons at Mount Sinai placed 4,096 electrodes on a living human brain using Precision Neuroscience's Layer 7 Cortical Interface — a world record, and double the previous mark of 2,048 that Precision itself had set. Notably, Layer 7 is a thin-film array that conforms to the cortical surface without penetrating tissue, part of a broader industry shift toward less invasive access to high-resolution neural data.

Bandwidth is scaling alongside channel count. In November 2025, the FDA cleared Paradromics to begin a clinical trial of its fully implantable Connexus device, aimed at restoring communication — as text or synthesized speech — for people with severe paralysis. Paradromics has claimed information transfer rates above 200 bits per second in preclinical work — roughly an order of magnitude beyond classic cursor-control interfaces, and enough to make decoded speech feel conversational rather than laborious.

The application frontier is widening too. Neuralink's Blindsight program — a microelectrode array implanted in visual cortex to restore a form of sight to blind patients — received FDA Breakthrough Device Designation in September 2024. Speech, movement, vision: each is now a live clinical program somewhere in the industry, not a hypothetical.

Our read: the bottleneck is shifting from hardware to decoding. Once thousands of channels are streaming, the question becomes what model turns spikes into intent — and that is fundamentally a machine-learning problem. It's why we treat decoder drift, calibration, and mental privacy as first-class research problems rather than engineering afterthoughts.