Phala

Phala is a confidential AI cloud platform that helps developers and organizations run AI workloads in hardware-protected environments. Its services use Trusted Execution Environments (TEEs), including Intel TDX and NVIDIA confidential computing GPUs, to keep data, prompts, models, and execution private while providing cryptographic proof of what ran.

The platform offers private LLM inference through OpenAI-compatible endpoints, a catalog of confidential models, GPU capacity including H100, H200, and B300 instances, and secure environments for AI agents. Customers can also deploy private AI training and data workloads, along with existing Docker Compose services, as confidential virtual machines. Phala provides command-line tools, deployment templates, APIs, documentation, and runtime attestation reports to help teams build, launch, and verify applications.

Designed for use cases such as financial services, healthcare research, legal AI, and enterprise software, Phala emphasizes privacy, auditability, and regulatory needs. The company reports SOC 2 Type I certification and HIPAA compliance, with ISO 27001 certification in progress. It also offers enterprise support and GPU marketplace access.