Rahil

OpenAI Revenue Breakdown > According to Kelvin Mu's research, OpenAI’s revenue is split evenly right now between: - B2C (GPT plus subscriptions) - B2B (API tokens) However, by extrapolating out to the future, one realizes that the future of the company rests on the B2B side. But is it enough? For B2C: He places the total TAM at $10-20B. While a big number, it’s small compared to the revenue of tech incumbents such as Google ($280B), Microsoft ($218B), and Meta ($117B). While B2C will decrease in importance in the future, it’s important for the company right now because it helps to subsidize the operating cost of the company. Here’s another way to think about it – the breakeven for the $20/month GPT+ subscription fee is about 50 pages of text output per day at the current token cost. I doubt the average GPT plus subscriber is using it that much. Hence, there is a lot of economic surplus leftover that’s used to subsidize other segments and costs of the business. For B2B: I place the total TAM at $200B+ under current token pricing. This assumes an addressable population of 2 billion and an ARPU of at least $100. (see slide for more details). $200B is a much more interesting number than $10-20B, but it also assumes a constant token cost which is probably unrealistic given the current trajectory of token pricing. Furthermore, we are seeing more competition from open-source models, which will likely reduce the TAM for closed-source models. So given these dynamics – on what basis can OpenAI scale to become the next Microsoft/Google? What other monetization strategies are available to them? Source: Kelvin Mu (I love his research)

There’s a major structural flaw in how the VC ecosystem works that we don’t talk about enough. It’s a flaw that creates confusion and bad advice for Founders. And it’s a flaw that makes no sense when you dissect it. 🧵👇