Phala Cloud
Confidential-computing cloud that runs private AI inference, training, and agents inside verifiable hardware-secured environments.
What it does
Phala Cloud is a confidential computing platform that runs AI workloads, agent backends, and GPU jobs inside trusted execution environments (TEEs), producing cryptographic attestations to prove what code actually ran. It offers CPU confidential VMs, GPU TEE capacity, and a catalog of privacy-preserving hosted LLMs.
Core features
Confidential VMs (Intel TDX) for Docker workloads
GPU TEE access to H100/H200/B300 hardware
Verifiable runtime attestation reports
Hosted encrypted LLM endpoints (Qwen, DeepSeek, GLM, etc.)
Familiar Docker Compose-based deployment CLI
Usage-based, pay-per-hour billing
Best for
→Running private AI agents with provable execution integrity
→Hosting sensitive AI inference workloads without exposing data to the cloud provider
→Deploying GPU-backed private model training or inference
→Building AI SaaS products requiring verifiable trust guarantees
Pricing
Confidential VM: from
$0.06/hour
GPU TEE: from
$3.80/hour
CVM disk storage
$0.000139/GB
tdx.small
$0.06/hour
tdx.medium
$0.12/hour
tdx.large
$0.23/hour
Toolspool rankingby global site rank