Platform tour
One protocol. One portal. Infrastructure you finally own.
OpenCloud is a portal built on top of the Internet Computer Protocol. It lets you compose independent compute nodes into a sovereign cloud platform tailored to your requirements — and then manage it, without a systems team.
What you get
A cloud engine is a fully-managed compute platform you own.
Your own execution environment
Motoko-authored apps — called canisters — run inside a virtualized runtime created by your chosen nodes. Only your code can touch your data; there is no shared operating system to compromise.
One-click deploys for agent-built apps
Caffeine, Claude Code, and other builders produce a signed .icp file. Drag it into your control panel and it deploys atomically across every node.
Usage-based, prepaid billing
Engines run against a prepaid balance of cycles. Top up with a credit card. No surprise invoices, no capacity-planning meetings.
A console built for reading
Live node health, per-canister cycle burn, compute history over hours or weeks, and one-button moves between operators.
From first engine to first ship
- 1.Sign in with a passkey.
- 2.Pick a starter node class. Small engines: ~$2,500/mo. Large production engines: ~$15k+.
- 3.Pick nodes from providers that match your geography and compliance needs.
- 4.Top up cycles with a card. Track burn in real-time.
- 5.Install an app, or connect Caffeine.ai and chat your first feature live.
What a cloud engine is not
- · Not a VM or container orchestrator.
- · Not a serverless function platform.
- · Not a managed database.
- · Not a place to host legacy Node/Python/Rails apps.
It is a new substrate for AI-built software — simpler, safer, and native to the way modern agents write code.
Compared to
How cloud engines stand apart.
| Capability | Traditional cloud (IaaS) | Serverless / PaaS | Cloud engine |
|---|---|---|---|
| Tamperproof from the OS down | No — patched OS stack | No — platform-trusted | Yes — no OS below protocol |
| Moves between providers without downtime | Weeks of migration | Vendor-locked | Seconds — swap nodes live |
| Rejects lossy data migrations | Your problem | Your problem | Verified by platform |
| Required systems team | Yes | Smaller, still needed | None |
| Billing model | Per-instance | Per-request | Prepaid cycles |
| Native app unit | Container / VM | Function / endpoint | Canister (code + state) |
Interoperation
Your canisters can talk to the wider network.
Canisters you choose to expose publicly can be addressed by canisters on other cloud engines — forming a larger, composable application surface. Engines running on customer-owned on-prem hardware receive reduced interop permissions, since their security guarantees are weaker.
Interop modelTrust classes
Verified operator engine
Nodes from KYC-verified providers in independent data centers. Full interop.
Cloud-backed engine
Nodes provisioned on AWS, GCP, or Azure by KYC-verified providers. Full interop.
On-prem engine
Customer-run hardware. Reduced interop permissions; cannot be called from others by default.
Ready when you are.
Sign in with a passkey — no credit card required until you top up your first engine.