Operator enrolment is not open.
One gateway, no nodes.
Each call runs through the Provn gateway and the upstream its operator configured. Sandbox and Terminal runs execute in QuickJS on WebAssembly. Provn has no node registry and no enrolment flow, and no outside machine serves any traffic.
Third-party enrolment.
An operator network needs vetting, scheduling, payouts and abuse handling, and Provn has built none of them. Enrolment stays closed until they exist. Provn has not published a wire contract for nodes either.
Two kinds of machine would matter.
If enrolment opens, these are the classes worth serving, because each has a buyer and a guarantee Provn could stand behind.
Open-model inference
A datacenter GPU with enough memory to keep an open-weights model loaded. A model that has to page in and out isn't worth scheduling.
Confidential sandbox
An x86 server with Intel TDX or AMD SEV. Provn plans a confidential tier, and without that CPU feature there is nothing to attest.
Consumer machines are out.
Desktops, laptops, home servers and CPU-only boxes won't make the list. Don't buy hardware for Provn, for either class above or for any other reason. Nothing on this page justifies spending money.
Three things Provn is not saying.
No earnings offer
Provn has set no rate, revenue share, payback period or utilization figure. Anyone quoting earnings for a network that hasn't opened is guessing with your money.
No date
Enrolment opens once there is demand for the compute and vetting and payout machinery to back it. Provn won't name a quarter before then.
No operators today
No third-party machine serves Provn traffic, so no one earns anything from Provn compute. If a page on this site suggests otherwise, tell Provn on X and Provn will fix the copy.
No form. Send a DM.
Provn hasn't built a signup flow, because there is nothing to sign up for. DM x.com/provn with the details below and you'll get a plain answer about where things stand. You won't get a rate card.
The hardware you have or can get: CPU model and TDX or SEV support, or GPU model and memory.
Location: datacenter, office or home.
Connectivity: a public endpoint, or behind NAT.
Machine count, and whether you own or rent them.