I'm posting a minor update of an informal paper I wrote a while ago using arguments from standard portfolio theory to look not at portfolios of technology companies but instead the technology portfolio of a given company — although then I turn around and look at portfolios of companies as a combination of the individual portfolios.
This all sounds byzantine (and, fair warning, this is the sort of paper that uses mathematical notation in the sincere belief that it makes things clearer), but I believe it provides a tentative model to explain how, despite but also because strong cultural drives towards innovation, both individual startups and investment funds tend to take less risk than would be optimal for them.
Implications for startup design are left as an exercise for the reader.
Technology portfolio modeling, generative AI, and optimal fund design [PDF link].