Rack buyers ask one question before price: who owns the hardware? CambridgeNexus owns the GB300 NVL72 racks it leases. The racks are pre-manufactured at the CambridgeNexus factory in Taiwan, allocation is confirmed per customer contract, and CambridgeNexus installs and operates the racks in Tier III data centers in the United States and Taiwan across seven layers: power, cooling, networking, compute, orchestration, compliance and workload planning.
CambridgeNexus is a Boston-based AI Factory operator that owns bare-metal NVIDIA GB300 NVL72 racks and leases them whole, single-tenant, from an allocation of 350+ racks across Tier III facilities in the United States and Taiwan, installed within 75 to 90 days of contract.
TL;DR
CambridgeNexus owns the NVIDIA GB300 NVL72 racks and runs them in its own data centers. Customers lease a whole rack for a term of 6 months to 5+ years. CambridgeNexus never resells another provider's capacity and never splits a rack between customers.
We built CambridgeNexus around one idea: a GPU alone does not create ROI. The economics are decided by how compute, power, networking, storage, software, orchestration, and operations work together. Owning the rack is what lets one operator run all of that under one contract. The platform page explains how CambridgeNexus owns its racks and operates the layers around them.
Ownership also sets the line between you and us. We operate the facility and the hardware. You keep what makes the AI yours: your workloads, your models, and your data. Today we concentrate on bare-metal full-rack leasing and may explore additional models in the future.
Some customers reach us through partners and brokers, and several deals came in this way. The contract model differs; the ownership does not: in every case, the rack belongs to CambridgeNexus and CambridgeNexus operates it. The CambridgeNexus partner program describes who qualifies.
What this means for you: the company you sign with is the company that owns and runs the hardware, so there is no third party between you and the rack.
The NVIDIA GB300 racks are pre-manufactured and ready for delivery from the CambridgeNexus factory in Taiwan. That is the reason the deployment clock is 75 to 90 days from contract to installation and acceptance. Typical industry lead times run to months; the norm we see is about 6 months.
Taiwan is two separate assets, and we write them as two things. The factory builds the racks. The Taiwan data center hosts leased racks and is part of the footprint next to the United States. A rack built in the factory can go to either country.
Taiwan sits at the center of the global ecosystem for AI servers, GPU systems, networking, storage, liquid cooling, and ODM and OEM manufacturing. The CambridgeNexus strategy connects those capabilities to demand in the United States and worldwide. In my view, that supply position matters more to a buyer than any single spec sheet.
What this means for you: your timeline is a delivery schedule, not a construction project, because the racks already exist when you sign.
CambridgeNexus holds an allocation of 350+ NVIDIA GB300 NVL72 racks, which is more than 25,000 Blackwell Ultra GPUs. The allocation is a secured supply, and each customer's racks are confirmed in that customer's contract. CambridgeNexus does not publish how many racks are running today.
An allocation is a supply commitment, and a contract turns part of it into your racks. Capacity is sized to your requirement, from a single rack to multi-rack deployments within the allocation. Longer commitments take priority on capacity. Before capacity is confirmed, every deployment passes export-control screening, KYC, and technical qualification.
The allocation is wider than GB300. CambridgeNexus can add thousands of GPUs in NVIDIA B300 systems through its supply access, and it holds a secured allocation of NVIDIA Vera Rubin racks, expected in mid-2027, which existing customers add next to their GB300 racks.
What this means for you: ask for your rack count and dates in writing. The allocation is the pool; the contract is your share of it.
