For startup investing the pure math explanation to invest in 1,000 companies makes more sense than in public markets. If you have 1 out of 1,000 startups return 1000%, 99 companies break even, and 900 companies go bust, you still make money due to the asymmetric one company generates. In public markets, two deviations of risk is satisfied with 20ish companies where three deviations being around 80 companies assuming a log-normal distribution. Adding more companies does not greatly lower the risk when the outcome isn't so binomial.
Up to a certain point I agree, but I do wonder about the problem that startups can appear out of nowhere if they think there is investment money to be had just for showing up. In other words, your very willingness to fund many startups, may cause the quality of startups to decline. The analogy to stock picking is not complete here because you buying stocks in many different companies (an index fund) doesn't cause more companies to go public. But, if it doesn't take much more than a slide deck and a good talker to take your money, you will easily have 1,000 startups show up that each have 0% chance of succeeding.
Not saying this is currently a problem, just that it could be if most VC's tried to take a "place chips on all the numbers" kind of strategy.
I assume you meant 1000x not 1000% (since the latter would only be 10x.
Is there any evidence that even as many as 1/1000 startups provide a 1000x return to VCs, considering that the return is only a fraction of a total exit?
Even then, assuming all investments were equal, that's a very meager return of 1099 on 1000. For a 10-year fund, that wouldn't even beat inflation, let alone compete with something like Treasuries.
An article that delves into more details (although there's some glossing-over, still):
