European fabless designer of CPUs, SiPearl, and computing company, Semidynamics, are joining forces in developing a European rack scale compute platform dedicated to advanced AI inference in the cloud.

SiPearl’s Arm-based CPU will provide the general purpose compute, orchestration and data plane hosting, whereas Semidynamics’ RISC V based GPU/AI inference ASIC will provide the acceleration for AI inference workloads. The two companies aim to offer a rack-scale system delivering the density expected from advanced global AI platforms. The design will be based on the Open Compute Project (OCP) standard, supporting interoperability and alignment with established cloud and data centre infrastructure practices.

The platform’s architecture will tackle high throughput, high reliability cloud deployments, yet at low power. This makes it highly suitable for enterprise inference server clusters and AI services that need consistent, large scale processing power. Among the target applications are AI inference in the cloud, especially when LLMs are involved and retrieval augmented generation (RAG) pipelines; enterprise scale inference in areas such as customer service automation and industrial analytics; and sovereign public sector workloads where data control and autonomy are essential.

The next step in the cooperation will focus on further chiplet-level integrations and the reference architecture.

Pundits state that Europe’s technological sovereignty lies at the heart of this collaboration, with key compute components being developed in Europe, thus reducing dependence on non-European “full stack” ecosystems.

“SiPearl is thrilled to see the impact of years of work in the European Processor Initiative and the EU sovereign ecosystem come to fruition with this platform. It demonstrates the systematic progress that Semidynamics and SiPearl have made individually and collectively, and will showcase the best of both companies,” said Philippe Notton, SiPearl’s CEO and Founder.