Posted in: Uncategorized

Honest and Naive Questions from a Generalist Seed VC Grappling with the Generative AI Revolution

I’ve been sitting on this post for a while, and I finally feel ready to post it. I have been a generalist VC for my entire career. I’ve invested through the early days of SaaS, the shift from desktop to mobile, the rise of web 2.0 and social apps, crypto (mostly sat that out), and various other mini-waves of tech innovation. In most of the aforementioned phases, there has been a clear opportunity for venture-funded startups to capture a meaningful portion of the value created by a new wave of innovation. Here’s my short summary of the various waves I’ve seen:

  • In web 1.0, the ability to do anything on the Internet felt novel – content, commerce, and communications were all rebuilt for the Internet, and being Internet native was a huge advantage. Many interesting companies were built and failed in this era. And many of the winners from this era (notably Amazon and Google) are still with us.
  • In web 2.0, some technological developments made the web feel more responsive and interactive. We also brought the concept of a social graph and social network to the web. Lots of new businesses were built here that were simply not possible before (notably Facebook).
  • In the iPhone-driven mobile era, many legacy desktop assets weren’t useful; you couldn’t just shove your desktop website into a mobile app and make it work. You needed to understand mobile UI/UX and distribution to succeed. Many, many cool companies were created in this era (notably Instagram, Uber, WhatsApp, and lots of other international messaging and communications apps).
  • Crypto just felt (and still feels) orthogonal to the world we had in web 2.0. We will see what crypto ultimately yields, but it feels like success in crypto largely meant building new businesses with new tech stacks for new audiences.

All of the waves above had really clear lanes for private company value capture. Specifically, private companies had access to some technology, expertise, or earned knowledge that gave them an edge against existing incumbents or new startups that waited for things to settle out before starting. I keep looking at what I see in the world of generative AI, and it feels to me like the dynamics feel much more like the lessons I learned from studying open-source software than what I learned from the big investing waves that created broad swaths of interesting, valuable companies.

One impression I have from studying open-source software is that the balance between value creation and value capture tilted heavily toward value creation. Many OSS companies created (and continue to create) tons of enterprise value without capturing much of that value. That is good for the Internet, good for society, and allows many businesses to flourish. But it can often produce businesses that are not as financially attractive as companies that focus on value capture. The early AI companies seem to have the best elements of OSS (massive value creation for the ecosystem) AND the best parts of SaaS (massive private company value capture). That feels really unique and interesting to me.

When I look at the current landscape of generative AI or next-gen AI generally, it feels like we have a few gigantic early brands that are creating the technology that many other companies are using. It feels to me like we are trending toward a world where there is a fairly concentrated set of technology suppliers who provide the plumbing to lots of application providers who use that technology to build new things. If you are a technology provider like OpenAI, this is a great thing – you will be a net supplier of technology to many people who figure out how to build big businesses using your tech. The more people who use your tech, the better it gets. But what if you are an application developer trying to build something that leverages the tech that OpenAI provides? Here’s where I get stuck as an investor.

I already have two writing tools that I use regularly, Notion and Mem.ai, both of which have integrated AI writing assistants. I assume that AI writing assistance will become a table-stakes feature for any writing tool. How do you build a defensible business around text creation powered by AI when everyone has access to the same tooling? I assume that Google Docs, Word, and every other text-based writing tool I use will soon incorporate an AI writing assistant. What does that mean for the opportunity to create standalone, valuable businesses powered by AI in this domain? If you swap domains, the story repeats itself.

Right now, it feels like most of the AI applications I encounter all use tools from a very narrow set of providers (mostly OpenAI). If everyone has equal access to the same tools, how do you build a defensible business? The only answer I’ve come up with is that these tools have to augment and amplify an advantage that a given company already has relative to its competitors. If a given company sees an advantage in using generative AI in its service, I assume its competitors are equally smart and will do the same.

If what I described above is true, a lot of the value will flow to the relatively small number of companies (like OpenAI) that provide the tech and tooling to people looking to build applications. If that’s true, where do you put your investment dollars to work outside of the few monopsony technology providers?

I am not an infrastructure/deep tech investor. I don’t invest in tech or tooling that makes it easier to create AI models, run AI or deep learning, or any of the core technology that will advance this field. I’m looking at the application landscape and trying to figure out where I should spend my time and energy. We already have companies that are using this technology to improve their products and services, but I am already invested in those companies and am very curious as to how it plays out for them. If you have thoughts or ideas, I’m all ears!

Leave a Reply

Your email address will not be published. Required fields are marked *

Back to Top