AMD Developer Cloud Is White-Labeled DigitalOcean | Generated by AI
Question: How does AMD Developer Cloud work with DigitalOcean? Looking at this snapshot deletion script that uses api.digitalocean.com, it seems like only the web portal (UI) is different but the API endpoints and backend/database are the same — is that correct?
Answer:
Yes, your inference is correct. AMD Developer Cloud is a white-labeled / OEM’d DigitalOcean deployment. Here’s the breakdown:
What’s Actually Happening
AMD Developer Cloud is built on top of DigitalOcean’s infrastructure platform, with AMD-specific GPU hardware (MI300X etc.) added to DO’s datacenter fleet. The relationship:
AMD Developer Cloud Portal (amd.com/en/developer/...)
↓ (just a themed frontend)
DigitalOcean Control Plane API ← api.digitalocean.com
↓
DigitalOcean backend / database (same droplet/snapshot/volume infra)
↓
AMD GPU hardware nodes in DO's Atlanta datacenter
Evidence From Your Script
Your script hits https://api.digitalocean.com/v2 with AMD_DEV_CLOUD_API_KEY — a DigitalOcean Personal Access Token issued through the AMD portal. The fact that this works confirms:
- The API is 100% standard DigitalOcean v2 API — no AMD-specific endpoints
- The token is a real DO PAT scoped to an AMD-provisioned DO account/project
- Snapshots, droplets, volumes all live in DigitalOcean’s backend database — AMD just provisions a DO account for you
What AMD Adds
| Layer | Who owns it |
|---|---|
| Web portal UI | AMD (themed, different from cloud.digitalocean.com) |
| GPU droplet SKUs (MI300X) | AMD (hardware + BIOS/driver config) |
| API endpoint | DigitalOcean (unchanged) |
| Account/billing | AMD (they resell DO capacity) |
| Snapshot/volume/networking DB | DigitalOcean |
| Support tier | AMD for GPU issues, DO infra underneath |
Practical Implication
Since the API is identical, any DigitalOcean SDK or tool works against AMD Dev Cloud with zero modification — just swap the token:
import digitalocean # pydo or python-digitalocean
manager = digitalocean.Manager(token=os.environ["AMD_DEV_CLOUD_API_KEY"])
snapshots = manager.get_all_snapshots()
droplets = manager.get_all_droplets() # lists your MI300X droplets
Same for doctl:
doctl auth init --access-token $AMD_DEV_CLOUD_API_KEY
doctl compute droplet list
doctl compute snapshot list
Your snapshot deletion script is a clean, correct implementation — the only “AMD-specific” thing is the token source. Everything else is vanilla DO API.