What Theo claims on YouTube is that each model they make is profitable over its lifetime and the weird numbers for the company are because they keep training more models, having to pay up front to buy a bunch of future profit each time basically
What Theo claims on YouTube is that each model they make is profitable over its lifetime and the weird numbers for the company are because they keep training more models, having to pay up front to buy a bunch of future profit each time basically
The problem is that OpenAI will have to keep doing this basically indefinitely, as otherwise open-source commodity models will catch up with them and offer an equivalent product at a much lower cost. If the company is reliant on having to keep paying more than it's making on its current model to train new models, I don't see how it can ever become a sustainable business.
Not even just open source models - other commercial LLM models as well. All of the big LLM companies (Google, Anthropic, OpenAI) are basically ocked in a cold war of having to continuously outspend the other providers or risk becoming irrelevant.
The big question for me is if people will ever be happy with a model that is "good enough", and can thus be optimized and run profitably over time without faling behind. Time will tell!
The bigger question is if you have a model that is "good enough" that is open source if there is a market for APIs delivering said models that able to justify the funding.
This claim that if they stopped training new models wouldn’t the old ones become stale as things are updated? Not sure how quickly that would occur but it does seem likely as the world moves fast
Future profit? Isn't that what every company in the red claims?
Ofcourse in the 21st century none of this actually matters people just want to buy low sell high.
If your amortised fixed costs plus your variable costs are greater than your revenue, that’s called making a loss. It’s like a shipping company saying the cost of delivery covers the cost of their staff and fuel, so they’d be profitable if only they didn’t need to pay for all these ships.
I would think that logic only works if opensource models like llama don't catch up? In other words the models don't become a commodity.