Every question, answer first.

Yes. Your private cloud runs on the devices you chose, and the placement is pinned before anything is scheduled. Failover follows the policy you set.

Yes. Triple X connects your racks or devices to a private cloud of your own and you manage them from the same console, API and CLI as our nodes.

No. Most teams start on Triple X Cloud nodes with the free tier and add their own devices later, or never.

Yes, inside one private cloud. Sensitive work on your devices, the rest on ours, with failover between them under your policy.

You choose region and zone at launch. No region list is published yet; the API and the CLI return the live one. Ask Triple X in writing for the current set.

Your private cloud’s own people, on its own private network. The platform underneath schedules and bills: it sees that a workload exists and how big it is, not what is in it.

Confidential inference for data no operator may see. The data stays unreadable while the model runs on it, and a receipt proves the run happened that way. See Sealed.

Nothing of yours is used to train any model. Inside a Sealed run, no one can read it in use, so the promise is a property of the hardware.

The platform operator and the providers that host this site and send email. GPU capacity, when quoted, names its operator before signature. See Security.

The transfer rate and nothing else: from $0.01 per GB out, no exit fee, in the contract. Every product uses a standard API, so you leave with a copy, not a migration project.

By the hour, per machine or node, plus outbound transfer. No minimum term. Budgets with alerts, and spend by service, region and resource. See Pricing.

One small machine for 750 hours a month, 10 GB of object storage, 1 GB of outbound transfer and one shared-CPU database. No credit card.

Tier B is index 90 against hyperscaler on-demand list at 100, Tier D is 70, Tier C 80, Tier A 115. You choose the tier at launch; most work lands on B.

No. The rate card is the rate card, hourly, with no minimum. The commitment you avoid at renewal is the saving.

Yes. 428 of 428 conformance tests; kubectl, Helm and your manifests work as they do anywhere. You can import a cluster you already run.

Work moves to other nodes under the policy you set, inside your private cloud first and across a network of 1,800+ nodes if you allow it. See Reliability.

By tier: 99.99% on A, 99.95% on B, 99.9% on C, 99.5% on D, as commitments in the master services agreement.

Yes: a REST API, one CLI, SDKs in TypeScript, Python and Go, and an MCP server for agents. See Developers.

Triple X Cloud is operated by TripleXPOS LLC, which contracts with you. The platform underneath is operated by its own company, named on the Sovereign and Security pages and in the footer notices.