I’m going to be honest about things that most VCs don’t say out loud. What keeps me up at night isn’t that the AI companies I back will fail. That means AI as a technology will be so successful that the business models of the companies I support will become completely irrelevant.
We are living in a period of real change in the industry. With the advent of AI-assisted application development, colloquially known as “vibecoding,” applications that once took years to build can now be replicated in weeks. And as big tech companies bundle AI capabilities into broader product suites, standalone chatbots lose pricing power overnight.
Startups in the human resources software space, which build digital tools to automate and manage employee-related tasks such as payroll and performance, will follow the same path as companies like Microsoft roll out similar products as part of broader bundles. Both developments highlight uncomfortable truths for venture capitalists right now. That means we risk paying too much for companies that look really special today but could become commonplace within a year, or worse, easily replicated by AI coding tools.
Let’s take a look at the pressure the SaaS subscription model (which has been the backbone of a range of venture capital-backed companies for years) is under in the public markets. Software stocks have lost more than $1 trillion in market capitalization during the so-called “SaaSpocalypse” that has engulfed capital markets since the beginning of this year. This is primarily due to concerns that AI agents will be able to perform entire workflows that previously required multiple SaaS subscriptions.
In other words, nothing I’m saying is theoretical. It’s actually already happening.
So how do we approach investing in this environment? My answer is to stop treating AI as a vertical and start treating it as a layer. and to focus on fundamental structures that technology cannot easily replace. Specifically I’m looking for three things. First, companies need to own customer trust and distribution. Second, it needs to be embedded in a system where real money moves. And third, you need to accumulate unique data that increases your efficiency over time.
In practice, this also means that you are more comfortable making multiple bets in a particular area than making one big bet on an AI-first business. Fraud compliance and security is one such area.
Exposure across the stack is key, from KYC and identity verification like our portfolio company Smile ID to transaction fraud monitoring like Orca. Over time, I expect these to overlap and the companies with the deepest data to win, regardless of which underlying model they use.
Similarly, I keep a close eye on agent platforms. Tools arrive on the market as generalized, but over time they become integrated into specific industry workflows. It will be difficult to replace companies that are building for specific niches.
But if I’m looking for the clearest silver lining in AI investing right now, I keep coming back to emerging markets, especially markets like Africa.
The continent is made up of 54 countries, each with its own regulatory environment and currency., And infrastructure. It is because of this fragmentation that global players in AI have largely ignored this in favor of the US and Europe. While this has long been considered a disadvantage to expanding business, it also offers local companies an opportunity to thrive while big tech companies focus on other regions.
Consider commodity trading. Platforms like Norrsken22-backed Sabi do more than just match buyers and sellers. It also coordinates supply, transportation, quality testing, storage, and financing across borders, where no system can reliably communicate with each other.
AI is being deployed to handle the tracking, tracking, and pricing of information flows throughout the chain. The company that solves this end-to-end becomes the default operating system for that market., The data that is accumulated during this process is extremely difficult for competitors to replicate.
This describes the pattern I’m noticing. These are companies that don’t just sell software, they run processes, accumulate local data that doesn’t exist anywhere else, and gain trust in the marketplace. the There is a lack of trust.
That’s not to say there’s little uncertainty in the industry at the moment. But we felt it was clear to stop asking, “Which AI companies should we support?” Instead, I started asking what problems leading companies were ignoring, and which companies were building solutions so deeply embedded in those problems that they wouldn’t be made obsolete by AI.
In emerging markets like Africa, these companies are everywhere. Investors just need to watch with interest.
The opinions expressed in Fortune.com commentary articles are solely those of the author and do not necessarily reflect the author’s opinions or beliefs. luck.
