Lyceum

What Lyceum does

Lyceum is a European GPU cloud platform founded in 2025 and headquartered in Berlin with an engineering hub in Zürich, providing infrastructure for AI training and inference workloads under EU data residency. The platform spans four product areas: serverless inference APIs (pay-per-token access to open-source models such as Llama 3.3 70B, DeepSeek V4 Pro, GLM-5.2, Kimi K2.6, and Qwen3.5 via an OpenAI-compatible interface), dedicated GPU endpoints with reserved capacity, on-demand GPU virtual machines and 8-to-8,000-GPU InfiniBand clusters for distributed training, and an AI-powered orchestration layer branded Pythia that analyzes workloads at the compiler level to predict memory requirements and select optimal hardware. Differentiators emphasized by the company include EU sovereignty and GDPR compliance, per-second billing, sub-30-second provisioning, and a stated 76% average GPU utilization versus an industry benchmark near 40%.

The business model combines usage-based pricing (per-token inference, per-second VM rental) with subscription-style committed contracts on clusters (3-24 month minimums). Distribution runs through a self-serve dashboard for startups, an AWS Marketplace listing for AWS-credit-bearing enterprises, and a dedicated enterprise sales team for regulated-industry buyers. Lyceum serves two primary customer segments: AI startups transitioning off hyperscaler credits and seeking sovereign EU compute, and ML infrastructure teams needing orchestration across hybrid on-premise and cloud environments. The company has publicly disclosed two customer case studies (SONALAB in audio AI and Albatross in product discovery) and raised €10.3M in pre-seed funding in June 2025 led by redalpine with participation from 10x Founders.

Name

Lyceum

Legal name

Lyceum Technology Germany GmbH

Website

https://lyceum.technology

Company type

Private

Founded year

2025

Operating status

Operating

Headcount range

11–50 employees

Short description

Lyceum is a European GPU cloud platform offering serverless inference, dedicated endpoints, GPU VMs, and large-scale InfiniBand clusters with EU data residency and AI-powered workload orchestration, targeting AI startups and infrastructure teams across Europe.

Ownership category

akta.pro rank

Product category

GPU Cloud Infrastructure

NAICS

Computing Infrastructure Providers, Data Processing, Web Hosting, and Related Services (518)

SIC

Services-Computer Programming, Data Processing, Etc. (7370)

akta.pro primary industry

AI Compute Cloud & GPU-as-a-Service (HDAAAAAK)

akta.pro secondary industries

Model Hosting, Serving & Inference Platforms (HDAAACAB), AI Compute Virtualization & Scheduling (GPU virtualization, cluster schedulers) (HDAAAAAG)

Keywords

• GPU cloud infrastructure
• Serverless AI inference
• Sovereign EU cloud
• Distributed GPU training
• AI workload orchestration

Headquarters

HQ city

Zürich

HQ country

Switzerland

HQ region

Europe

Offices2 records

Markets served

GTM type

B2B

Offering type

Software

Cost components

Technology or R&D, Infrastructure, Personnel, Operations, Marketing or Sales

Revenue model

1. Serverless Inference: Pay-per-token pricing for inference API access to open-source models. Customers pay based on input/output token volume with no GPU management overhead.
2. Dedicated Endpoints: Reserved GPU capacity for production models with consistent latency guarantees. Billed per GPU-hour with committed usage options.
3. On-Demand GPU VMs: Full root access GPU virtual machines with per-second billing. Customers provision 1-8 GPUs on demand for training, inference, or general compute workloads.
4. Large-Scale Clusters: 8 to 8,000 GPUs with InfiniBand networking for distributed training. Minimum 3-month contracts with custom pricing based on GPU type and duration.
5. Serverless Training: Python and Docker execution on GPUs without infrastructure management. Per-second billing for code execution with automatic hardware selection.

