DARPA Is Preparing Firefox, a Program for a 100-Channel Neural Interface Without Surgery

DARPA Is Preparing Firefox, a Program for a 100-Channel Neural Interface Without Surgery

On September 1, 2026, DARPA will hold an in-person meeting with teams that may participate in the future Firefox program. The agency aims to develop a system capable of using light to read cortical signals through an intact skull.

A neural interface converts signals from the nervous system into commands for an external device. In the Firefox notice, DARPA states that it has already developed a noninvasive coherent optical system that detects neural signals at cortical depths in humans. Firefox is intended to extend this technology into a multichannel interface.

The agency has set targets of 100 channels, a spatial resolution of 50 micrometers, and a temporal resolution of 100 microseconds. Spatial resolution indicates the smallest area of the cortex that the system can distinguish. Temporal resolution indicates how close together in time two signal changes can occur while remaining distinguishable. The program will run for 36 months and consist of three annual phases. An independent DARPA partner will evaluate each team’s results at successive stages.

Firefox divides this work into two related engineering tasks. Dynamic scanning must direct light toward multiple regions of the brain. The other task is to process the large volume of measurement data. DARPA is seeking architectures that go beyond conventional digital data processing.

The completed N3 program faced the same physical constraint: optical, acoustic, and electromagnetic signals scatter and weaken as they pass through the skin, skull, and brain tissue. N3 set a target of operating with 16 independent channels within a 16 mm³ volume of neural tissue and achieving temporal precision of up to 50 milliseconds. Firefox takes this problem as its starting point and sets new measurement parameters.