High-performance compute. Hosted in Ireland.

Your own GPUs, or our model API.

Dedicated sovereign compute

Your own physically isolated RTX PRO 6000 GPUs, hosted in Ireland and operated by us. For work where data privacy, compliance and performance are not negotiable.

• Physically isolated, single-tenant hardware
• Inference and data stay in Ireland
• Zero data retention by design
• Fixed monthly or contracted pricing
• Any model, full control over the stack
• Bare metal, private AI and the API on one account

Sovereign model API

Open models behind one OpenAI-compatible API, on our own GPUs in Ireland. Switch from a closed frontier API by changing one line, and pay a fraction of the price.

• Drop-in OpenAI compatibility: one line of code
• A fraction of the cost of closed frontier APIs
• The right open model for each job, not one size for all
• Zero data retention on every request
• Pay per token, or commit to a monthly volume
• Shared API, or a dedicated endpoint when you grow

from openai import OpenAI

# Point your existing client at tandemcompute.
# Inference in Ireland, zero data retention.
client = OpenAI(
    base_url=TANDEMCOMPUTE_API_URL,  # issued with your API key
    api_key="YOUR_API_KEY",
)

reply = client.chat.completions.create(
    model="qwen/qwen3.6-35b-a3b",
    messages=[{"role": "user", "content": "Summarise this contract."}],
)

Sovereign by default

Inference and data stay in Ireland, under EU jurisdiction. Zero data retention: prompts and outputs are never stored or used for training.

Dedicated when it matters

Whole GPUs, never sliced or oversubscribed, for bare metal and dedicated endpoints. Start on the shared API and move to your own hardware on the same API.

Priced below the frontier

Open models at a fraction of what closed frontier APIs charge. Contract rates for committed capacity and volume.

Accelerated compute, powered by NVIDIA.

NVIDIA RTX PRO 6000 Blackwell

A universal GPU for AI inference, fine-tuning, rendering and simulation. 96 GB of memory with fifth-generation Tensor Cores and fourth-generation RT Cores, in Dell servers at every site.

Open models, kept in Ireland.

Data stays in Ireland

Inference runs on tandemcompute hardware in an Irish Tier III facility.

Zero data retention

Prompts and outputs are never stored, and never used for training.

Drop-in OpenAI compatibility

Point your existing OpenAI client at tandemcompute. One line of code changes.

A fraction of the cost

Well below closed frontier APIs, with the right model for each job.

Committed volumes

Monthly token commitments, or dedicated endpoints at contract rates.

One supplier

The same account covers bare metal, private AI and the API.

The rest of the pipeline, beside your GPUs.

CPU-only instances on the same private network as your GPU instances, so GPU capacity stays free for training and inference.

see cpu pricing →

AI applications and agents

App backends, serving logic and orchestration that sit alongside your GPU workloads.

Data preprocessing

Tokenisation, feature engineering and data loading on dedicated CPU.

Offline and batch inference

Bulk evaluations, document processing and inference that is not latency-critical.

Automation and tooling

Evaluation harnesses, scheduled jobs, pipeline scripts and CI/CD.

More than instances. Run, build, scale.

Run

Serve and fine-tune models in production

Build

Train and serve custom models

Scale

Dedicated capacity in Ireland