1Mby1M Virtual Accelerator AI Investor Forum: Heriberto Diarte, Co-Founder and Managing Partner at Catalyzer Ventures

Heriberto Diarte, Co-Founder and Managing Partner at Catalyzer Ventures, offers a point of view that is refreshingly down to earth. His fund is $100M, tightly focused on specific sectors where the partners have deep domain knowledge, and they aim to build capital efficient businesses that can make money off smaller exits instead of mindlessly chasing portfolio-making unicorns. 

Sramana Mitra: We are going to start our discussion with Heriberto Diarte, co-founder and managing partner of Catalyzer Ventures. Heriberto is a longtime friend of mine, and we’ve had many offline conversations on many different topics related to the industry. Heriberto had a stint as a corporate VC prior to raising his own funds with a co-founder. So he’s going to tell us more about his journey. We recently had dinner at his house, and we had quite a bit of discussion on some of the changes that are happening in the industry, and I invited Heriberto to come to the program and discuss his thoughts on how the industry is evolving. 

Welcome, Heriberto. Great to have you.

Heriberto Diarte: Thank you, Sramana. I’m delighted to be here in your program. It only took me 11 years to pass the test to be good enough to come on your program. So I’m very pleased to be here now and look forward to the discussion.

Sramana Mitra: So Heriberto, let’s start with a bit of an introduction of your fund, your background, your co-founder’s background, because I think those are some of the uniquenesses that you bring to the table through this fund.

Heriberto Diarte: Oh, yes, absolutely. So we are two co-founders, but three partners, and we all have two things in common. First of all, we’re serial entrepreneurs. I’ve started five companies, three have done very well, two died, and my co-founder has started six companies and sold those companies for 10X. And then the other co-founder started two companies, both of them very successful.

The other thing we have in common, we are operators. I’ve been a CEO of large industrial companies. So has my partner, Christina, and she’s on the board of big companies like Cisco, DuPont Electronics, now called Qnity, and other large companies. Our third partner is currently the CEO of the largest independent hydropower operator in the United States. So, we’re all ex-CEOs, ex-entrepreneurs. We bring that to the table.

On top of that, my two partners are scientists. They’re very technical people who have PhDs, patents, and have created things. Their startups were based on technology they developed and can really understand technology deeply. For an early-stage fund, that is a very important part of the skill set. I don’t have that part of the skill set. If you tell me, “This is going to work. This exactly is going to work like the founder is saying,” I know what to do with it, and how big can this be, and how to sell it, and what is the potential of that. So together, we complement each other very well. So we’re a very different type of VC. We are founders and operators with capital that invest and know the companies deeply.

Sramana Mitra: Now, I understand your investment thesis is vertical AI. So please tell us more. Double-click down on that investment thesis and tell us more on what you are thinking, how you are framing the vertical AI investment thesis.

Heriberto Diarte: Right. Vertical AI is a very important part of our focus areas. It’s not the only one. I’ll give the overview.

But talking about vertical AI, AI has gone through three distinct stages. The first one was the advent of the foundation models like the LLMs, like ChatGPT and others. Then we move into this craze of infrastructure deployment to be able to have those data centers.

And now we are entering the phase of vertical AI, which is applications. All the LLMs and all this infrastructure to particular use cases that create value. We are going to focus on that. Our strategy for how to do AI is, we’re not touching the foundation models. That’s for people who are much more capitalized. That race is well on. We don’t intend to discover the next Anthropic.

What we will invest is a little bit in the middle, when we find a good technology that can help de-bottleneck the rollout of the infrastructure in data centers. For example, passive cooling technologies, where you don’t have to spend electricity to cool your data center, and you can put all that energy into more compute power, or access to power today, and things like that. Things that can really help the rollout of the infrastructure. So that’s one thesis.

The other one is on vertical AI, where you can use AI to optimize, for example, a factory, get all the data and all the information of when the factory produces at the highest quality with the lowest cost, with the lowest waste, and so on. Look at those patterns and try to control the factory to replicate exactly that, so that you always have the optimum product with the highest quality and the lowest cost.

These are the sort of things we like. For example, we don’t like these very simple wrappers. We like what is called expert systems or things that use proprietary data and a lot of information, or, for example, the code of a country to be able to use and create something valuable. 

Another example would be in generative AI, a company that has all the code of how you build a water treatment plant. Because there’s a lot of code and restrictions and things you have to do, and then they have a huge database of the exact details of how water plants are built in that particular country. In combining these two, plus the plan of where is the land and what is the capacity, they can generate the blueprint and the bill of materials of the new plant, as well as if you had hired Jacobs Engineering and paid them 400 hours of engineering time to develop this plan, you can do it in minutes.

So that is the type of applications we’re looking for… where they use AI plus expert knowledge, and they tap into restrictions, and they optimize all of that.

This segment is a part in the series : 1Mby1M Virtual Accelerator AI Investor Forum