What’s the current state of verifiable compute? Feels like everyone talks about it but nothings actually usable

Trying to understand where we actually are with verifiable compute. The pitch makes sense, cryptographic proof that computation happens correctly without exposing the underlying data. BUT every project I look at is either:

-Pure research or academic

-vaporware with a token attached

-so technically complex that adoption seems impossible

ZK proofs, TEEs, secure enclaves. lots of approaches but what’s actually being used in production? Especially for ai workloads where this seems most needed