EWeek reported that Nvidia’s GTC 2025 conference gave industry leaders a live demonstration of a minimally invasive AI-powered neural interface. The demonstration was conducted by Synchron, which claims its brain-computer interface technology is superior to rival projects, including Elon Musk’s much-hyped Neuralink.
Image source: Synchron
The Synchron brain sensor is installed in the brain without surgical intervention. It resembles a stent for dilating blood vessels and is inserted into the jugular vein in the neck. The operation is simple: the stent is brought through the vein to the motor area of the cerebral cortex, where it reads its activity. A signal transmitter is installed under the skin on the patient’s chest, from where the data is transmitted wirelessly to a computer. Such implants have been installed in four Australian citizens and six US citizens.
At GTC, Synchron demonstrated the seamless integration of several new technologies: Apple Vision Pro, Nvidia’s Holoscan multimodal data processing platform, and Sentrode, Synchron’s proprietary interface that connects a person’s thoughts to the physical environment. The patient’s desires (brain activity) were captured by a stent implant and transmitted to the Vision Pro headset’s AssistiveTouch system.
«“We build a model of the brain using generative pre-training methods that learn directly from neural data, abstracting away the fundamentals of the human mind to build features that improve the lives of our users,” said Synchron CEO and founder Thomas Oxley. “This is possible thanks to our ability to scale large data sets, promising to make BCI as common as getting a stent.”
During the demonstration, a user named Rodney used the implanted system to control the temperature, lighting, and music in his home using his mind, something the company had been demonstrating in experiments long before the event.
The company’s goal is to help people with paralysis and other severe physical disabilities control technology hands-free and better interact with the world around them. Data collected in real-world settings will help train “cognitive AI” designed for both medical and broader use. To that end, Synchron is working with Nvidia on the Chiral concept, a future brain-computer interface system that combines real-time neural processing with advanced artificial intelligence. To realize this vision, three key goals will need to be achieved.
First, develop a system for neural decoding of motor functions in real time, allowing users to control digital environments directly through brain implants with minimal latency. Second, add information about the real world using the Nvidia Cosmos environment to map the environment and understand realistic physics. Third, create an AI model based on anonymized user data. All of this will form the basis of Chiral, a “self-improving universal” model for both medical and general use.
Synchron calls this evolving model “cognitive AI.” The ultimate goal is to go “beyond intent recognition, laying the foundation for moving from intent to real-time action” and create a “full-scale core model of the brain.” It’s a bold idea, but the company acknowledges that the concept is still hypothetical.