DFINITY pitches 'programmable cloud' model
DFINITY announced Cloud Engines that promote a programmable-cloud approach where nodes can be scaled by region, private infrastructure participation is supported, and roughly 80% of revenue can flow directly to providers. The pitch reframes cloud providers as a decentralised marketplace rather than a single rented stack (x.com).
DFINITY is trying to turn cloud computing into something that looks less like renting one company’s server farm and more like buying capacity from a market of competing operators. In its new “cloud engine” push, it says a customer could choose where nodes sit, including a single region, and still run on Internet Computer infrastructure. (dashboard.internetcomputer.org) The pitch is not just “decentralized cloud” in the abstract. DFINITY’s own governance proposal describes a cloud engine as a private subnet with a control panel, where the owner can pick replication settings and choose nodes in one geographic region for legal or performance reasons. (dashboard.internetcomputer.org) That is a direct answer to a very old enterprise complaint about blockchains. Most chains spread workloads everywhere by default, while most companies need to know which country or region their data and compute live in before legal teams will sign anything. (dashboard.internetcomputer.org) DFINITY is also promising switchable infrastructure underneath the app. Its proposal says a cloud engine could start across Amazon data centers, move to Google, use a mix of both, and later shift onto traditional node-provider hardware without interrupting the hosted application. (dashboard.internetcomputer.org) That is why the company keeps using the word “sovereignty.” On the Internet Computer homepage, DFINITY now frames the product as a way to escape vendor lock-in by running on the network or on-premises, instead of being trapped inside one cloud provider’s stack. (internetcomputer.org) The money model changed to match that story. In February 2026, DFINITY proposed replacing fixed node rewards for this new category with a revenue split where 80 percent of cloud engine revenue goes to the nodes providing compute and 20 percent is used to burn Internet Computer tokens. (internetcomputer.org, edgen.tech) That makes node operators look less like passive validators and more like cloud suppliers competing for customers. The governance text says associations of node providers will be able to market cloud engines based on their own nodes, in the same way internet service providers sell access over their own network segments. (dashboard.internetcomputer.org, dashboard.internetcomputer.org) This is also tied to DFINITY’s 2026 “Mission 70” plan. The white paper says the foundation wants to cut Internet Computer inflation by at least 70 percent by the end of 2026, and one of the main demand-side bets is that cloud engines will drive more token burn through real compute, storage, and bandwidth usage. (internetcomputer.org) There is already a concrete test of the idea on-chain. A proposal opened on April 8, 2026 to create the first test cloud engine with four cloud nodes spread across Amazon Web Services, Google Cloud, and Microsoft Azure, specifically to test protocol changes like firewall isolation and cross-network shielding. (dashboard.internetcomputer.org) So the real claim here is bigger than a product launch. DFINITY is saying cloud should be programmable at the infrastructure layer too: pick the region, pick the operator mix, move the workload later, and let the revenue flow mostly to whoever actually supplies the machines. (dashboard.internetcomputer.org, internetcomputer.org)
