A tiny brain chip is being heralded as a game-changer for how we interact with computers and treat neurological conditions. The new implant promises a future where brain activity can be captured and interpreted with unprecedented speed, offering new options for epilepsy, spinal cord injuries, ALS, stroke, and blindness, while enabling more fluid control of devices. Its appeal lies in being incredibly compact yet capable of transferring data at very high rates, creating a minimally invasive, high-bandwidth bridge between the brain and external systems.
The device, a brain-computer interface known as the Biological Interface System to Cortex (BISC), was developed through a collaboration among Columbia University, NewYork-Presbyterian Hospital, Stanford University, and the University of Pennsylvania. It centers on a single silicon chip that forms a wireless link to external computers, enabling real-time read and write access to neural signals.
A Nature Electronics paper published on December 8 describes BISC’s architecture, which comprises the chip implant, a wearable relay station, and the software that runs the platform. Ken Shepard of Columbia University notes that traditional implantables are bulky canisters filled with electronics. By contrast, BISC is a single, ultra-thin integrated circuit that can sit between the brain and the skull, resting essentially like a thin sheet of tissue.
Turning the cortex into a high-bandwidth gateway
Shepard collaborated with Andreas S. Tolias of Stanford, who brings vast experience training AI on large neural datasets—some gathered with BISC—to evaluate decoding performance. Tolias explains that BISC effectively converts the cortical surface into a fast, minimally invasive portal for read/write communication with AI and external devices. He emphasizes that the chip’s scalable, single-chip design could enable adaptive neuroprosthetics and brain-AI interfaces to address a range of neuropsychiatric conditions, including epilepsy.
The project’s clinical partner, Dr. Brett Youngerman of Columbia/NewYork-Presbyterian, stresses the potential to transform how neurological illnesses are managed, from epilepsy to paralysis. He, Shepard, and Dr. Catherine Schevon, an epilepsy specialist at NewYork-Presbyterian/Columbia, recently obtained NIH funding to explore BISC as a treatment for drug-resistant epilepsy. Youngerman notes that maximizing information flow to and from the brain while keeping surgical impact minimal is the key to effective BCI devices, and that BISC excels on both fronts.
Shepard adds that advances in semiconductor technology have made it feasible to shrink room-sized computing power into a pocket-sized form factor, and the same progress is now enabling medical implants that occupy very little space while performing complex tasks.
Next-generation BCI engineering
BCIs work by interfacing with the electrical language neurons use to communicate. Conventional medical BCIs typically rely on multiple discrete components—amplifiers, data converters, radio transmitters—packed into a relatively large implanted canister, with connections that may require skull openings or chest implants and wires reaching the brain.
BISC changes the equation. All functions are consolidated onto a single CMOS chip thinned to 50 micrometers, occupying roughly 3 cubic millimeters—less than one-thousandth the volume of typical implants. The flexible chip can contour to the brain’s surface and hosts a micro-electrocorticography array with 65,536 electrodes, 1,024 recording channels, and 16,384 stimulation channels. Because it’s produced with standard semiconductor manufacturing, mass production is feasible.
A complete system is embedded on the chip, including a radio transceiver, wireless power, digital control, power management, data conversion, and both recording and stimulation analogs. An external relay station supplies power and data via a custom ultrawideband link capable of up to 100 Mbps—more than a hundredfold faster than existing wireless BCIs. In effect, the relay station acts as a bridge between any computer and the implant, operating much like a WiFi device.
BISC also ships with its own instruction set and software environment, creating a specialized computational platform for brain interfaces. The high-bandwidth recording enables advanced machine-learning and deep-learning approaches to interpret complex intentions, perceptual experiences, and brain states.
“By integrating everything on a single piece of silicon, we’re proving that brain interfaces can be smaller, safer, and dramatically more powerful,” says Shepard.
Advanced semiconductor fabrication
The BISC chip was manufactured using TSMC’s 0.13-micron Bipolar-CMOS-DMOS process, a three-in-one technology that fuses digital logic, high-current analog, high-voltage analog, and power devices into a single mixed-signal IC. This combination is essential for BISC’s performance and efficiency.
From lab to clinic
To move toward real-world applications, the team partnered with Youngerman at NewYork-Presbyterian/Columbia University Irving Medical Center to design surgical methods for safely placing the ultra-thin implant in preclinical tests and to confirm stable, high-quality recordings. Early intraoperative studies in humans are already underway.
“These initial studies provide crucial data about how the device behaves in a surgical setting,” Youngerman explains. “Implants can be inserted through a small skull incision and slid onto the brain’s surface without penetrating tissue or requiring wires tethering the implant to the skull, reducing tissue reaction and signal degradation over time.”
Preclinical work in motor and visual cortex areas was conducted with Tolias and Bijan Pesaran of the University of Pennsylvania, both renowned in computational and systems neuroscience.
Pesaran notes that BISC’s extreme miniaturization opens up exciting possibilities for integrating other modalities—such as light and sound—into future implantable systems.
The DARPA Neural Engineering System Design program supported BISC, drawing on Columbia’s microelectronics prowess and the neuroscience capabilities of Stanford, Penn, and the surgical infrastructure at NewYork-Presbyterian/Columbia.
Commercialization and future AI integration
To push the technology toward clinical use, Columbia and Stanford researchers spun up Kampto Neurotech, a startup founded by Columbia engineer Dr. Nanyu Zeng. Kampto Neurotech is producing research-grade versions of the chip and seeking funding to bring the system closer to trials in humans.
“BISC represents a fundamentally different approach to building BCIs,” Zeng asserts. “Its capabilities surpass those of competing devices by orders of magnitude.”
As AI advances, BCIs are gaining momentum not only for restoring lost functions in people with neurological disorders but also for potential future enhancements that could augment natural abilities through direct brain-to-computer communication.
“By combining ultra-high-resolution neural recording with full wireless operation and pairing that with advanced decoding and stimulation algorithms, we’re moving toward a future where brain and AI systems interact seamlessly—benefiting research and everyday life,” Shepard reflects. “This could reshape how we treat brain disorders, how we interface with machines, and ultimately how humans engage with AI.”
