A Path Forward for Seed VC

My Partner Stephanie Palmeri described it as “the series finale cliffhanger of posts” and lots of folks are eagerly anticipating part II. 

Well, here it is. But think of this as episode 1 of the new season. I’ll share part of the “answer,” but most of it will remain in the mystery box.

And let’s be honest, it would make no sense to share my specific “answer” publicly. It’s up to each investor to read this and figure out what the right path is for themselves.  

But here’s how I think about the way forward. Ultimately, it comes down to a few simple and somewhat obvious things:

1. The magnitude and progression of the AI supercycle
2. Defending and increasing share through sustaining innovations
3. Betting the farm on some form of disruptive innovation

The AI Supercycle

I have previously compared the VC industry’s evolution to that of the beer industry. However, there is one major flaw in this comparison: the beer industry is not one that inherently rides the waves of technological innovation. So, while it’s possible to draw many lessons from these sorts of industries, the backdrop of technology supercycles clouds an otherwise simple picture. 

There are two patterns to these tech supercycles that I believe will be repeated this time around. The first is the expansion of value that these technologies create. By driving step-function increases in productivity, efficiency, and capabilities, tech supercycles expand the pie dramatically. And over time, technologies penetrate broader and broader corners of the economy, such that their effect becomes more and more far-reaching. The pie is expanding, and that creates more economic surplus available for various players. In my prior post, I argued that seed funds are getting boxed out by YC and mega-funds, but in an expanding market opportunity, this remaining subset may still be very, very lucrative for the right-sized players. Fundrise did an interesting analysis that showed this multiplicative effect over the last four tech waves, and I think it’s reasonable to expect that this will persist. 

The second pattern is that these cycles have a predictable progression. A new technology creates a market interruption that leads to a period of frenzy that propels a rapid installation period. One or several boom and bust cycles may occur, followed by a prolonged deployment period as the technology permeates markets and society. A lot has been written about this cycle, usually referencing the work of Carlota Perez and her book “Technological Revolutions and Financial Capital”*.  There’s a great discussion on this topic on Stratechey, and here’s a great illustration from the book:

During the installation period, capital concentrates into core infrastructure build-out and enabling technologies where seed investors tend to underperform because of capital intensity and the extremely high concentration of outcomes. This is true for where value has accrued in the AI landscape thus far (GPUs, data centers, foundation models, etc). However, because this super-cycle is building on top of prior cycles that gave us ubiquitous computing, the rise of the application layer is happening very quickly. There is an argument to be made that the soil for seed investors focused on the application layer is about to get very, very fertile, and will only become more so in the decades to come. Again, from the Fundrise analysis, value begins at the infrastructure layer, but eventually expands into the application layer, which is typically larger and has a broader set of winners in different domains.

Skeptics may say that the leading infrastructure players will inevitably win the application layer as well. After all, the most successful AI application in the world is OpenAI’s ChatGPT. But I’d compare this to the era when the major internet platforms like Google and Yahoo looked like they were going to dominate every application that mattered on the internet. It may take some time for meaningful new application companies to find their footing and carve out defensible positions, but I think it’s a good bet to believe that we will have at least as many new companies of consequence built in this tech super-cycle as the ones that preceded it. This is the landscape that is most promising for seed investors. 

It’s helpful to remember that when dedicated seed funds emerged in the mid 2000s, most other investors were highly skeptical. Seed investing had been tried in the past, and the investments made by Josh Kopelman, Steve Anderson, Michael Dearing, and Mike Maples seemed like inconsequential “toys” compared to the core infrastructure or platform companies that the large VCs were targeting. Similar to today, early seed VCs started to invest in a period when the application layer of the internet, mobile, and cloud was just starting to emerge. Several mega-companies seemed destined to dominate the entire market, and the barriers to entry for these “web 2.0” applications seemed very thin. Raising seed funds in that era was extraordinarily difficult, but it ended up being the exact right time to be in business as a right-sized, early-stage investor. I believe we are about to enter a similar moment today on the cusp of an application layer explosion in this AI supercycle. 

Sustaining Innovation 

Even if the investing substrate of the AI super-cycle provides fertile ground for seed funds, success is not a given. Many legacy (and emerging) venture funds in the mid- 2000s failed to evolve to the new conditions of the market and failed to harness the tailwinds effectively. This is a similar moment for seed investors, and all firms are faced with the urgency to pursue sustaining innovation, disruptive innovation, or to try to execute on both. 

In a nutshell, a sustaining innovation is one that improves existing products to serve existing customers. It’s doing the same thing better, perhaps using new techniques or modern technologies, but improving on the common vectors of competition.

If you agree with my view that seed investing is getting squeezed by YC and the mega-funds, then it is obvious that the pressure is very high to pursue sustaining innovations vigorously just to hold ground. Not being aggressive will mean ceding market position pretty quickly to these two formidable forces, as well as the other seed players and new entrants into the space.

So, what does sustaining innovation look like for seed investors? Don’t overthink it–sustaining innovation will be about leveraging AI to transform the way VCs source, select, win, and support portfolio companies. 

The most straightforward application of AI will be around sourcing. How can firms get in front of the right founders before anyone else? How can you either figure out patterns of the right people or traction signals that others aren’t thinking about yet? And how do you align yourself organizationally to be able to act on these signals? This is mostly still unknown, but in very little time will become table stakes in an increasingly competitive market.

There are also many opportunities to incorporate AI into investment selection as well as portfolio support. From better decision-making to more data-driven deal evaluation to better ways to drive hiring or sales for portfolio companies, we have barely begun to see how AI systems can supercharge the major corners of the seed investing business. 

