The actual article with non incendiary headline is actually interesting. It sounds like Tesla’s data center buildout is slower than their chip appetite, or they are shifting responsibility and resources to xai. I wonder how their internal chip project is going.. normally a surplus of h100s might seem like they have some winning custom silicon, but I imagine if they did we would be hearing a LOT about it.

With Tesla dealing with declining demand, and the cultural transition to a larger and more stable company, I understand why Elon would peel off AI, finance it with commercially reasonable contracts from Tesla, twitter, and spacex, and boom - you’ve got like the second highest revenue ai company in the world with arguably some of the most valuable proprietary data in the world, especially for the robotics / physical world stuff that’s coming, and it could be operating in that sweet spot he seems to like of massive growth operational + research scale out startup.

I wonder if xai needs custom hardware to win. I don’t think so. But I’m also unsure what winning looks like: AGI? Home assistant robot? Should be a fun five years to see what they end up doing.