Most AI Companies Won’t Exist in Five Years. Here’s How to Spot the Ones That Will

Everyone wants AI exposure right now. The problem? It’s nearly impossible to get it right.

The marquee private companies—OpenAI, Anthropic, Harvey—are available only to a select few mega-funds. Meanwhile, the rest of the private markets are drowning in startups that slapped “AI-powered” onto their pitch decks. The real issue isn’t the hype—it’s that the crystal ball is cloudier than ever. Distinguishing which companies will still exist in five years is harder than it’s ever been.

For founders: Adding AI to your deck won’t save you. Foundation models are getting better and cheaper by the quarter. And while you’re building your “AI-first” product, incumbents like Salesforce and Workday are integrating AI faster than you can acquire customers. Distribution beats innovation when the innovation can be copied in a sprint.

For investors: FOMO is not a strategy. Most of what you’re seeing will either get commoditized by foundation models or absorbed as a feature by incumbents. The middle is disappearing.

So what do we look for? Four things:

1. A system of record, not a product. We’re not interested in point solutions or tools. We’re looking for systems of record that sit at the center of operations, where customer, transaction, operational, or compliance data is stored and trusted as the source of truth. The software should be embedded in how the product or service is delivered, not just how it is analyzed. If the system disappears, the customer should feel real operational disruption, not just inconvenience.

2. Greenfield territory, not incumbent battlegrounds. There’s no point trying to unseat Salesforce or Workday in their core markets. We prefer companies that integrate with incumbents, complement them, or build systems of record in adjacent or newly emerging workflows rather than trying to replace them. The best opportunities are in neglected or underserved categories—where there’s no dominant player to fight. There’s also a massive opportunity to replace legacy SaaS tools that have failed to incorporate AI. When the old guard can’t or won’t adapt, AI-native platforms can capture entire markets

3. Large “AI” TAM. The market needs to be big enough to build a meaningful business. But TAM calculation in AI requires new thinking. Traditional SaaS spend doesn’t tell the full story. AI unlocks capabilities and replaces work that was never software-addressable before. The value delivered—and pricing potential—can far exceed what legacy tools captured. AI-native companies can expand the pie, not just capture a slice of existing spend.

4. An A+ team. This matters in every investment, but especially in AI. The technology is evolving so rapidly that teams need to be exceptional at both building product and adapting to shifting landscapes. And here’s the catch: AI has lowered the cost of building software—products can now be developed for a tenth of what they used to cost. Inferior teams raising modest amounts can flood the market with functional-but-forgettable products. The barrier to entry is lower, so execution quality matters more than ever.

Our recent investments reflect this thesis. Procurement Sciences is building a system of record for procurement—a massive, overlooked category with no dominant player. AI enables their solution, but the moat is becoming mission-critical infrastructure.

FieldPulse is another example: a fast follower to ServiceTitan in the fragmented home services professional market. They’re integrating AI features into their platform, but the real value is capturing an untapped segment and becoming the operational backbone for their customers.

Burro, our investment in AI robotics, replaces and supplements human labor, not software.

These aren’t “AI companies.” They’re companies using AI to build durable businesses in large markets that incumbents have ignored.

The AI hype cycle will pass. The companies that survive won’t be the ones with the best models—they’ll be the ones that built something customers can’t live without.

Build companies, not features. And if you’re investing, have the discipline to pass on everything else.