Phase 01 / Dedicated capacity
Your first allocation.
A bigger horizon.
Start with SicurAI’s planned 32-rack, 2,304-GPU NVIDIA GB300 deployment in Italy. This is the first phase of Muxaro, with further deployments planned at the same campus.
Shape your capacity brief.
Explore an indicative requirement. Smaller allocations remain subject to the final partitioning plan and commercial minimum.
Your indicative requirement
32 NVL72 racks · March–April 2027
Each block represents one requested rack. This is a sizing illustration, not live available inventory.
Final configuration, availability, term and allocation are subject to agreement.
The service boundary
Define the whole offer.
A useful proposal makes responsibilities and inclusions explicit.
| Compute | Dedicated bare-metal infrastructure based on the planned GB300 NVL72 configuration. |
|---|---|
| Software operation | Customer-operated software and control plane as the proposed initial model. Handover, access and support responsibilities agreed together. |
| Facilities & energy | Facility and energy scope defined in the proposal, with the operating conditions and responsibilities stated. |
| Storage & connectivity | Capacity, performance, external connectivity and any transfer charges specified in your agreed scope. |
| Support & acceptance | Maintenance, escalation, acceptance tests and service levels documented before commitment. |
| Commercial terms | Tailored quotation based on capacity, term, start date, included services and payment structure. |
Procurement questions
Before you reserve.
Is the GB300 cluster operating today?
The GB300 deployment is planned. Target service availability is March–April 2027, subject to final configuration, delivery and commissioning. The target refers to the proposed service, not a statement that GPU capacity is operational today.
Is this the full planned size of the campus?
No. The 32-rack, 2,304-GPU cluster is the first deployment. It represents approximately 25% of the energy allocation envisaged for the fully developed Muxaro campus, with further deployments planned at the same location. Additional capacity, configuration and timing will be confirmed as the later phases progress.
Can I reserve the entire cluster?
Full-cluster discussions for the first deployment are a priority. Final availability, exclusivity, allocation and reservation terms are confirmed through technical and commercial diligence.
Can I request fewer than 2,304 GPUs?
Yes. Share your requirement. Smaller allocations depend on the final partitioning plan, minimum commitment and supported operating model; the sizing tool is not a promise of availability.
Is this an on-demand GPU cloud?
The initial proposition is dedicated reserved bare-metal capacity. Instant self-service instances and a managed inference API are separate services and are not part of this launch offer.
How is the delivery date agreed?
Your capacity proposal will define the target schedule, dependencies and acceptance endpoint. A final contractual date follows confirmation of the configuration and delivery scope.
What happens when I send an enquiry?
The team reviews your capacity, date and workload requirements, then discusses technical fit and commercial scope. An enquiry is non-binding and does not reserve capacity.
Bring your next deployment into focus.
Tell us the capacity you need, your deployment window and what you want to build.