Keynote: Future-Proofing Confidential AI
summary
Cloud architecture adds hardware-enforced TEEs across CPUs and GPUs (H100, Blackwell B200, and new mid-size GPUs in public preview) and a confidential external key manager that gives customers key possession and per-agent auto keys. A live demo shows agents storing distilled memory to encrypted SSD and then deterministically crypto shreding long‑term memory by destroying keys. Intel reports AMX gives ~6x for small models, plans platform PQC support culminating in post-quantum by 2029, and promises line‑rate TDX connect attestation.
AI-written summary · about
chapters (11)
- 0:00Confidential AI future
- 1:05Customer demands
- 2:33Hardware-backed compute evolution
- 4:02Sovereignty and post-quantum crypto
- 7:12Live demo: confidential AI agents
- 12:25Deterministic data destruction
- 13:07Intel TDX role
- 15:56Hardware isolation attestation
- 17:46Data sovereignty beyond geolocation
- 18:29Post-quantum crypto roadmap
- 19:12Future of confidential computing
related talks
37mConfidential AI: Enabling Trustworthy Ecosystems in the Age of Agentic Intelligence - Session
30mKeynote Panel: The Future of...
28mGenerative AI + Proprietary Data
46mOpen for AI: Secure paths to data collaboration volume and diversity
19mHow Can We Have Private Access to Intelligence?
06mSponsored Keynote: How to Protect AI Workloads in Cloud Native
01h 11mKeynote - Securing AI: Building Trust in the Era of AI
36mCTO Panel: The Impact of Confidential Computing with Intel, Azure, AMD & NVIDIA