Pricing tiers

Go-to-market motion3 records

Distribution channels4 records

Marketing channels6 records

Core offering

Lyceum operates a European GPU cloud platform providing serverless inference APIs, dedicated endpoints, on-demand GPU virtual machines, and large-scale InfiniBand GPU clusters (8-8,000 GPUs) for AI training and inference workloads. The platform features an AI-powered orchestrator (Pythia) that analyzes workloads at the compiler level to predict VRAM, runtime, and optimal hardware selection, with EU data residency and GDPR compliance built in. Customers access compute via per-second billing, Docker-native execution, and an OpenAI-compatible API for drop-in replacement of hyperscaler services.

Product overview

Lyceum is a GPU cloud platform providing a unified platform for AI inference and training workloads. The core offering spans four product areas: (1) Inference — Serverless Inference (pay-per-token API to open-source LLMs) and Dedicated Endpoints (reserved GPU capacity for production); (2) Training & Execution — Serverless Training (Python/Docker on GPUs without infrastructure management) and Dedicated GPU VMs (on-demand with full root access); (3) Infrastructure — Large-Scale Clusters (8–8,000 GPUs with InfiniBand for distributed training) and Cloud Notebooks (Jupyter on cloud GPUs); (4) Orchestration — Pythia AI Scheduler (compiler-level workload analysis predicting VRAM, runtime, and optimal GPU selection, with multi-cloud unified queuing and cloud bursting). The platform is characterized by per-second billing, EU data sovereignty (GDPR compliance, EU data residency), instant provisioning (~18s for VMs, ~28s for clusters), Docker-native execution, and automated hardware selection. It targets AI startups, infra teams, and enterprises across Europe.

Differentiator

Problem solved

Functional benefit

Products and services

• Serverless Inference Pay-per-token API access to pre-hosted open-source LLMs with no GPU management required. OpenAI-compatible interface supports streaming responses, multi-model hosting, batch and async inference pipelines, and scale-to-zero for cost efficiency. Targets AI teams wanting drop-in migration from US-based inference providers.
• Dedicated Endpoints Reserved GPU capacity (1x to 8x NVIDIA L40S, A100, H100, H200, or B200) for production model inference with 12ms latency, 2,847 requests/min throughput, 99.99% uptime SLA, and 83% GPU utilization. Targets production AI applications requiring consistent low-latency inference.
• Serverless Training Run Python scripts or Docker containers on GPUs without managing infrastructure. Per-second billing with automatic hardware selection via the Pythia AI scheduler. Jobs provision in seconds with no idle costs for early-finishing jobs. Targets ML engineers iterating on training workloads.
• Dedicated GPU Virtual Machines On-demand GPU virtual machines with full SSH root access, provisioned in approximately 18 seconds. Available with NVIDIA L40S (48GB), A100 (80GB), H100 (80GB), H200 (141GB), B200 (192GB), and B300 (262GB) GPUs at $1.19-$7.99/hr. Standard H100 spec: 32 vCPUs, 256GB RAM, 1TB NVMe. Storage at $0.10/GB/month.
• Large-Scale GPU Clusters GPU clusters scaling from 8 to 8,000 GPUs with InfiniBand interconnect for distributed training. Supports NVIDIA GB300 (288GB), B300 (288GB), GB200 (192GB), B200 (192GB), H200 (141GB), and H100 (80GB). Custom pricing with minimum 8 GPUs and minimum 3-month contract. Full cluster provisioning in approximately 28 seconds.
• Pythia AI Orchestration AI-powered orchestration layer that analyzes code at the compiler level to predict VRAM requirements, runtime, and optimal GPU type before execution. Automatically selects the best GPU (H100, A100, L40S) and location (on-prem vs. cloud) to prevent OOM errors and maximize utilization. Also provides unified multi-cloud job queuing across on-premise clusters, AWS/GCP/Azure, and Lyceum Cloud with automatic cloud bursting.
• Cloud Notebooks Jupyter notebooks hosted on Lyceum cloud GPUs, enabling interactive ML development and experimentation without managing infrastructure. Targets ML engineers for prototyping and exploratory work.
• EKS Hybrid Nodes Integration Custom Kubernetes integration that provisions AWS EKS Hybrid Nodes on-demand, plugging Lyceum GPUs directly into a customer's existing EKS cluster. One-click hybrid node provisioning without manual installation or specialized networking software. Allows customers to replace AWS GPU instances with Lyceum GPUs without changing their cloud setup.

