# TEEs solve verifiability and confidentiality at \~20% overhead

- **Speaker:** Greg Osuri
- **Attribution:** attributed
- **Said:** 2025-12-03
- **Theme:** Decentralized AI — https://www.gregosuri.com/predictions/decentralized-ai/
- **Source:** The Truth About Decentralized AI and the Future of Compute (TEACHMEDEFI)
- **Watch at:** 00:17:25 — https://www.youtube.com/watch?v=CWA3IAr9sc4&t=1045s
- **Canonical:** https://www.gregosuri.com/predictions/decentralized-ai/2025-12-03-tees-solve-verifiability-and-confidentiality-at-20-mtg6/
- **Cite as:** Greg Osuri, "TEEs solve verifiability and confidentiality at \~20% overhead," The Truth About Decentralized AI and the Future of Compute (TEACHMEDEFI), 3 Dec 2025, 00:17:25. https://www.youtube.com/watch?v=CWA3IAr9sc4&t=1045s

## Quote — verbatim

> A big favorite for verifiability today is trusted execution environments, where it is vendor-attested workloads, essentially attested at the chip level, and with memory encryption it's impossible to at least see inside a workload when it is encrypted at the memory level.

## Context — site annotation, not spoken

He measures ~20% TEE overhead as "acceptable," pairs TEEs with MPC and Akash's audited attributes for data-residency needs, and says "TEE solves 99% of the problems" for verifiability and confidentiality.