Vantage Compute vs Hosted HPC Services
Sending research out to be hosted moves data outside your boundary. Vantage Compute runs on hardware you own.
| Hosted HPC services | Vantage Compute | |
|---|---|---|
| Where workloads and data run | On the provider's infrastructure | On your own racks or in your own cloud account |
| Identity | Varies by provider | Your own identity provider (SSO and RBAC), one identity on every cluster |
| Access policy | Varies by provider | Deny-by-default, then granted per role before a job runs |
| Audit trail | Varies by provider | Every action logged and tied to one identity |
| GPUs you already own | Varies by provider | Idle GPUs found and shared between teams |
When each option fits
Hosted HPC services: the better fit when
- You have no hardware and no plan to buy any.
- Your data is allowed to leave your boundary.
- You need capacity for a short project and have nobody to run it.
Vantage Compute: the better fit when
- Data has to stay inside your boundary, on hardware you own or in your own cloud account.
- You already own GPUs and want them used.
- You have to show who ran what, where, and on whose approval.
Frequently asked questions
Does data leave our boundary with Vantage Compute?
Data never leaves your boundary unless you choose to use cloud. Your models, your hardware, your jurisdiction.
Can Vantage Compute still use cloud capacity?
Yes, when you choose. The same control plane runs on AWS, Azure, Google Cloud or your own racks: burst out, stay in region, or run fully on-prem.