My understanding of the TEE attacks is that what we thought cost ~$100k actually only cost ~$1k. Either way, we all knew that current TEEs require Proof-of-Cloud, and any one serious in TEEs has said this long before these attacks came. It's janky, but it's what we got for now. With Proof-of-Cloud the attack cost prob moves well into the millions (10s of millions? what would it cost to get undetected access to do surgery to an Azure node?), plus reputational stake. I'd still be reticent to trust it for correct execution of a blockchain protocol, but as a practical privacy solution for shared state with execution proven in ZK, it seems totally reasonable. Looking forward to alternative solutions maturing. That means (1) FHE+MPC and (2) secure-by-physics TEEs