Do you think that it really is at the end of the day more about profit growth and operating margins? Or in some cases, will this be akin to what we saw when the automobile replaced the horse and buggy, when the refrigerator replaced the ice delivery man, when the spreadsheet replaced the calculator? That we'll have to re-underwrite these companies with an assumption of a terminal value of zero.
I think it's a continuum. I think some of the lighter weight applications, it's going to be hard for them to go sell successfully in the enterprises that are trying these models and coming up with building that software on their own with cloud. And one of the things we're going to look for in the next 6 to 9 months is what is happening to sales cycles of these companies. And then I think it's a continuum all the way to businesses that are in much more regulated markets where you have interconnections into states and federal policies. And think about the financial infrastructure, things like that, Claude isn't just going to be able to build the workflows and replace. So I think it's a continuum where it goes from re-rating to is this really, there's fundamentally a better way to do things.
People have often thought about things like regulation as a liability. In this case, it could actually be an advantage, a moat, right?
No, absolutely. Look, I think if you take a step back and think about the proverbial Silicon Valley or the tech ecosystem building every company in Silicon Valley, I would describe it as a C-corp writing code. Right? If they were not in the regulated markets and it was sort of core infrastructure, sort of simple productivity applications, there were these C-corporations with engineers writing code. But in the world where code self-writes, which is where we're headed, what is the purpose of those entities? So I think what you're seeing is just the transformation, the overall opportunity, innovation is not going to stop. There's a whole AI stack that's emerging that allows for enterprises to take advantage of AI, and then there's all this transformation of these industries that is happening using these models. So there's a lot to do in the context of where investment needs to go. But there's a category of companies where the AI will do that work for us. I think instead of giving us software to be more productive so that human cognition can be leveraged in enterprises, software will just do that work for us.
As you noted, this isn't a thought experiment for you personally. General Catalyst has invested heavily in AI, both in AI foundation model developers Anthropic and Mistral as you mentioned, as well as hundreds of companies that are applying AI in various ways to make business more efficient and more productive. Now, I'm aware of the fact that at least a couple of GC investments have already succumbed, if you will, to AI evolution. How vulnerable is the rest of your portfolio as you see it now? Going back to that conversation you were telling us about having last night.
Yeah, look, I think we have, so at GC we have been guided by two big themes simultaneously. One has been around applied AI and the evolution of the infrastructure and diffusion of AI into enterprises. The other is around global resilience, which does come down to how do countries where we do business, so US, Europe, India, all the allied sort of markets, how do key industries become, sort of, countries have strategic autonomy in them. So what happens in healthcare and defense and energy and manufacturing, financial services. So we have a large part of our portfolio focused on what are the regulated markets. We sort of went deep into them over the last 15 years knowing that these were places where founders could actually have long-term moats. And in fact, it was driven by an observation that the technology platforms had such an advantage in sort of unregulated markets that it was important for us to diversify. It was actually a conscious choice we made about 14, 15 years ago, right? And so we're fairly insulated in that regard. We also went fairly deep into hardware as a result. If you think about defense, if you think about manufacturing, we ended up building a lot of companies, companies like Anduril and you know, Saronic and Helzing and Rafi here in India. Healthcare, we're deep in, we're building hospitals here with PB Health for example. So I think we are fairly diversified, but our core SaaS portfolio, you think about content management or you think about productivity or you think about sort of core data infrastructure, we are thinking about it as what do you do with these companies? I mean it's a small part of our NAV, but it's a part of our NAV where a lot of what I addressed are questions we are actively thinking through on how can AI be an advantage for those businesses and if not, then how do they persist in the long term and on a cash flow basis with reasonable multiples can actually be good investments.
When we talk about moats, we're talking about sustainable competitive advantage, and it is clear that in some cases those moats are disappearing. We've talked about at least one area in which the moats might be durable: regulated industries or software providers to regulated industries where they're deeply entrenched in a regulatory structure. Where else do you see moats as being durable?
Yeah. Look, so if you look at the software application in enterprise, the thing we look for from a moat standpoint is is there proprietary data that the models aren't really trained on. And the second thing is is there learning going on in these systems? I mean a lot of the systems records aren't really used in a way that knowledge is essentially being captured. It's really data and there workflows on top. And so if there isn't sort of true secret sauce of the enterprise being captured in those systems, then they are vulnerable because migrations can be handled by AI easily. Data is your data. You can post-train these models on your data and then if there's not a learning, then those workflows can actually easily be replaced by observing how the enterprise functions. So I think the way you build software in this next phase, how you think about products, will evolve from how a lot of these sort of traditional software companies used to build.
One of the central debates in AI evolution, and we're going to hear a lot about it over the next couple of days, is around safety and responsibility. And that is one in which you personally have played a leading role with your 2022 book Intended Consequences and also with the Responsible Innovation Labs which you founded in 2023. Now, Anthropic, which, as Hemant has noted, General Catalyst is an investor, has arguably put a higher premium on safe and responsible AI than some of its rivals. However, on this very issue, Anthropic is caught in a battle with the US government at this very moment over how Claude, its LLM, is being used. I'm sure you would agree there is a lot at stake both financially and philosophically.
