AI investing can be surprisingly unprofitable even when you're correct
The path to liquidity is paved with weird licenses and dividends
It can be surprisingly difficult to make money from investing, even when you have a pretty good idea what the future will look like. There is, at this point I think, a reasonably high probability that AI is going to be a genuinely new technology paradigm (side-note: I still feel it’s more accurate to compare AI to computing than the internet, as the latter was connectivity and distribution whereas AI is more about productivity-time will tell). But two recent M&A deals show the complexity of investing and actually making money from investing in AI trends.
Meta acquired a 49% stake in Scale AI for ~$15b in cash. Their last two rounds (2021, 2024) had valued the company at ~$7b and ~$14b respectively. So while on paper this is a ~$30b valuation but on the basis of having taken the CEO and basically distributed cash to everyone at the last round valuation, it’s unclear what the remaining 51% of the company is worth (probably more than zero but let’s see). So investors after 2021 make between 2x their money and breakeven.
Google announced a deal to license Windsurf’s core technology and hire some of its team for $2.4 billion in cash. Their last round (2024) valued the company at $1.25b. Somehow, investors are receiving part of this licensing deal and also keeping their equity in the company. Again, a similar analysis of an okay outcome and kind of weird tail situation.
Scale and Windsurf were not the first (and won’t be the last) AI companies to be hired and licensed. A lot of commentary around the deal structure is based on ‘oh well Lina Khan and the FTC wouldn’t let us buy companies so we have to do it this way’. Yes maybe. But the investor reality is that the buyers of these companies are simply not pricing them where the investors are.
You can be right with your instinct and still not make (much) money. The single biggest challenge in the AI supercycle is where to invest to capture value in a way that doesn’t rip your face off or just give you pretty normal (and sub-S&P 500) returns. The more general sobering point is that EVEN if you are a VC fund in the right companies AND you get exits, you may not actually be generating enough returns for your investors.
But look, maybe I’m being a little mean? There is generally no such thing as bad liquidity, especially in illiquid assets. I say this cautiously but there does genuinely appear to be some real (albeit still cautious) momentum around both exits (albeit primarily driven by Meta) and IPO pipelines (Figma, Navan, Discord etc). So it might even end up being quite a good year for distributions. This is an investor point but founders should care a lot-DPI drives market confidence.
It’s intriguing to look at the topic of tokenisation in this context. One narrative which you could project onto the markets is that if the IPO mechanism for investor liquidity isn’t going to return fully, then private company tokenisation (which Robinhood appears to be leading the charge for) is going to be the way we unlock liquidity from retail investors (assuming we can’t sell them more access to private market funds). I’m absolutely not saying that this literal set of words is being uttered but it is kind of the effect which this series of events would generate.
Other reading and half-formed thoughts
OpenAI’s Amazon archaeology initiative has created wild levels of inner-ancient-history-nerd excitement for me. It has the distinct fingerprints of Nat Friedman and given the sheer scale of our known unknowns in that region, it will certainly produce some remarkable new discoveries.
I had not really thought about stablecoin as a convenient US treasury sink but it does make sense. Intriguing to consider the longer term ramifications of this.
I just finished Patrick McGee’s excellent Apple in China book (it is as good as the reviews suggest). I am intensely curious how much Warren Buffett truly understood what was going on here when he invested in Apple in 2016.
A long interesting read about the UK’s preposterous stance on air conditioning.
Everyone in the world wants more dollars and USD stablecoins are the easiest way to get them.