Phala
Confidential AI cloud running agents, LLMs, and GPU jobs inside hardware-backed TEEs.
About Phala
Phala is a confidential compute cloud that runs AI agents, private LLM models, and GPU workloads inside hardware-backed Trusted Execution Environments (Intel TDX and NVIDIA Confidential Computing) so secrets and data stay private during processing. It lets developers deploy existing Docker Compose workloads into CPU or GPU confidential machines and emits runtime attestations that cryptographically prove what code ran. It targets regulated sectors like finance, healthcare, and legal, as well as decentralized/Web3 AI use cases.
Screenshots
Commonly Cited Strengths & Limitations
Strengths
- Hardware-level data privacy via TEEs
- Verifiable execution through cryptographic runtime attestation
- Compatible with existing Docker workloads and AI frameworks
Common Use Cases
- Running private AI data analysis and model training under regulatory compliance
- Providing verifiable data protection for enterprise AI SaaS products
- Building decentralized and autonomous AI agents for Web3, DeFi, and IoT
- Secure fine-tuning and inference of large language models on sensitive data
- Processing sensitive financial/trading data with AI while maintaining compliance
- Multi-party collaboration on medical data with sealed PHI under HIPAA
Details
- Pricing Model
- Paid
- Starting Price
- $3.20/GPU/hr
- Team Size
- Individual, Startup, Business, Enterprise
- Category
- AI & Machine Learning
Key Features
- Confidential VMMove existing Docker Compose workloads into CPU or GPU confidential machines with a verifiable runtime.
- Agent SandboxRuns agent backends in a confidential VM with sealed keys, private memory, and verifiable execution.
- Confidential GPU MarketplaceLaunch H100, H200, and B300 GPU capacity with TEE-backed runtime proof and public attestations.
- Confidential LLM ModelsOpenAI-compatible endpoints for private LLM models with encrypted prompts and verifiable runtime state.
- AttestationEmits runtime measurements and hardware quotes that software can verify to prove what code ran.
- GPU TEECombines Intel TDX with NVIDIA Confidential Computing for confidential GPU execution.
Official Pricing
Phala's published plans
H200 GPU
$3.20/GPU/hr
- 24 vCPU
- 141GB VRAM
- Intel TDX + NVIDIA CC
B300 GPU
$5.60/GPU/hr
- 16 vCPU
- 288GB VRAM
- Intel TDX + NVIDIA CC
Real Pricing from Verified Users
See what people actually pay for Phala
Reviews
0 reviews
No reviews yet. Be the first to share your experience!
Switching Stories
Real migration experiences with Phala
Discussion
Ask questions, share tips, or discuss this tool with other users.
No discussion yet. Start the conversation!
Agentic CLI coding tool by Anthropic
More in AI & Machine Learning
The agentic IDE powered by Codeium AI
