Very interesting. One of the challenges of the startup ecosystem is the traditional challenge with finance: a very limited historical perspective. Most of the people who stick around have grown wildly successful, and most of those who got blown up are out of the game. I enjoyed the insight into how a shift from low/no marginal cost will be a sea change, and think you are dead on. Thanks for sharing.
Hi Lawrence, I largely concur. However, I do find a contradiction in points pertaining to VCs investing in more companies (large portfolio size) and the increased focus on capital efficiency and margins. Owing to the former and the large Opex associated with gen AI models (including high and linearly expanding inference costs), VCs might invest more in few. In other words, the ticket sizes might increase while the recipients of those ticket sizes might sharply decrease with funding going to those managing costs efficiently.
I'd also go one to say that this is the inversion of the effect of AWS and cloud computing. Renting hardware made startups reduce their initial experimentation costs, leading to VCs investing in a large number of startups with lower ticket sizes. The difference was low costs associated with using AWS and cloud computing services.
That clearly is not the case with Gen AI, as you have mentioned in the beginning.
one banger after another - legend
Ha, all bangers, all the time
Very interesting. One of the challenges of the startup ecosystem is the traditional challenge with finance: a very limited historical perspective. Most of the people who stick around have grown wildly successful, and most of those who got blown up are out of the game. I enjoyed the insight into how a shift from low/no marginal cost will be a sea change, and think you are dead on. Thanks for sharing.
Hi Lawrence, I largely concur. However, I do find a contradiction in points pertaining to VCs investing in more companies (large portfolio size) and the increased focus on capital efficiency and margins. Owing to the former and the large Opex associated with gen AI models (including high and linearly expanding inference costs), VCs might invest more in few. In other words, the ticket sizes might increase while the recipients of those ticket sizes might sharply decrease with funding going to those managing costs efficiently.
I'd also go one to say that this is the inversion of the effect of AWS and cloud computing. Renting hardware made startups reduce their initial experimentation costs, leading to VCs investing in a large number of startups with lower ticket sizes. The difference was low costs associated with using AWS and cloud computing services.
That clearly is not the case with Gen AI, as you have mentioned in the beginning.