Show me the money: how do I make $$$ from AI?

I don't think there's a great VC investment thesis to be had in AI, bar a few opportunities of recent vintage.

A while back I finagled an invite to a weekly Zoom call of a bunch of venture capitalists and venture capital-adjacent people. Today we discussed AI, among other things, and the host of the call asked me a bunch of questions about this Substack newsletter and AI more generally. One of his arguments was that OpenAI’s valuation makes no sense. And, on a discounted cash flow basis, which is to say, a public market basis, he’s absolutely correct.

But I also think that framing misses the point about why AI is so important. AI is not important because there is money to be made. AI is important because it is a general technology, akin to electricity or the printing press. Further, over time it will likely eclipse both of those technologies in importance.

That’s a bold statement, and to the extent that it’s correct, tens of trillions of dollars will flow from the observation over time. But here’s the thing: I’m not sure that there’s a VC-style investment thesis to be had with AI. Aside from some notable VC-related AI investments of recent vintage, I think the vast majority of AI-related returns are going to accrue to the large tech companies and non-tech companies that build their own. I know that a lot of VCs will disagree with this—they want a pipeline of future decacorns to spray cash at.

The returns to large non-techs (think Walmart, Exxon, Pfizer, JP Morgan, etc.) will be from operational efficiencies that better technology can create. The returns to large tech companies will be more cash extracted from consumers and businesses.

Open source AI is certainly interesting and worthy paying attention to, but it’s also not VC-investable.

And as a technology scales and matures it becomes cheaper to instantiate. Imagine a future point where spinning up your own AI tooling is as trivial as spinning up a server with AWS is today. Technologies commodify over time. In fact, you can sort of do this today, if you don’t mind diving into the command line.

AI is, as I mention above, best understood as a general technology, akin to electricity or the printing press. There are no electricity or printing press billionaires. But trillions in wealth have been created from businesses which use these technologies. That’s the play to be had: what businesses become possible, given sufficiently powerful AI tech? Were I a VC, I’d be seeking out those companies whose existence depends on advanced AI tech. An example of a company whose existence depends on advanced AI tech could be a pharma company which uses AI to model drug molecules.

More generally, I think that the near term of AI, meaning the next five or so years, will be one in which OpenAI/Microsoft, Google, FB, Amazon, and Apple, along with NVIDIA, become even more dominant. And large, cash-rich non-tech companies like Walmart, Pfizer, Exxon, JPMorgan, etc., will deploy billions into developing their own internal AI tooling and re-orient their operations around AI. Open-source AI tooling will develop in parallel, and will act as a kind of check on some of the worst impulses of proprietary software developers. Large corporations and other organizations are going to have a very hard time re-structing their operations to take advantage of AI tech, but they will have to do so over the next decade or so if they want to survive through the end of the 2030s.