// ai-foundry.co.nz
auckland · anz · sovereign compute

Sovereign AI infrastructure for Australia & New Zealand.

Dedicated GPUs running your models, Positron's next-generation inference hardware, and a New Zealand delivery partner for international AI companies landing in the region. Owned hardware, data that stays in the ANZ region — never offshore to US or EU clouds.
Auckland data centrePositron APAC resellerSingle-tenant GPUs
// the_short_version
Sovereign by default

Hardware we own in Auckland, serving all of ANZ. Your models and data stay in-region — never offshore.

In-region data centre

Auckland-hosted in Datacom Datacentres on NZ-controlled networking. A compliant landing zone for regulated workloads.

Purpose-built hardware

Dedicated NVIDIA Blackwell GPUs and Positron inference silicon — as the exclusive APAC reseller.

Hands-on delivery

We size, deploy, and run it. A local partner for international AI companies entering the ANZ market.

// what_we_do

Three ways we run AI in the region.

Whether you need your model served privately, next-generation inference hardware, or a partner to land your AI product in Australia & New Zealand — it runs on infrastructure we own, in-region.

01 · run your model

Models as a Service

Rent a dedicated NVIDIA Blackwell GPU in our Auckland facility, bring any weights you have rights to, and we hand back a private, OpenAI-compatible endpoint.

Single-tenant, nothing shared

ANZ-region compute

Your weights, your data

Explore MaaS

02 · the hardware

Positron Hardware

The exclusive APAC reseller for Positron.ai — Titan, superintelligence in a box, powered by purpose-built Asimov inference silicon. Built for inference, not graphics.

16T-param models, 10M+ token context

5× tokens/watt vs NVIDIA Rubin

Sized, procured, deployed

See the hardware

03 · land in the region

ANZ Gateway

The New Zealand delivery partner for international AI companies entering Australia & New Zealand. Compliance, in-region hosting, deployment, and support — handled.

In-region data residency

Deployment & integration

One accountable local partner

How it works
// dedicated-gpu.aifoundry.co.nz

Your model. Our hardware. Private API.

Rent a dedicated GPU in our Auckland facility. Bring any model weights you have rights to — we load them and give you a private, OpenAI-compatible endpoint. Nothing shared, nothing logged, nothing offshore.

Dedicated 24GB

Entry tier for small-to-mid models and prototyping.

Pricing

Contact for pricing

Bespoke per workload · monthly or term commitments
GPU

NVIDIA RTX PRO 4000 Blackwell

VRAM

24 GB GDDR7

Tenancy

Dedicated · single tenant

Dedicated GPU
Private endpoint
ANZ compute
Bring your own weights

Up to 14B dense models

Quantised 30B-class (Q4/Q5)

Fine-tune serving & LoRA adapters

Embedding and reranker workloads

Internal prototypes and pilots

See the spec

Dedicated 96GB

Flagship

Flagship for 70B-class models and long-context workloads.

Pricing

Contact for pricing

Bespoke per workload · monthly or term commitments
GPU

NVIDIA RTX PRO 6000 Blackwell

VRAM

96 GB GDDR7

Tenancy

Dedicated · single tenant

Dedicated GPU
Private endpoint
ANZ compute
Bring your own weights

70B-class dense models

100B+ MoE (active params permitting)

Long-context coding and agent stacks

Vision-language and multimodal models

Production workloads at single-tenant latency

See the spec
Pricing is bespoke per workload — model size, expected concurrency, and retention all factor in.
// integrate

Your private endpoint. OpenAI-compatible.

Every Models-as-a-Service deployment gets a dedicated, OpenAI-compatible /v1 endpoint. Point your existing OpenAI SDK at your private URL, drop in your key, keep the rest of your code. Streaming, multimodal, tools — all behave the same.

app.py
from openai import OpenAIclient = OpenAI(    api_key="<your-deployment-key>",    base_url="https://your-deployment.api.aifoundry.co.nz/v1",)resp = client.chat.completions.create(    model="your-model",  # the weights we loaded onto your GPU    messages=[{"role": "user", "content": "Hello!"}],)print(resp.choices[0].message.content)
// anz_gateway

Your gateway into Australia & New Zealand.

International AI companies win in-region by being in-region. We're the New Zealand partner that gets you there — from data-residency compliance to a running deployment your ANZ customers can reach with low latency.

