Photonic chip company Q.ANT has partnered with Ionos to launch a pilot program that will allow companies, researchers, and developers to test and evaluate their AI models on Q.ANT’s photonic hardware.

In a statement, the partners said the program will open up a new way of computing, supporting developers to build “leaner AI models” that make direct use of complex nonlinear mathematical functions, thus enabling them to run faster.

Approved applicants will gain access to Q.ANT’s second-generation Native Processing Unit (NPU) via the Ionos website. The processors are hosted on a bare-metal server in an Ionos data center in Germany, and users will be able to use the hardware to run and evaluate their own workloads with their own data and software.

Ionos did not specify which data center would host the hardware but, according to the company’s website, its German data center footprint includes a facility in Berlin, with Ionos also leasing space from Equinix in Frankfurt. Globally, Ionos operates 32 data centers.

Additionally, developers will be able to explore new algorithms and AI architectures that result from the NPU’s ability to compute complex nonlinear functions. The partners claim that this will allow AI models to deliver “comparable or better results” with fewer parameters, fewer operations, and fewer resources.

The program is also able to identify where photonic processing offers the greatest advantages, providing the necessary validation needed for broader commercial deployment.

“A new computing architecture only becomes commercially relevant when customers can test it against the workloads that matter to them. When we launched our SDK, developers could begin building applications for photonic computing. Now, with Ionos, they can take their work and run those applications on the real processor,” said Dr. Michael Förtsch, founder and CEO of Q.ANT. “This matters because the conversations around AI are shifting from what the models can do to what it costs to run them. Photonic computing can make the next generation of AI models leaner: complex functions that inflate digital models can be computed with fewer data and parameters with light, resulting in less energy requirements for the same workloads.”

Achim Weiss, CEO of Ionos, added: “Promising computing technologies from Europe need sovereign European infrastructure to make the leap into business use. In the pilot program, Q.ANT’s photonic processor runs in an Ionos data center under German jurisdiction. Users can explore processing with light on their own workloads. The results will show which applications benefit most and will feed directly into our future development work.”

Based in Stuttgart, Germany, and founded in 2018 as a spin-off from industrial firm TRUMPF, Q.ANT has been developing photonic chips built on Thin-Film Lithium Niobate (TFLN). The processors work by executing mathematical operations directly in the optical domain using photonic integrated circuits, which Q.ANT says eliminates on-chip heat generation and cooling requirements.

In July 2025, the company raised €62 million ($72m) in a Series A funding round co-led by Cherry Ventures, UVC Partners, and imec.xpand.

Ionos is part of Virt8ra, a testbed sovereign Edge cloud platform in Europe that launched in January 2025. In May of this year, the company became the first commercial cloud data center customer for Q.ANT’s Native Processing Server (NPS).

According to Q.ANT, its NPS offering integrates “seamlessly” into data centers as a plug-in co-processor, and claims the hardware can offer up to 30x energy efficiency and 50x performance improvement, with the potential to increase data center capacity 100x without the need for “complex active cooling systems.”