Definitely, expect future GPUs to run only signed LLM models with mandatory online attestation and reporting.