Intelligence, on contract.
Lume is building the first independent contracted compute platform: deploying large-scale AI clusters against long-term offtake contracts, while maximising speed and flexibility.
Independently owned and operated
Long-term contracted capacity
Stranded energy sourcing
Key Metrics
40 MW
Contracted capacity across 3 sites
GB300 NVL72 / B300
Cluster generations
Q1 2027
First capacity targetted
24/7
Managed operations
The Model
The IPPA.
In energy, a Power Purchase Agreement lets a buyer lock in capacity before a plant is even built.
We apply the same logic to compute: an Intelligence Purchase Agreement (IPPA) is a long-term contract for dedicated AI capacity, priced and secured before deployment, not rented after the fact. We deliver intelligence to end customers, we do not just sell GPU time.
We apply the same logic to compute: an Intelligence Purchase Agreement (IPPA) is a long-term contract for dedicated AI capacity, priced and secured before deployment, not rented after the fact. We deliver intelligence to end customers, we do not just sell GPU time.
The Problem
We Solve
/ 001
Capacity Gap
No operator currently combines independent ownership, European energy access, and long-term contracted AI capacity at scale.
/ 002
Execution Trap
Enterprises and hyperscalers needing dedicated AI capacity face long
procurement cycles, limited hardware, and the complexity of building
private infrastructure themselves.
procurement cycles, limited hardware, and the complexity of building
private infrastructure themselves.
/ 003
Control and governance
Strategic AI workloads increasingly require clarity over location, access, operations and long-term capacity.
Lume is the only contracted compute platform built to resolve all
three at once.
three at once.
What we build
IPPA
Intelligence Purchase Agreement
THE COMMITMENT
GPU Clusters
NVIDIA GB300 NVL72 and B300
The engine
Networking
InfiniBand, low-latency
The connective tissue
Data Centre
Tier III, purpose-built
The vessel
Power Source
Grounded European energy
The advantage
Operations
Availability, security and support
The assurance
Capacity
Total: 40 MW
Stranded energy, unlocked.
We build contracted AI capacity directly on top of power that already exists.
We build contracted AI capacity directly on top of power that already exists.
/ SITE 001
Available Q1 2027
10 MW GB300 NVL72
/ SITE 002
Available Q1 2027
5 MW B300
/ SITE 003
Available Q3 2027
25 MW GB300 NVL72
The Moment
AI demand is accelerating
Dedicated compute is becoming strategic infrastructure
Power is the constraint
AI capacity can only scale where power, cooling and connectivity are available
Capacity takes time to deploy
Sites, hardware and commissioning require long lead times
Contracted, not queued
Every Lume site is secured under an IPPA before a single GPU is deployed. Customers know exactly what they're getting, and when