Summary
- Photon Bridge raised €10 million from new and existing investors.
- Its architecture combines multiple lower power lasers for optical communication.
- Production yield, manufacturing scale and commercial qualification remain important development tests.
Photon Bridge has raised €10 million to develop laser sources for optical connections in AI computing systems, addressing manufacturing challenges associated with supplying large numbers of photonic components for increasingly complex processor networks.
The Eindhoven company confirmed the funding on 21 October. The round was co-led by Amadeus APEX Technology Fund and 360 Capital, with additional investment from existing backers Innovation Industries, Faber and Brabant Development Agency, known as BOM.
Photon Bridge is developing technology that combines light from multiple lower power semiconductor lasers rather than relying on a smaller number of increasingly powerful individual devices.
The company says this approach can reduce the manufacturing difficulties associated with producing very powerful laser sources and improve the ability to scale production. Its architecture uses multiplexing to combine optical outputs and is designed to support multiple wavelengths.
According to Photon Bridge, the platform is intended to provide up to 40 milliwatts of optical power per wavelength per fibre while allowing manufacturing across multiple indium phosphide foundries.
These are stated design capabilities rather than independently established volume production results. The company has not disclosed commercial shipment volumes demonstrating that the technology is already supplying the largest AI computing installations.
Optical communication is increasingly important as AI systems require many processors to operate together. Moving data between processors, switches and storage introduces demands that can affect both system performance and electricity consumption.
Electrical connections continue to serve important functions within computing equipment, but their design becomes more challenging as data transmission requirements increase. Optical links provide a route to transferring information using light, although they introduce their own requirements for transmitters, receivers, lasers and packaging.
One developing approach is co-packaged optics, in which optical connectivity is integrated close to the electronic devices handling network traffic. Shortening some electrical connections may offer performance and energy advantages, but only where the optical components can be produced and integrated at a suitable cost.
The laser supply chain has traditionally served telecommunications and other specialist optical markets. AI infrastructure could create substantially greater unit demand, particularly if optical connections move closer to the processors within large computing clusters.
Photon Bridge cites an Omdia forecast suggesting that more than 20 million data centre AI accelerators could ship annually by 2028. Such estimates indicate potential market scale, but they are forecasts rather than confirmed orders for its products.
Manufacturing yield is central to the company’s approach. Producing large numbers of consistent optical components requires processes that can tolerate normal variations in materials and fabrication while delivering acceptable performance.
A design that performs well in development testing must still meet requirements for reliability, thermal management, packaging and integration with customers’ systems. These factors can determine whether an optical component becomes commercially viable even when its underlying technology is promising.
Photon Bridge’s intention to use multiple indium phosphide foundries may also provide manufacturing flexibility, although qualifying a process across different production locations requires testing and process control.
The new funding will support the company’s development programme as suppliers compete to address networking requirements created by larger computing installations. For European data centre infrastructure, the eventual contribution would be through the networking equipment installed within and between facilities rather than through direct construction of additional capacity.
The next evidence of commercial progress will be validated product performance, manufacturing qualification and customer adoption at meaningful production volumes.
