Skip to content

The hardware behind your AI

The starting system can fit on a desk. The right hardware depends on the models, the information and how your team uses it.

A compact setup with defined roles

An example setup pairs a DGX Spark with a Mac mini. Your pilot may use a different, comparable configuration. We select the hardware around the workload and agree it before kickoff.

Your private foundation

Illustrative private compute: enclosure, acceleration, memory, storage and system connections.
Illustrative compute
Final hardware selected for your workflows

AI, running on your own hardware.

Compute runs the selected models locally. Capacity depends on the models and the work your team needs them to do.

Example setup / Local model compute

NVIDIA DGX Spark

Runs the selected local AI models. Its unified memory gives those models room to work with the context your workflows need.

Compute
GB10 Grace Blackwell
Unified memory
128 GB
Footprint
150 × 150 mm
Companion system

Apple Mac mini

In this example, Mac mini hosts workflow services, connectors and access tools. The final devices and their roles are confirmed during scoping.

Example hardware specifications: NVIDIA. The interactive drawing is generic illustrative compute, not a DGX Spark model.