Enterprise AI servers, sourced and deployed with you

We help you source and deploy systems for South African sites. The route and the delivery cost depend on the system and where it is going, so we confirm this when we talk.

Lead times vary by system and configuration, and we would rather not print one here that may not hold. We confirm this when we talk.

NVIDIA's datasheets list a three-year support tier for the DGX B200 and DGX B300. Warranty on the GPU servers, and how local support works for your system, we confirm this when we talk.

Yes. On a GPU server you choose the GPUs, processor, memory and storage, and on a DGX system the support term and NVIDIA AI Enterprise. If you need something that isn't listed, tell us, and we confirm this when we talk. Some GPU and chassis pairings are marked "Compatibility confirmed on your quote", which means we check them before you commit.

We don't publish any. If you are a university, school or start-up, say so on the form. Whether special pricing applies, we confirm this when we talk.

The configurator gives you an instant indicative price for the build you pick, excl. and incl. VAT. A quote is where we confirm the final price, compatibility and terms for that build. Configuring a system doesn't commit you to anything.

Tell us who needs to learn what: running the system, deploying models, or building on them. What we can offer your team, we confirm this when we talk.

A DGX B300 draws up to 15.1 kW in 10 rack units, according to NVIDIA's datasheet, so power, cooling and rack space need planning early. We can talk through your site and the network around the system. How far the design work goes on your project, we confirm this when we talk.

Yes, GPU as a Service is the option to ask about if you would rather not own or house a server. Availability, location and pricing, we confirm this when we talk.

These systems are imported, so their cost moves with the exchange rate, freight and clearing. We show indicative prices in rand, excl. and incl. 15% VAT, so you can plan with them. Your final price is confirmed on your quote.

Running models on your own hardware, on your own site, means you decide where your data sits and who can reach it. That helps, but it doesn't make you POPIA compliant by itself, and we don't give legal advice. For the cloud option, where data is held, we confirm this when we talk.

Deployment is part of what we help with, and we can set up open-weight models and the software that runs on them. Who handles the physical installation, and what it costs, we confirm this when we talk.