Emerging Brain-Computer Interface Industry Across Chips, AI, and Regulation
From surgically embedded chips, adaptive neuroheadsets, and AI-steered interfaces to an intensifying geopolitical contest
In summer 2016 Noland Arbaugh, a student of Texas A&M University, suffered spinal cord injury during lake diving. This accident changed his life forever, leaving him paralysed from the shoulders down. In January 2024 Neuralink in collaboration with Barrow Neurological Institute turned Noland into the first recipient of a Neuralink brain chip “Telepathy”.
This wireless system, which monitors the electrical activity of neurons and transmits those signals as instructions to an external device, helps Arbaugh control the computer for daily tasks. Now he’s capable of sending emails, playing chess or Civilization 6 and even work, all through the brain-computer interface (BCI).
In this article: How it started — The Utah Array — BCI Patent Landscape — Non-Invasive BCIs — Minimally Invasive BCIs — Invasive BCIs — BCI & AI Fusion — China’s vs. West’s BCI Roadmap — Ethical Flashpoints
“It’s just made me more independent, and that helps not only me but everyone around me. It makes me feel less helpless and like less of a burden,” said Arbaugh.
Noland’s story underlines the growing BCI relevance. Surging interest in BCIs has created Around 25 clinical trials of implantable BCIs are already in progress, and MIT Technology Review readers have voted these devices as the addition onto the magazine’s 2025 “10 Breakthrough Technologies” list. Still, BCIs have come a long way until they started transforming people’s lives for the better.
How it started
The history of brain-computer interfaces began in 1875 when British doctor Richard Caton discovered electrical activity in the brain with a galvanometer. Inspired by Caton, Hans Berger, a German psychiatrist, recorded the first electroencephalograph (EEG) from a child undergoing brain surgery in July 1924. Berger initially used conductive silver wires placed on the boy’s scalp, later developing scalp-mounted sensors that enabled non-invasive EEG recordings. He published his findings in 1929, introducing a method still widely used today for diagnosing epilepsy, brain tumors, and strokes.
Research expanded over the following decades. In 1964, Spanish neurophysiologist José Manuel Delgado captured public attention by halting a charging bull with radio-controlled electrodes stimulating the animal’s motor brain structure. That same year, neurophysiologist W Grey Walter demonstrated a direct link between brain signals and intention. In his experiment, participants wearing EEG sensors pressed a button they believed advanced slides in a projector. Little did they know, the projector responded only to their brain activity.
How did BCIs get their name? The term first appeared in 1973 when computer scientist Jacques Vidal explored the possibility of using brain signals to communicate with computers and control external devices like prosthetics. In 1977, Vidal validated his idea by demonstrating participants could mentally guide a cursor through a digital maze using EEG-recorded brainwaves.
While Vidal’s vision was accurate, EEG proved too imprecise for detailed control. By the 1990s, researchers realized clearer signals required electrodes placed directly on or within the brain tissue. Irish neurologist Phillip Kennedy created one of the earliest implanted interfaces in the late 1990s. Kennedy’s device, involving a glass cone with Teflon-coated wires, directed neurons to grow into the cone and enhance signal detection. With this, Kennedy enabled a patient with locked-in syndrome (severe paralysis that restricts voluntary movement and speech) to control a computer cursor by painstakingly typing out words. By the early 2000s, monkeys implanted with early BCI devices successfully controlled robotic limbs, laying the groundwork for future human applications.
The Utah Array & Playing Pong via Brain Activity
In 2004, the first clinical trial of a BCI, BlackRock Neurotech’s BrainGate, demonstrated significant progress. 25-year-old Matthew Nagle, with paralysis, used a BrainGate interface to control a computer cursor and play Pong using his brain activity. The device, a product of featured a small chip called Utah Array (named after BlackRock’s location in Salt Lake City) implanted into the brain’s motor cortex, connecting neurons to a computer through an external cable. Researchers John Donoghue and Leigh Hochberg led the trial, showcasing Nagle's ability to change TV channels, read emails, draw digitally, and operate a robotic hand. Another recipient of the chip, Nathan Copeland, who got his Utah Array implant in 2015, still holds the record for the longest continuous use of a brain-computer interface.
BCI advancements continued through the 2010s. In 2011, researchers reported a woman with locked-in syndrome controlling a cursor more than 1000 days post-implantation, a record at that time. In 2012, participants with tetraplegia used neural implants to control robotic arms, performing tasks like picking up a coffee bottle and drinking from a straw
BCI nowadays come in 3 main types:
Invasive BCIs: involve surgically implanting electrodes directly into a patient's brain tissue. Due to the significant risks associated with surgery, these interfaces are best suited for individuals aiming to recover from severe conditions like paralysis, significant injuries, and neuromuscular disorders;
Minimally invasive BCIs: involve surgical intervention to directly access the brain tissue and don't need to deeply penetrate it, a balance between safety and signal potency;
Non-Invasive BCIs: mostly involve wearable devices with electrical sensors acting as communication channels between the brain and a machine. These systems produce weaker signals since they do not directly contact brain tissue. Non-invasive BCIs are better suited for applications like virtual gaming, augmented reality, and controlling robotic systems and other technologies.