Q.ANT and IONOS pilot photonic computing
The collaboration enables developers to test AI workloads on TFLN-based photonic processors hosted in commercial data-centre infrastructure.
Q.ANT and IONOS have launched a joint pilot programme to make photonic computing hardware accessible through commercial bare metal servers, allowing developers to test artificial intelligence workloads on physical photonic processors.
The initiative provides selected participants with access to Q.ANT's second-generation Native Processing Unit (NPU), hosted in an IONOS data centre in Germany.
The processor uses thin-film lithium niobate (TFLN) photonic integrated circuits to perform complex mathematical operations directly using light.
The programme aims to evaluate the potential of photonic computing to improve AI processing efficiency by reducing the number of computational operations and parameters required for certain workloads.
Applications include matrix multiplication, nonlinear mathematical functions, industrial image processing, scientific simulation and robotics.
The launch follows Q.ANT's introduction of an open-source software development kit in September 2026, which enabled developers to build and simulate applications for its photonic architecture.
The new pilot extends this capability to physical hardware, providing an opportunity to assess performance under real-world operating conditions.
Q.ANT's processors have already been deployed for evaluation at the Leibniz Supercomputing Centre and Jülich Supercomputing Centre, where researchers have investigated their potential for high-performance computing and AI applications.
The pilot begins in November 2026 and will be available free of charge to a limited number of selected participants. Developers must submit a proposed use case and demonstrate the technical expertise needed to operate a bare metal server.
Participants will receive technical support from both companies, with results helping to guide further development of Q.ANT's photonic processors and software ecosystem.
The collaboration represents another step towards commercial adoption of PIC-based computing architectures as data-centre operators explore alternatives to conventional electronic processing.
