Sam Altman on how enormous inference demand will finance OpenAI's frontier training without requiring high margins.

"We will have so much usage of our models that we do not need to be a gigantically high-margin business to be able to afford model training.

So much of our future compute plans will be used to sell inference to customers that, even if we can enjoy a modest margin on trillions of dollars of revenue, we can afford to train some giant models."

---

He even expects intelligence itself to become fungible, shifting durable advantage toward whoever operates the largest, cheapest compute fleet.

Distillation becomes less existential if copied capabilities compress prices while OpenAI's total usage expands faster than its margins shrink.

----

From "Invest Like The Best" YouTube channel, (full video link in comment)