The United States is investing in the development of photonic chips for artificial intelligence, but it will have to break the laws of physics.

DARPA launches the PICASSO program – “Revolution in Photonics”

The U.S. Department of Defense (through DARPA) announced the start of the PICASSO project (Photonic Integrated Circuit Architectures for Scalable System Objectives). The goal of the program is to overcome fundamental physical barriers that prevent photonic computing systems from realizing their potential: low latency, near‑zero power consumption, and high speed.

Why it matters now
Photonics is still in a “seed” stage of development. Key limitations:

As a result, modern photonic chips can only perform simple linear operations, and the depth of computation is limited by the need to convert light to electricity and back. This negates the advantages of photonics in hybrid solutions.

DARPA’s mission
1. Create circuit designs that bypass the above limitations using already available photonic components (without requiring new materials or devices).
2. Ensure predictable operation of large photonic circuits so they become practical for AI and other compute‑intensive applications.

Program structure
| Phase | Duration | Task |
|-------|----------|------|
| 1 | 18 months | Demonstrate feasibility of the approach. |
| 2 | 28 months | Develop and demonstrate working solutions. |

* Total budget: about $35 million.
* Applications due by March 6, 2026.

DARPA believes that circuit designers have already proven their ability to create reliable and robust electronic architectures; now we need to replicate that success for photonic circuits.

Outcome
The PICASSO program is a challenge: to move photonics from a “seed” state into a mature technology capable of competing with traditional computing systems. Success here could pave the way for widespread use of photonic chips in artificial intelligence and other high‑performance tasks.