IN Brief:
- AI systems are increasing the bandwidth and electrical power required to move data between processors, switches, and memory.
- Solinide uses silicon-nitride microcombs to generate multiple wavelength channels from one integrated optical source.
- The €4m seed round will support commercialisation, European manufacturing readiness, and expansion of engineering and prototyping capability.
Solinide Photonics has raised €4 million to commercialise its silicon-nitride photonic integrated circuits and prepare the technology for scalable production. The Gothenburg company is developing optical microcomb sources intended to supply multiple wavelength channels from a single integrated device, with AI data centres and co-packaged optics providing its principal target markets.
The seed round was co-led by Navigare Ventures and PSV Hafnium, with participation from Chalmers Ventures, Turbine Capital, Norrsken Evolve, and Almi Greentech Fund. Solinide says the capital will support product readiness, commercialisation, engineering recruitment, in-house prototyping, and manufacturing preparation in Europe.
The underlying device addresses a growing optical-interconnect problem. Wavelength-division multiplexing allows several data channels to share one fibre, but conventional implementations can require multiple discrete lasers to generate the required wavelengths. Increasing the number of lasers raises component count, electrical power, control complexity, thermal load, and the number of optical sources that have to remain stable throughout system life.
An optical microcomb replaces that bank of emitters with a resonant photonic structure driven by a pump laser. Non-linear optical effects inside the resonator generate a series of precisely spaced wavelengths, creating multiple carriers that can be modulated independently for data transmission.
Solinide builds its photonic circuits on silicon nitride, a material used for low-loss integrated optics and compatible with wafer-scale semiconductor processing. The company says its current microcomb technology has demonstrated optical conversion efficiency above 60%, while a packaged rack-mounted system has produced 28 wavelength channels.
Those figures are relevant because laser efficiency becomes a system-level problem once optical links are deployed by the thousands. Electrical power consumed by light sources ultimately becomes heat inside the data-centre rack, joining the much larger thermal load generated by processors, memory, switches, and power-conversion hardware.
Co-packaged optics intensifies the requirement. Moving optical engines closer to processors or network switches shortens the highest-speed copper connections, reducing electrical loss and equalisation overhead, but it also brings photonics into a much hotter and denser package environment. Laser supply, fibre coupling, thermal stability, packaging yield, and serviceability become part of the semiconductor system rather than separate transceiver problems.
A shared multiwavelength source could reduce the number of individual emitters required around that architecture, although it replaces one form of complexity with another. The generated wavelengths must remain sufficiently stable and uniform, while the pump source, resonator, coupling, control electronics, and packaging all have to operate reliably under changing temperature and load.
Solinide has already taken a second step towards manufacturing by opening a multi-project-wafer service around its silicon-nitride process. MPW access allows several designs to share one fabrication run, reducing the cost of prototyping and giving external teams a route to develop photonic circuits around the same process technology.
That is useful if the company intends the platform to extend beyond its own microcomb products. Photonics processes become more commercially valuable when design teams can build against a defined manufacturing stack rather than relying on one-off fabrication recipes that remain closely tied to the originating laboratory.
The next phase will expose whether the device performance transfers cleanly into repeatable hardware. Fibre attach, optical coupling, wafer variation, automated test, thermal stability, packaging yield, and lifetime qualification tend to become more important as volumes rise, and none can be solved by conversion efficiency alone.
Solinide’s funding therefore sits at the point where the technical proposition has to become a manufacturing one. Generating dozens of optical carriers on one chip is attractive; supplying those devices with consistent wavelength, power, yield, and lifetime inside production AI infrastructure is the more demanding commercial test.
