Everyone is an AI company now. Here’s how we tell the difference.
AI startups raised $102 billion in the third quarter of 2026, 64% of all venture funding worldwide, according to Crunchbase. With that much capital flowing to one word, it’s no surprise the word now appears in almost every pitch deck we see.
Some of those companies are building remarkable technology. Others are using “AI” the way an earlier generation used “cloud” or “blockchain”: as a label that raises valuations faster than it raises revenue.
The cost of getting it wrong
The practice of overstating AI has a name: AI-washing. It isn’t harmless marketing. In 2024 the SEC brought its first enforcement actions over misleading AI claims, and later charged the founder of AI hiring startup Joonko with fraud.
The starkest example is Builder.ai, once valued at $1.5 billion and backed by Microsoft. Its “AI” coding assistant was largely powered by around 700 human engineers, and an audit found 2024 revenue closer to $50 million than the $220 million executives had claimed, eWEEK reported. The company entered insolvency in 2025.
The lesson isn’t to avoid AI. It’s to do the work.
Five questions we ask
When a company tells us it is an AI business, these are the questions we ask:
- What does the AI actually do? We want a specific answer: what goes in, what comes out, and what a person would otherwise do by hand. “It uses AI to optimize” is not an answer.
- Would the product still work without it? If the AI could be removed and customers wouldn’t notice, it’s a feature or a slogan, not the core of the business.
- What do they own that others can’t copy? A thin layer on top of someone else’s model can be rebuilt in a weekend. Durable AI companies own something harder: proprietary data, deep domain models, or science that took years to get right.
- Is it measurably better? Faster, cheaper, more accurate, by how much, and proven where? We look for results in a customer’s hands, not only in a demo.
- Do the economics work? Model and compute costs can quietly eat a company’s margins. We want to see that the business makes more money as it grows, not less.
What real AI looks like
The strongest AI companies we know rarely lead with the word. They lead with the problem. Physics-based AI that runs a full reservoir simulation in seconds instead of days is valuable because of what it saves an engineer, not because of what it’s called.
That’s the standard we hold ourselves to. We do deep diligence on the technology, including the parts labelled AI, and we are rigorous about the team behind it. AI will reshape nearly every industry we invest in. The founders who win will be the ones whose claims hold up on the first day of diligence and the thousandth day of operation.