Lyceum

Lyceum builds the infrastructure layer intended to make GPU compute as usable as any other cloud service, beginning in Europe. From its headquarters in Berlin and a footprint of European data centres, the company serves AI and machine-learning teams with compute that can be provisioned in seconds and billed per token or per second - pricing made deliberately transparent, with EU data residency and GDPR compliance as the default.

The product surface is broad enough to span the full lifecycle of an AI workload. Inference is delivered through APIs and dedicated endpoints. Training and execution run serverless, with support for Python and Docker. Teams needing direct control can take on-demand GPU virtual machines with full root access, while large-scale GPU clusters support distributed training at scale. The stated aim is to let teams build products rather than manage infrastructure.

Work at Lyceum sits at the intersection of GPU compute infrastructure, cloud computing, inference APIs and distributed training. Engineers deal in the practical problems of that domain - provisioning latency, execution environments, cluster-scale training and metering - rather than abstractions above it.