2 ms·
Loss-leaders tend to grow fast. More importantly, you need a moat to benefit from being a loss-leader and get back to profitability, which OpenAI does not have.
by minimaxir 2y ago
Loss-leaders tend to grow fast. More importantly, you need a moat to benefit from being a loss-leader and get back to profitability, which OpenAI does not have.
- loandbehold 2y agoChatGPT is the best foundational model. The barrier of entry is high because of high compute costs.
- minimaxir 2y agoHaving the best model, which is debatable even in GPT-4o's case, is not a moat.
- mjamesaustin 2y agoBut if training the model is building the moat, and doing so is highly unprofitable, then the only way for OpenAI to maintain its moat is to continue training better models than the competition, each of which will cost even more money to train and push them further into the red.