Phala

A confidential AI cloud platform that runs AI workloads inside hardware-backed Trusted Execution Environments (TEEs), solving the problem of keeping data and models private during processing while providing verifiable execution. It targets developers, IT teams, and enterprises that need to deploy AI agents, LLMs, and GPU jobs with privacy guarantees. The platform is positioned as a verifiable alternative to traditional cloud providers, offering cryptographic attestations that prove what ran without exposing secrets.

Key features

• Confidential VM deployment
• GPU TEE support (H100, H200, B300)
• Confidential AI model endpoints
• Attestation API for verification
• Docker Compose compatibility
• CLI for deployment and management
• OpenAI-compatible LLM endpoints
• Private inference and training