Skip to main content
The Singularity Cloud Network is decentralized confidential compute: independent operators run hardware-attested Trusted Execution Environments (TEEs), and you rent that compute by the request or by the hour — paying in crypto, with no account required. What makes it different from a normal cloud: every workload runs inside a verifiable enclave, and for AI inference your prompts are end-to-end encrypted — the network relays only ciphertext, so not even the operator (or us) can read them.

​
Three surfaces

SGL Grid

Decentralized, confidential AI inference. OpenAI-compatible, pay-per-call.

SGL Machines

On-demand cloud compute instances, deployed in minutes.

SGL Processors

Serverless functions inside a TEE. (Coming soon.)

​
Why it’s different

  • Confidential by default — workloads run in hardware TEEs; AI prompts are end-to-end encrypted (see Confidential compute).
  • Verifiable — each node is hardware-attested; inference responses carry an attestation you can check.
  • Pay per use, in crypto — per-request or hourly, via x402 (USDC) or prepaid credits. No subscription, no KYC. See Billing.
  • Decentralized — compute is served by independent, staked operators, not a single provider.
  • Open & compatible — the Grid speaks the OpenAI API, so existing SDKs work by changing a base URL.

​
Get started

Use the Grid

Send your first inference with the OpenAI-compatible API.

Deploy a machine

Spin up a cloud instance from the catalog.

Provide compute

Run a node, serve workloads, and earn.

How billing works

Credits, x402 pay-per-call, and API keys.