The operating model·

AI Factory Platform:

Seven Operating Layers

CambridgeNexus is an AI Factory operator. We own bare-metal NVIDIA GB300 NVL72 racks and run every layer around them as AI Factory as a Service (AIFaaS), so customers lease whole racks and run production AI without building any of it themselves.

Talk to us

A GPU alone does not create ROI

Worldwide AI spending is projected to pass $2.5 trillion this year, according to Gartner. Investment alone does not create business value.

The economics are decided by how compute, power, networking, storage, software, orchestration and operations work together. That is the problem the CambridgeNexus AI Factory is built around: moving AI from ambition to production, economically. The seven layers below are how.

The seven layers of the CambridgeNexus AI Factory

Every deployment runs on the same stack. You lease the NVIDIA GB300 racks; CambridgeNexus operates everything around them.


One operator across all seven layers means your dates are commitments, written into the contract you sign.

See how a deployment runs

  1. Power

    Facility power provisioned for 140kW per rack. We contract it and manage it end to end. You never call a utility.

  2. Cooling

    Direct liquid cooling built for GB300 density. We install it, monitor it, and maintain every loop. Your team never handles coolant.

  3. Networking

    High-bandwidth fabric inside the rack, private connectivity out. We configure both for every deployment. You connect and run.

  4. Compute

    Bare-metal NVIDIA GB300 NVL72 racks: 72 Blackwell Ultra GPUs and 36 Grace CPUs, one customer per rack. This is the layer you lease.

  5. Orchestration

    Scheduling and utilization management, operated by us. You get a tuned cluster without running a single tool.

  6. Compliance

    SOC 2 Type I, with GDPR and HIPAA status published on the trust center. Export screening on every deployment. We carry the audits.

  7. Workload planning

    You describe the run. We propose capacity, site, and configuration to match. Sizing is our job, from day one.

From inquiry to production

Six steps. Every stage has a named owner and a date in your contract.

The six steps from inquiry to a production CambridgeNexus deployment
Step Stage What happens
Step 1 Inquiry

You send rack count, term, location preference and workload to the CambridgeNexus team.
Step 2 Qualification

Commercial review and technical discovery run in parallel.
Step 3 Site proposal

CambridgeNexus proposes the facility fitting your workload, latency, and compliance requirements. Countries and states are listed on our website, while the location of your data center will be disclosed in the contract.
Step 4 Contract and downpayment

The contract lasts from 6 months to 5+ years. Downpayment is the default. 
Step 5 Installed and operational

75-90 days (or faster, depending on rack availability) from contract signing to launch.
Step 6 Ongoing operations

Monitoring, orchestration, SLA-backed support and quarterly reviews for the life of the lease.

FAQ

AI Factory as a Service is the CambridgeNexus operating model. You lease whole NVIDIA GB300 NVL72 racks or B300 systems; CambridgeNexus operates every layer around them under one contract: power, cooling, networking, compute, orchestration, compliance and workload planning. See the GB300 NVL72 rack

CambridgeNexus operates orchestration and utilization management as one of the seven layers. Scheduling and capacity planning are built into the lease.

CambridgeNexus operates the facility and six layers around your hardware: power, cooling, networking, compute, orchestration, and compliance. You manage what makes the AI yours: your workloads, your models, your data. Every lease is a whole bare-metal NVIDIA GB300 rack. One customer per rack; data isolation is physical, enforced by the hardware itself.

75-90 days from contract to installed and operational, or faster depending on rack availability. Racks are pre-manufactured and allocation is secured, so your timeline is a delivery schedule: every stage has a named owner and a date in your contract. How a CambridgeNexus deployment runs

Four things: rack count, contract term, residency or latency requirements, and your workload. We return with one document containing rate, capacity, site, and committed dates. Reserve capacity

Reserve your GB300 capacity

Tell us the rack count, the term, location, and your workload requirements. We'll come back with the deployment plan: capacity, site, and dates.

A technical conversation, from the first call