What Are Cloud Engines?
Cloud Engines are the compute layer that turns a plain-language description into a running, self-owned application on the Internet Computer — no servers to rent, no cloud account to manage, and no vendor holding your data.
01The problem with renting compute
Traditional cloud computing works by renting someone else's computer. You open an account with a provider, hand over a credit card, and get a virtual machine that lives in a data center you will never see. The provider holds the keys: it can read your data, change its prices, suspend your account, or shut the machine down. Your application runs, but it is not really yours. Cloud Engines exist to remove that landlord.
02What a Cloud Engine actually is
A Cloud Engine is a self-contained compute environment that runs on the Internet Computer, a decentralized network of independent node machines spread across many data centers and geographies. Instead of a virtual machine on one company's hardware, a Cloud Engine runs as a canister: a unit of code and state that is replicated across the network and advanced only when a supermajority of independent nodes agree on the result. There is no single operator who can quietly change your code, read your state, or switch you off.
03What Cloud Engines enable
Because a Cloud Engine is replicated, tamper-resistant, and paid for from its own cycles balance, it enables things a rented server cannot. Your application keeps running even if the company that built it disappears, because the network — not a vendor — hosts it. It can hold its own cryptographic keys through chain-key cryptography, so it can sign transactions and act on other blockchains without a bridge or a custodian. It can serve web content directly to a browser, so users reach it over ordinary HTTPS with no wallet or plugin. And it can be upgraded by its owner while its data persists, so the application evolves without a migration project every time.
04Why this matters for sovereignty
The sovereignty thesis is simple: the software you depend on should be controlled by you, not by a gatekeeper who can turn it against you. Cloud Engines are that thesis made practical. You own the canister, you control its upgrade keys, and the network enforces the rules you wrote. No cloud provider can read your users' data, no platform can deplatform you, and no single point of failure can take your application offline. That is the difference between renting compute and owning it.
05How to access and use Cloud Engines
Open Caffeine AI in your browser and sign in with Internet Identity — a passkey-based login that needs no password and no email address.
Describe the application you want in plain language: what it should do, who uses it, and what data it stores.
Review the generated application in the live preview and refine it by describing changes in the same conversation.
When it looks right, deploy it. Caffeine provisions the Cloud Engine for you — the canister, its cycles, and its HTTPS address — so there is no server to configure and no cloud account to create.
Open the deployed URL to use your application, and return to Caffeine any time to describe further changes; each change is applied to the same running engine with its data preserved.
06What you do not have to manage
With a Cloud Engine you do not rent a server, patch an operating system, configure a load balancer, or set up a database. You do not manage TLS certificates or DNS records. You do not pay per request, because the engine holds cycles and pays for its own computation — the reverse gas model. What remains yours is the part that should be yours: the application's logic, its data, and the authority to change it.