Sabi raises $50M to build a wearable brain-computer interface that converts thoughts into prompts

Sabi, a Palo Alto startup, has raised $50 million to build a wearable brain-computer interface that converts thoughts into prompts. The device is a beanie, not a brain implant.

The round was announced on October 9, 2026. Its pitch is simple: if you want to talk to AI with your mind, you shouldn’t need a neurosurgeon to make it happen.

A hat with a lot going on underneath

Sabi’s flagship product is a beanie embedded with dense arrays of miniature EEG sensors. EEG, or electroencephalography, reads the brain’s electrical activity from outside the skull, so nothing gets implanted.

Sabi says the beanie packs between 70,000 and 100,000 miniature sensors, each about 1 mm in size.

The goal is to decode internal thoughts into digital text or commands. Sabi’s estimated decoding speed is around 30 words per minute.

A baseball cap variant is also in the pipeline, pitched as a more discreet option.

Who is writing the checks

The $50 million round drew a recognizable lineup. Investors include Khosla Ventures, Accel, Initialized Capital and DST Global.

Kevin Weil, a former chief product officer at OpenAI, also participated.

Sabi plans to use the capital in two areas. The first is improving its hardware capabilities. The second is developing its proprietary “Brain Foundation Model,” a large AI model trained to understand brain signals.

Sabi claims the model is built on the world’s largest neural decoding dataset, totaling approximately 100,000 hours of data collected from around 100 volunteers.

From stealth to shipping in one year

Sabi was founded around 2023. It emerged from stealth in mid-April 2026, around April 16, with plans to bring brain-computer interfaces into the consumer wearable market.

The funding announcement came roughly six months later. Shipment of the product is anticipated by the end of 2026.

Sabi is also emphasizing end-to-end encryption and has made commitments to external audits.

The Neuralink comparison

Sabi is positioning itself directly against surgical brain-computer interfaces like Neuralink. Those systems require invasive procedures to place hardware inside the head.

The tradeoff is signal quality. Reading the brain through the skull is harder than reading it from the inside, which is exactly why Sabi is stacking so many sensors and leaning on a large AI model to fill in the gaps.