Trust
We rent out GPU capacity to a defined set of customers. The page below describes how we decide who we work with and what we ask of them.
Who we work with
We work with European AI companies, private financial institutions, and research organisations. Customers come to us through introductions or direct outreach. We do not run a self-service platform, and we do not take customers we cannot identify and verify.
Who we don't work with
We do not provide services to:
- government, military, defence, intelligence, or law-enforcement bodies
- surveillance or mass facial-recognition projects
- autonomous weapons development
- customers in jurisdictions restricted under U.S., Swiss, EU, or UN sanctions
- workloads prohibited by NVIDIA's end-use policies
If a customer's use case falls outside what we support, we say so up front.
Onboarding
How we onboard.
Before we grant access, we go through these steps. It usually takes a few days. We don't compress this for speed.
Entity verification
Corporate documents and beneficial ownership.
Sanctions screening
Screening against the relevant restricted-party lists.
Workload review
A direct conversation about the intended use case.
Written agreement
Acceptable use, no sublicensing, no re-export.
Export control
Our hardware is NVIDIA, and we operate it under NVIDIA's end-use terms. We pay attention to U.S., Swiss, and EU export-control rules — particularly when a customer is operating across borders. If a workload is sensitive, we'd rather pause and clarify than process it and find out later.
Hardware ownership
PT Gede Artha Samudraowns the hardware. We don't resell it, lease it to other operators, or transfer it. Customers get access to compute under contract — they don't get the equipment.
Ongoing
Trust isn't only an onboarding step. We can suspend or terminate access if we discover a misalignment with the agreement, the sanctions environment changes, or we have reasonable grounds to believe a customer is operating outside what was disclosed.