Quantifiable outcome

• 38% shorter queue times than industry average
• +5 more outcomes

Named customers2 records

Segments2 records

Ideal customer profiles3 records

Technology focussed Yes

API detail

Has API

Yes

API docs

API detail

Core technology

AI maturity

App detail

Integration2 records

AI capability7 records

Feature5 records

Partnerships

Two partnerships are on record, tiered customer.

• AlbatrosscustomerTechnology or Integration · 24 April 2026AI/Product Discovery company using Lyceum serverless GPU for multi-daily model retraining. Lyceum engineers built custom EKS Hybrid Nodes integration allowing Albatross to use Lyceum GPUs within their existing Kubernetes cluster. Processes billions of user interactions monthly.
• SONALABcustomerTechnology or Integration · 7 April 2026European audio AI company building voice AI tools for media post-production. Uses Lyceum infrastructure for development workloads with plans to expand into GPU training as they scale. Non-negotiable requirement: guaranteed EU data residency.

Scale indicators9 records

Recent moves6 records

Expansion highlights6 records

Direct peers

• CoreWeave: Large independent GPU cloud provider offering on-demand H100/H200 clusters, serverless inference, and dedicated endpoints. Most direct comparable to Lyceum's product surface, though at much larger scale and primarily US-based.
• Lambda Labs: GPU cloud specialist offering on-demand GPU VMs, clusters, and inference for AI workloads. Closely comparable product portfolio and target customer base to Lyceum, with per-hour GPU pricing.
• RunPod: GPU cloud platform with per-second billing, serverless endpoints, and on-demand GPU pods. Closely matches Lyceum's pricing model and developer-focused positioning.
• Modal: Serverless GPU platform for AI/ML workloads with developer-focused ergonomics and per-resource billing. Comparable to Lyceum's serverless training and inference offerings.
• Together AI: GPU cloud with focus on open-source LLM inference and training, offering serverless endpoints and dedicated clusters. Directly comparable inference product surface and open-source model focus.

Regional players

• Scaleway: European cloud provider (France-based) offering GPU instances and sovereign infrastructure. Comparable European footprint and GDPR positioning, though broader cloud portfolio rather than AI-specialized.
• OVHcloud: European cloud provider with GPU instances and strong sovereign cloud narrative. Comparable European data residency positioning and enterprise customer base.
• Hetzner: European hosting provider increasingly offering GPU cloud products with cost-competitive pricing. Comparable regional positioning targeting European customers seeking non-US providers.

Broad incumbents

• Amazon Web Services (EC2 / Bedrock): Hyperscaler offering GPU instances (P5, G5), Bedrock inference, and EU-only regions. The primary competitor Lyceum positions against on cost, sovereignty, and procurement friction.
• Google Cloud (GCP / Vertex AI): Hyperscaler with GPU VM offerings (A3, G2) and Vertex AI inference. Competes with Lyceum on price-performance and AI workload tooling, with EU regions for sovereignty.

Market position

Strengths5 records

Weaknesses5 records

Competitive moat5 records

Key risks6 records

Key highlights7 records

Customer concentration

• LinkedInlinkedin.com/company/lyceum-technology

Compliance1 record

Revenue estimate

Valuation estimate

Number of profiles

Profiles9 records

Funding overview

Funding rounds1 record

Investors2 records

Funding detail is available on the Subscription and Enterprise plan.Contact sales →

M&A

Investments

M&A and investment is available on the Subscription and Enterprise plan.Contact sales →

Frequently asked questions about Lyceum