I don’t know Is it wrong to think that in optimal contract theory, there is a dual agency relationship and therefore payments should go both ways? Of course, optimal contract theory often fails.
Ben Thompson I have a hypothesis (gate):
This sounds like a play to gain share of users and awareness with the potential to upsell them to a subscription. In other words, it’s the same model that OpenAI has on its websites and apps. Moreover, this partnership also explains why OpenAI is the only option to launch if Apple doesn’t pay. For example, Google probably wanted to get paid for Gemini and Anthropic for Claude. We’re not going to pay for it, especially if (2) OpenAI is making aggressive moves to build a consumer business and become a durable brand and winner in the consumer space. In short, my updated current thinking is that both Apple and OpenAI are betting that very large language models are becoming increasingly commoditized. That means Apple doesn’t have to pay to access the model, and OpenAI is seeing scale and shared consumer awareness. This is the best way for a sustainable business.
If I were to guess, it looks like the major payments are going from Apple to OpenAI, right? We know that Apple is very good at monetizing its customers. OpenAI will not have a comparable track record, no matter how fast it grows over the next few years. So there’s a model where Apple pays for access to AI and then charges more for the iPhone later. If enough people ask for it, you could get cheaper phones without these services. Of course, OpenAI also charges users, but only if they exceed their quotas, and then they face the “problem of additional subscription fatigue” in a way that Apple doesn’t.
It’s just a hypothesis. The fact that we don’t know speaks to the limits of economic reasoning.
