Yup! Big beautiful burning pile of cash. Though there’s some nuance with subsidizing inference (which is supposed to make them money long term) vs training (which we’re supposed to ignore I guess).
And they can’t stop because they want to IPO soon, and stopping this madness would signal loudly that they are bleeding cash, don’t really have a business model, and might not be worth $2 trillion after all.