This post is interesting, but I don't think it is entirely on target (the latter portion could be, but I'd need more evidence to be swayed). The author essentially paints a picture where open source startups are losers on their own, but that the smartest investors fund them to provide infrastructure to the community, like they are cooperating in a giant prisoner's dilemma of some sort. Also, that by investing in infrastructure you gain insight and access (this is the part I may agree with).
I don't think that investors (even Andreesen Horowitz) look at ANY investment and think, "Well, some of my investments are going to fail, I can take on this one and accept the hit to my returns." Getting LP commitments is very competitive, and you are stack-ranked. You can't afford to make investments that you don't think are going to be winners. To be fair, if there is anyone that could do this, it's Andreesen Horowitz, but that doesn't make them smarter, it's because they have the benefit of the doubt with LPs.
He says that these smarter investors will fund an open source project to multiply the returns of their other investments, but that ignores a likely truth. The opportunity cost of investing in the open source company to benefit from the infrastructure isn't that the infrastructure doesn't exist. It is that it is created slightly more slowly and you still get to benefit from the majority of its power, or that you are forced to use the next best infrastructure, which, if we are being honest, is quite nearly as good. I think it's likely that the next best alternative to npm is going to be 90%+ as good for your project with the same developers. Optimizing on a thin margin like that with respect to your other investments doesn't seem like good strategy.
Another factor to consider is that there have been open source companies that provided good returns to investors. MySQL sold for 20x invested capital, for example. You don't need to appeal to tenuous arguments about ancillary benefits of an investment when it could simply be that a company sponsoring an open source project with a boatload of traction just happens to be, from a risk-reward perspective, well-positioned to make returns in all sorts foreseen and unforeseen ways. We also don't know the valuation or deal structure, which could mean that the $2.6m investment is quite a bit more favorable to the investors than one's first impression.
I believe the strongest point he made is the last one, "They get insight and access at unprecedented levels into the future of Node, at the cost of something they can already afford to lose." Much of the effort in being a VC actually relates to "access." If you can get into some of the hot node deals by being known to be a shop that favors node, I could see how that might be worth it, though I'd need more evidence to be convinced they are thinking this way or that it's a good strategy. (Andreesen Horowitz can probably get into any deal they want, which is not the case for True Ventures) But that's different from thinking that by investing in npm you are going to make your other companies stronger by making npm available to them.
As miscellaneous commentary on his examples: GitHub is minting money, so investing in them was a crazy good decision no matter what the reason. Even if you think the liquidity event could be challenging for some reason, it still provides good reward for risk, and ultimately there are ways to handle lack of liquidity events (I also suspect that Andreesen Horowitz are more flexible in this regard for structural reasons, venture capital partnerships are quite complex in the details). Jeff Bezos doesn't have LPs so he can invest in 37Signals/Basecamp even if they are never selling, he just punches his dividend checks like the other owners, which can be just as good a return. The focus on liquidity events is a quirk of the standard fund structure rather than inherent to the asset class.
> I don't think that investors (even Andreesen Horowitz) look at ANY investment and think, "Well, some of my investments are going to fail, I can take on this one and accept the hit to my returns."
I'm not sure that's what the OP was saying. a16z is a $2.5b fund, so spending $100MM is only 4% of their fund. The point that he's making, is that the micro risk gained on just GitHub decreases the macro risk overall the whole portfolio. In other words, keep the assumptions concrete and businesses that utilize those assumptions will flourish.
Imagine you're an auto company, and money stops going into road construction, and therefore less people use the roads. A smart auto company may invest in construction companies. The only argument here is now do these constructions really need any extra money to keep them going? (I would argue no, they're already making enough money) In the same vein, does GitHub need more money to be motivated? (the answer is probably no, but as you point out, it's an extremely safe bet) That being said, it doesn't hurt to throw a small percent of your fund into road construction.
I don't think that investors (even Andreesen Horowitz) look at ANY investment and think, "Well, some of my investments are going to fail, I can take on this one and accept the hit to my returns." Getting LP commitments is very competitive, and you are stack-ranked. You can't afford to make investments that you don't think are going to be winners. To be fair, if there is anyone that could do this, it's Andreesen Horowitz, but that doesn't make them smarter, it's because they have the benefit of the doubt with LPs.
He says that these smarter investors will fund an open source project to multiply the returns of their other investments, but that ignores a likely truth. The opportunity cost of investing in the open source company to benefit from the infrastructure isn't that the infrastructure doesn't exist. It is that it is created slightly more slowly and you still get to benefit from the majority of its power, or that you are forced to use the next best infrastructure, which, if we are being honest, is quite nearly as good. I think it's likely that the next best alternative to npm is going to be 90%+ as good for your project with the same developers. Optimizing on a thin margin like that with respect to your other investments doesn't seem like good strategy.
Another factor to consider is that there have been open source companies that provided good returns to investors. MySQL sold for 20x invested capital, for example. You don't need to appeal to tenuous arguments about ancillary benefits of an investment when it could simply be that a company sponsoring an open source project with a boatload of traction just happens to be, from a risk-reward perspective, well-positioned to make returns in all sorts foreseen and unforeseen ways. We also don't know the valuation or deal structure, which could mean that the $2.6m investment is quite a bit more favorable to the investors than one's first impression.
I believe the strongest point he made is the last one, "They get insight and access at unprecedented levels into the future of Node, at the cost of something they can already afford to lose." Much of the effort in being a VC actually relates to "access." If you can get into some of the hot node deals by being known to be a shop that favors node, I could see how that might be worth it, though I'd need more evidence to be convinced they are thinking this way or that it's a good strategy. (Andreesen Horowitz can probably get into any deal they want, which is not the case for True Ventures) But that's different from thinking that by investing in npm you are going to make your other companies stronger by making npm available to them.
As miscellaneous commentary on his examples: GitHub is minting money, so investing in them was a crazy good decision no matter what the reason. Even if you think the liquidity event could be challenging for some reason, it still provides good reward for risk, and ultimately there are ways to handle lack of liquidity events (I also suspect that Andreesen Horowitz are more flexible in this regard for structural reasons, venture capital partnerships are quite complex in the details). Jeff Bezos doesn't have LPs so he can invest in 37Signals/Basecamp even if they are never selling, he just punches his dividend checks like the other owners, which can be just as good a return. The focus on liquidity events is a quirk of the standard fund structure rather than inherent to the asset class.