Right, but not before they try to fire people again, fafo cycle continues!

For this specific case they want to train local smaller models I think, like we didn't have that before the llm boom