I know that this sort of innovation doesn’t sound particularly unique or revolutionary. When I have spoken to LPs about AI, most say that every VC firm is talking about doing something here, so it’s not that exciting. But that comment misses the point of a sustaining innovation. In a hyper-competitive market, you need to do the obvious at a very high level just to keep pace. 

Nearly every VC firm will talk about how much they incorporate AI into their work, but I’d argue that very few will successfully implement it in a practical way. This is because almost all funds are a federation of lone-wolf individuals that spend most of their time independently hunting for their next deal. Executing on this innovation with excellence requires a level of organizational focus and coordination that is completely foreign to a typical partnership structure. It’s not just about hiring a few developers or data scientists. The magic is in how the technology is transformed into a product and how it is tightly incorporated into firms’ strategy and process. This is much easier said than done, and the proof will be in the outcomes.

Disruptive Innovation

Perhaps more exciting, but also more daunting than executing on a sustaining innovation, is executing on a disruptive innovation. While every seed investor should be thinking hard about disruptive strategies, this is especially true for funds that are new to the industry. Existing players have the advantage in pursuing sustaining innovations thanks to scale, relationships, portfolio size, etc. Newer entrants generally have more pressure to figure out a way to compete in disruptive ways. 

Surprisingly, I personally don’t see that much disruptive innovation coming from the newer seed VCs. In fact, we’ve strangely seen the opposite: a convergence of strategies. When you speak to a new seed manager about their approach, you tend to hear the same thing you would expect from ChatGPT in terms of portfolio size, ownership targets, return expectations, follow-on strategy, etc. On one hand, you can blame the LP community for forcing every peg into the same round hole. But really, the blame is on the VCs for not having the creativity or courage to try to do things differently. 

What would disruptive innovation look like in seed investing? I’ll return to the beer industry for clues. The answer likely isn’t a more flavorful double-IPA with more obscure hops and better brewing techniques. That would be the realm of a sustaining innovation. Instead, a disruptive innovation involves winning on a completely different vector, and hoping that it’s a vector that ultimately matters. For example:

• Non-alcoholic beer (Athletic Brewing)
• Alcoholic seltzer (High Noon)
• Hard-core water (Liquid Death)
• DTC + vibe (Treehouse)

What might the corollaries be in seed investing? Some obvious examples:

• Investing in non-core geographies
• Investing in non-standard founder profiles
• Targeting a different sweet spot in the risk/return spectrum
• Targeting market segments in speculative gray areas
• Pursuing a different economic model or investment security
• Offering a standard deal and automated decision-making
• Very, very tight sector specialization with no regard for diversification

The challenge with disruptive innovation is that you have to get a lot of things right at the same time. You have to be non-consensus AND have the right timing AND have the right execution. A lot of the ideas above have been tried before in some fashion, but most have failed. Is it because the ideas are fundamentally flawed? Non-alcoholic beer was not a new idea, but no one was all that excited about O’Doul’s when Athletic Brewing came around. Was it the wrong premise? The wrong execution? Or just bad timing?

Similarly, a non-core geo “rise of the rest” strategy has been tried many times before. And although isolated examples of winning companies in second or third-tier cities abound, most of the firms that specifically targeted these markets failed to capture this value. Is the premise wrong? Or was it timing or execution? Hard to know.

I also think that some of these experiments can be pursued by individual partners if not by entire firms. Perhaps some of these non-standard strategies can work beautifully, but it’s tough to build an entire firm around them. Or perhaps you need to combine a few different disruptive innovations together to enjoy a sufficient portfolio effect. Either way, individual investors or entire firms need to make some tough calls about what disruptive innovation to pursue to drive meaningful alpha in their returns.

More Confident Weirdos 

Ultimately, the challenge for seed funds is finding a way to be distinct and to execute on that distinctiveness at a very high level. It may still be possible to play the classic seed VC game if you are deeply experienced, have a proprietary network, and are extremely disciplined about picking your spots. This allows you to tactically win enough non-YC, non-mega-fund investments at reasonable prices and with enough regularity to drive strong performance. And if you supercharge your efforts with the sustaining innovation of data and AI, you’ll probably have the foundation of a very good firm.

However, the number of firms that can execute on this is probably <20. For everyone else, you might as well bet the farm on some sort of disruptive innovation. You will still have to execute at a high level with the best available technologies, but at least you can do it in a less crowded environment and potentially capture a segment of the market that is meaningfully underserved.

Ironically, this is what seed VCs were doing early on when they themselves represented a disruptive innovation. As Dan Gray wonderfully commented on my last post, we need “more confident weirdos” who “float in the unbounded negative space of possibility.” I love the sentiment, but I’d make a slight modification. 

Most founders don’t really want to deal with weirdo VCs. They want to work with professionals who are focused, trustworthy, do what they say, and make their businesses better. Startups are hard enough, and there is no need to be distracted by an investor’s eccentric nonsense. The benefit of an industry that has matured and become more institutional is that the standards are set higher for professionalism, ethics, service levels, and follow-through. 

What we really need are investors with conviction about their weird taste, pursuing weird strategies, and supporting founders pursuing weird but potentially magical things. Couple that with great dependability, professionalism, practical support, and hustle, and you get a potentially winning combination. 

*Note: I realize that Carlota talks about tech cycles in a longer-term perspective. In her view, the PC, internet, mobile, cloud, and AI innovations are all part of one same 50+ year cycle. But even in that context, she tends to think that we are in the early deployment phase. I think that there are both large cycles and small cycles that follow an analogous progression at different time scales.

Addendum:

1. A Crisis Moment for Seed VC
2. One Question, Two Vastly Different World Views