Modular infrastructure for
private AI compute.
AIZ designs modular compute enclosures for organisations that need dedicated AI capacity, clearer infrastructure control, and a faster route than conventional facility-led buildouts. Alongside distributed edge nodes, the platform direction includes engineering adaptation for next-generation inference servers using GPU, NPU, FPGA, and ASIC accelerators.
Based in Christchurch, New Zealand, AIZ is positioned for customers evaluating cleaner power contexts, stronger infrastructure control, and a more regional deployment model.
Start with the deployment path if you already have a workload, site, power context, or private infrastructure requirement in view.
AIZ has also announced a Phase 1 R&D funding agreement with New Zealand’s Ministry of Business, Innovation and Employment (MBIE) under the New to R&D Grant Scheme. See newsroom update.
Why private AI infrastructure needs a different delivery model.
| Metric | Traditional DC | AIZ Modular Deployment Model |
|---|---|---|
| PUE | 1.4–1.8 | Air target <1.19 · Thermal target <1.12 |
| Form factor | Rigid custom builds | Modular enclosure · standardised deployment footprint |
| Deployment | 12–24 months | Weeks — pre-engineered site deployment |
| Cooling | Legacy CRAC / CRAH | Purpose-designed thermal architecture |
| Scalability | Proprietary lock-in | Modular — expand by repeating validated node templates |
| Carbon intensity | High | Designed to align with efficient power availability |
Indicative engineering targets currently referenced by AIZ include Air Node performance below 1.19 PUE and Thermal Node performance below 1.12 PUE. Final deployment outcomes depend on configuration, validation scope, and site conditions.
AI deployment infrastructure
must be more deployable, more repeatable, and more site-aware.
Organisations deploying modern AI workloads often need dedicated compute capacity on timelines and site conditions that conventional data-centre delivery models do not always serve well.
Deployment constraints are not only about servers. They also involve power access, site readiness, thermal control, and the practical work required to bring infrastructure into service.
AIZ focuses on the physical infrastructure layer: modular enclosures, controlled thermal behaviour, and repeatable deployment methods for private AI infrastructure.
Built for organisations evaluating dedicated AI infrastructure.
Regional Operators
Operators that need deployable GPU infrastructure without the time, permitting burden, or cost structure of a conventional facility project.
Private AI Labs
Teams that need dedicated infrastructure control, predictable workload placement, and a clear path to private AI capacity outside shared-tenancy environments.
Industrial Sites
Industrial and warehouse environments with suitable floor area, power access, and operational constraints that favour enclosure-led deployment.
Energy-Adjacent Projects
Projects evaluating how site and power availability can be aligned with modular AI infrastructure in regional or energy-linked deployment contexts.
Where the current AIZ model is most relevant.
Private Inference Capacity
For organisations that need dedicated infrastructure for inference, data-sensitive workloads, or controlled model-serving environments.
Site-Constrained Deployments
For deployments where speed, floor-space realities, or retrofit constraints make a conventional data-centre build impractical.
Regional AI Infrastructure
For operators exploring dedicated AI infrastructure closer to power availability, workload demand, or regional service requirements.
Next-Generation Inference Servers
For server and solution partners adapting NPU, FPGA, or ASIC-based inference platforms to defined rack, power, thermal, network, and operational requirements.
From distributed node deployment to energy-linked AI infrastructure.
Deploy compute closer to site and power.
AIZ is building a distributed node model that combines modular compute enclosures, repeatable deployment standards, and site-level infrastructure planning. The objective is to make AI infrastructure easier to place where suitable power, sites, and workload demand already exist.
Convert power availability into AI infrastructure value.
For selected deployments, AIZ sees an infrastructure pathway in linking energy access, modular infrastructure, and AI workload production. This energy-linked infrastructure model can include energy-to-compute arbitrage opportunities where operators or site owners want to turn available power and land into productive AI infrastructure capacity.
Structured discussions for site-led and strategic opportunities.
AIZ can explore partnership pathways for energy owners, site owners, regional operators, and strategic counterparties. These discussions may include deployment planning, distributed node participation, and structured partnership models where there is a clear infrastructure fit.
If you control site and power, AIZ can help evaluate the infrastructure path.
AIZ is not limited to selling a modular compute enclosure as a standalone product. We also evaluate how site access, power availability, and distributed node deployment can form part of a broader infrastructure pathway.
If you are evaluating whether available land, power, or site capacity can support AI infrastructure, AIZ can assess whether a distributed node pathway is technically, operationally, and commercially relevant.
Request Partnership DiscussionCommon questions about the AIZ platform.
What does AIZ provide?
AIZ provides modular AI infrastructure centred on enclosure-led private compute deployments, including distributed edge nodes and an engineering adaptation pathway for inference servers using GPU, NPU, FPGA, and ASIC accelerators.
What deployment environments does AIZ target?
AIZ focuses on industrial, commercial, warehouse, and regional environments where dedicated AI compute needs to be deployed faster and with less site-specific construction than a conventional data-centre buildout.
What PUE targets does AIZ communicate publicly?
AIZ publicly describes indicative engineering targets including an Air Node target below 1.19 and a Thermal Node target below 1.12. These figures are not universal guarantees and depend on configuration, validation scope, and operating conditions.
Have a site, workload, or deployment path in view?
Start with an infrastructure assessment.
The best early inputs are expected workload, likely available power, site type, timing, and whether the opportunity is a direct deployment or a broader site-led infrastructure pathway.