Strictly Confidential: Use Cases & Demos of AI Workflows in TEEs
summary
Demo shows end-to-end implementation: Google Cloud adds confidential G4 (NVIDIA Blackwell RTX Pro 6000) and expands C3→C4 CPU families, runs AMD SEV confidential VMs plus confidential GPU mode and H100 support, and ships an open‑source prompt encryption SDK. Live demo blocks a non‑confidential model from connecting, successfully attests CPU+GPU claims, and serves private medical prompts over an attested TLS channel. Exact claims: PCI/NVLink encryption, instance key/OIDC attestation flow, no code changes and minimal perf impact.
AI-written summary · about
chapters (10)
- 0:00Confidential computing meets AI
- 1:26Confidential AI use cases
- 3:51Companies leveraging confidential AI
- 5:15Google Cloud’s confidential portfolio
- 7:46New confidential AI infrastructure
- 9:35Building blocks for confidential AI
- 12:24Securing prompts with SDK
- 16:20Demo setup for confidential AI
- 18:49Live demo of confidential AI workflow
- 22:59Summary and next steps
related talks
47mTowards Confidential AI for the Masses!
36mCTO Panel: The Impact of Confidential Computing with Intel, Azure, AMD & NVIDIA
44mGPU-less, Trust-less, Limit-less: Reimagining the Confidential AI Cloud
42mSponsored Session: Confidential Computing - New Capabilities for New Workloads- 42mSponsored Session: Confidential Computing - New Capabilities for New Workloads
37mConfidential AI: Enabling Trustworthy Ecosystems in the Age of Agentic Intelligence - Session
13mProvably Secure AI for Health Data – Powered by Confidential Computing
11mLightning Talk: Computing Confidentially in the Clouds