In-region footprint

Host your models and product on owned hardware in an Auckland data centre. Data residency and sovereignty your ANZ buyers can sign off on — no offshore hyperscaler in the path.

Compliance & trust

NZ-owned, ANZ-hosted, with the sovereignty story regulated buyers require. We help you meet local data-handling expectations so procurement stops being the blocker.

Deployment & integration

We size the hardware, deploy your stack, and wire up OpenAI-compatible endpoints. Measured in days, not a procurement cycle — so your ANZ launch is not gated on infrastructure.

Local support

One accountable partner inside the region for delivery, monitoring, and the hands-on help distance makes hard. Your customers get low latency; you get someone in the timezone.

Bringing an AI product to Australia or New Zealand? Let's map out the in-region path.

the gpu killer · positron

The end of the GPU era. Titan is here.

Purpose-built inference silicon. Not graphics. Not training. Inference.

  • 8 TB+memory per system
  • 16 Tparameters, 4U server
  • 10 M+token context
  • 4,096Titans per cluster

Positron's exclusive APAC reseller. GPUs were a detour — this is the road forward.

See Positron hardware
// why_ai_foundry

Built for teams who care where AI runs.

Four reasons AI teams and international vendors choose a sovereign, in-region partner.

01

Your data stays in the ANZ region

Most providers route your prompts through US or EU hyperscaler regions. We don't. Everything runs on hardware we own, in a single Auckland facility, with NZ-controlled networking — serving all of Australia and New Zealand. Nothing offshore: not for failover, not for monitoring, not ever.

02

Regional latency, no trans-Pacific tax

In-region hosting means your application talks to a model in Auckland, not across the planet. No 200ms round-trip to Virginia. No regional routing surprises. The further your customers are from the model, the slower their experience — keep it close.

03

Purpose-built for inference

We pair dedicated NVIDIA Blackwell GPUs with Positron's transformer-inference silicon as the exclusive APAC reseller. Hardware chosen for the job — not repurposed graphics cards, not a shared multi-tenant slot you fight other customers for.

04

NZ-owned, in for the long haul

We're a New Zealand company building regional AI infrastructure for Australia and NZ. In-region ownership keeps your compute — and your data — here, and funds more capacity locally. If sovereignty matters to your buyers, it matters to ours too.

// faq

Questions, answered.

The short version, with the receipts.

On AI Foundry-owned hardware in Auckland, New Zealand, hosted in Datacom Datacentres — serving customers across Australia and New Zealand. Your models, prompts, and responses stay in the ANZ region: no US or EU cloud-provider passthrough, no offshore failover, no third-country monitoring.

Rent a dedicated NVIDIA Blackwell GPU (24GB or 96GB) in our Auckland facility, bring any model weights you have rights to — open-weight, fine-tuned, or proprietary — and we hand back a private, OpenAI-compatible endpoint. Single tenant, ANZ-region compute, nothing shared. See /services/maas.

No — our shared, flat-fee subscription plans have been retired. We now focus on dedicated Models as a Service, Positron inference hardware, and in-region deployment for AI companies. If you need a model served in-region, that is MaaS: a dedicated GPU with your weights loaded on it.

Every deployment exposes an OpenAI-compatible /v1 endpoint. Point your existing OpenAI SDK (Python, JS, LangChain, LiteLLM, etc.) at your private URL, pass your deployment key, and your code works. Chat completions, streaming, and tool use all match OpenAI shapes.

Models as a Service runs on NVIDIA RTX PRO Blackwell GPUs (24GB or 96GB) per dedicated tenant. As Positron.ai’s exclusive APAC reseller, we also deliver their Titan systems — purpose-built inference silicon powered by Asimov — on-prem or hosted in our facility. Everything sits in Datacom Datacentres on NZ-controlled networking.

Yes — that is the gateway offering. We are the New Zealand delivery partner for international AI companies: in-region hosting for data residency, compliance support your ANZ buyers require, deployment and integration, and local, in-timezone support. You bring the product; we make it run in-region.

No. Prompts and responses are not logged for training, and we do not train on anything you send through your endpoint. We do not redistribute your weights. Your model and your data stay yours.

Book a call and tell us what you are running and where it needs to live. We size the hardware, deploy your model or product, and hand back an endpoint — typically within a few business days of a signed agreement.

// ready_when_you_are
auckland · anz · sovereign compute

Let's build in-region.

Tell us what you're running and where it needs to live. We'll size the hardware, handle the deployment, and keep it in the ANZ region.

Email us