SciTransfer
NEUROPIC · Project

Ultra-Low Power Programmable Light-Based Chips for AI and Data Processing

digitalPrototypeTRL 3

Imagine a computer chip that uses light instead of electricity to move information, like a tiny city of fiber-optic roads. Instead of using heat-heavy switches, it uses microscopic mechanical parts that move to redirect light. This makes the chip run much cooler and use far less energy than today's AI hardware.

By the numbers
thousands
programmable nodes
4
potential business cases
7
consortium partners
The business problem

What needed solving

Current AI and data processing hardware consumes too much power and generates excessive heat. Existing programmable light-based chips are either too power-hungry or too small to be practical.

The solution

What was built

A programmable photonic chip architecture using nanoelectromechanical technology and optimized silicon nanofabrication processes.

Audience

Who needs this

AI chip designersData center infrastructure providersAutonomous driving hardware engineersQuantum computing architects
Business applications

Who can put this to work

Cloud Computing
enterprise
Target: Hyperscale Data Center Operator

If you are a data center operator dealing with massive electricity bills and overheating servers — this project developed nano-electromechanical programmable photonic circuits that reduce power consumption compared to conventional technologies.

Automotive
enterprise
Target: Autonomous Vehicle Manufacturer

If you are an EV maker dealing with the high energy drain of onboard AI processors — this project developed a compact, low-power photonic chip architecture that can process complex data for autonomous systems more efficiently.

Quantum Computing
SME
Target: Quantum Hardware Startup

If you are a quantum processor developer dealing with the need for precise, scalable signal routing — this project developed a programmable platform scaled to thousands of nodes to manage optical interconnections.

Frequently asked

Quick answers

How does this reduce operational costs?

Based on available project data, the technology uses nanoelectromechanical switches which offer ultralow power consumption compared to existing programmable photonic approaches, reducing energy overhead.

Can this be produced at an industrial scale?

The project is developing scalable platforms and advanced silicon nanomachining to enable the creation of circuits with thousands of programmable nodes.

What is the IP and licensing status?

Based on available project data, the project is in the research phase and does not specify licensing terms, though it aims to establish foundations for four business cases.

How does it integrate with current AI software?

The project focuses on neuromorphic computing, using photonic systems with nonlinear dynamics to perform information processing tasks similar to brain-inspired AI.

What is the development timeline?

The project period runs from March 1, 2023, to February 28, 2027.

Consortium

Who built it

The consortium consists of 7 partners across 5 countries, showing a balanced mix of academic and commercial interests. With 2 SMEs and a 29% industry ratio, the project has a direct link to the market, while the 3 research centers and 2 universities provide the deep technical expertise in silicon nanotechnology required for this high-risk development.

How to reach the team

Contact AGENCIA ESTATAL CONSEJO SUPERIOR DE INVESTIGACIONES CIENTIFICAS in Spain

Next steps

Talk to the team behind this work.

Contact us to identify the specific business cases emerging from this photonic research.