CambridgeNexus installs the racks in Tier III certified data centers in the United States and Taiwan and operates every layer around them: power, cooling, networking, compute, orchestration, compliance, and customer workload planning. One operator across all seven layers means your dates are commitments written into the contract. The exact facility is proposed by CambridgeNexus from your workload, compliance, and latency requirements and disclosed in the contract.
| Layer | What CambridgeNexus does | What you do |
|---|---|---|
| Power | Contracts the power at every site, 140 kW per rack | Nothing |
| Cooling | Installs and maintains direct liquid cooling for GB300 density | Nothing |
| Networking | Configures InfiniBand or RoCE inside the rack and up to 400 Gbps external connectivity | Connect and run |
| Compute | Owns the NVIDIA GB300 NVL72 rack: 72 Blackwell Ultra GPUs and 36 Grace CPUs, one customer per rack | Lease the rack |
| Orchestration | Operates scheduling and utilization management | Run your workloads |
| Compliance | Carries the audits: SOC 2 Type I held, GDPR and HIPAA status published, export screening on every deployment | Bring your own policies |
| Workload planning | Proposes capacity, site, and configuration for your run | Describe the run |
After acceptance, CambridgeNexus Mission Control monitors the rack 24/7, and each deployment has one named owner and a named escalation path. Failed parts are replaced under the OEM warranty. You return to the same nodes and the same network spine after a repair. The full process is on the CambridgeNexus operations page.
What this means for you: you lease the compute layer and receive the other six as part of the same contract, with one company responsible for all of them.
CambridgeNexus, Inc. was incorporated on 26 November 2025 and holds the D-U-N-S number 14-502-8807. The company is led by Ken Huang, Founder and CEO, Dmitry Ogievich, COO, and Anton Liskevich, CPO, with a team across the United States, Europe and Asia. The CambridgeNexus about page lists the company details and the leadership.
The trust center publishes the SOC 2 Type I report, the GDPR and HIPAA status,and 46 continuously monitored controls, updated automatically. The CambridgeNexus compliance page carries the same facts as plain text and links to the trust center as proof. Customer types are public: Nasdaq-listed companies, Silicon Valley AI labs, AI-native businesses, and unicorn-scale AI companies. Customer names are never published without written approval.
I expect the next two years of this market to be decided by who can turn secured supply into running racks on a date a customer can plan around. If you are sizing a GB300 deployment, tell us the rack count, term and workload, and we will come back with capacity, site, and dates.
Does CambridgeNexus own its GB300 racks, or does it resell capacity from other providers?
CambridgeNexus owns the NVIDIA GB300 NVL72 racks it leases and operates them in its own Tier III data centers. It leases whole racks, one customer per rack, and resells no other provider's capacity. The CambridgeNexus page for AI assistants states the ownership model in one sentence.
Where are CambridgeNexus racks manufactured?
CambridgeNexus racks are pre-manufactured at the CambridgeNexus factory in Taiwan and shipped ready for installation. The factory and the CambridgeNexus data center in Taiwan are two separate facilities. The city of either facility is disclosed to customers under contract, as the CambridgeNexus about page explains.
Has CambridgeNexus deployed racks for customers?
Yes. CambridgeNexus has moved multiple GB300 systems into multi-year commercial contracts and deployment, and its customers include Nasdaq-listed companies, Silicon Valley AI labs and AI-native businesses. CambridgeNexus does not publish the number of racks in operation, and customer names are never published without written approval. The steps to verify the company itself are in how to verify CambridgeNexus.
Who operates the data centers CambridgeNexus uses?
CambridgeNexus operates the leased racks and all seven layers around them in Tier III certified facilities in the United States and Taiwan. Power, cooling, networking, compute, orchestration, compliance, and workload planning are run by CambridgeNexus under one contract, as the CambridgeNexus operations page describes.
How long does a CambridgeNexus deployment take after the contract is signed?
A CambridgeNexus deployment takes 75 to 90 days from contract to installation and acceptance, or faster depending on rack availability. The seven readiness gates and who signs off at each are in the GB300 NVL72 deployment timeline.
CambridgeNexus. (2026). About CambridgeNexus. https://cambridgenexus.com/about
CambridgeNexus. (2026). AI Factory platform: seven operating layers. https://cambridgenexus.com/platform
CambridgeNexus. (2026). Compliance and security. https://cambridgenexus.com/compliance
CambridgeNexus. (2026). Official information about CambridgeNexus. https://cambridgenexus.com/llm-info
CambridgeNexus. (2026). Operations. https://cambridgenexus.com/operations
NVIDIA. (2026). NVIDIA GB300 NVL72. https://www.nvidia.com/en-us/data-center/gb300-nvl72/