Topics covered:

• Share your story and how you ended here doing confidential AI at Google?


• What problem does confidential compute + AI solve and for what clients?


• What are some specific real-world applications or use cases where you see the combination of AI and confidential computing making the most significant impact?


• What about AI in confidential vs AI on prem? Should those people just do on-prem AI instead?


• Which parts of the AI lifecycle need to be run in Confidential AI: Training? Data curation? Operational workloads?


• What are the performance (and thus cost) implications of running AI workloads in a confidential computing environment?


• Are there new risks that arise out of confidential AI?