I can’t help thinking the SoftBank #goose metaphor is a better model for AI than they think it is. Gold eggs don’t lay eggs, but normal goose eggs hatch, become geese, and then lay more eggs. If you want geometric growth (you know, the thing that they’re selling in all of the other slides) then you want a goose that lays eggs that hatch and become geese. If you have a goose that lays gold eggs, you briefly have a load of gold, but eventually the goose dies and stops producing gold. Oh, and, because transmutation of elements isn’t actually a thing even for geese, you’re only able to get gold eggs out of your robot goose with an egg factory inside because you’re feeding it with gold, mostly from melting down the eggs.

It’s a very succinct summary of SoftBank’s current portfolio of investments.

EDIT: He’s also complaining that the market has decided the value of the fund is half the value of the fund’s assets. But that’s because the value of the assets is entirely made-up nonsense. These are not valuations based on real concrete assets, nor on any plausible earnings projections. They are not publicly traded things where at least the value is some consensus of what people think it is. The valuation is based on something that is deliberately misleading. And SoftBank’s investors are marginally less stupid than the manager, so they don’t buy it. Which is also promising: the market thinks SoftBank’s OpenAI investment is worth a lot less than Altman is trying to pretend it is.