The later the investment stage, the lower the expected multiple, but also I don't know that 20x is a crazy expectation to have about OpenAI? It might not happen, but people really fixated on that "might", which is not what venture investing is about.
20x isn't crazy, but its what they need just to reach parity P/E with Google (assuming their costs remain flat). In order for the investment to make sense they have to grow much much more than that to account for the extra risk, otherwise you're better off buying Google stock.
If we throw out some conservative numbers, and assume costs will rise super modestly, you have to believe OpenAI's earning will grow 50-100x for the investment to make sense. They'd have to maintain their current growth rate for 5+ years, but I wouldn't be surprised if their revenue growth is already slowing
