The AI Funding Gold Rush Is Minting Winners and a Lot of Roadkill

TL;DR — Capital is rushing into anything that can spell AI, on terms that look detached from the underlying businesses. A handful of these companies will be generational. Many are thin wrappers around someone else’s model with no defensible moat, and the gap between the two is about to get very visible.

There’s a particular energy to a gold rush, and right now it’s pointed at AI. Founders can raise eye-watering sums on a deck and a demo. Valuations have unhooked from anything resembling revenue. If you can credibly say the letters “A” and “I” in a sentence, money finds you.

Some of that money is building the genuinely important companies of the next decade. A lot of it is lighting itself on fire in slow motion.

Cheap to start, brutal to defend

The uncomfortable truth under the boom is that it has never been easier to build an impressive AI demo, and rarely been harder to build a defensible AI business.

The reason is the same trend driving everything else in our AI coverage: the underlying models are a commodity you rent. That’s wonderful for getting started and terrible for staying ahead. If your whole product is a thin layer over a model anyone can call, then anyone can call it, and your moat is a weekend project for a competitor. The ones that last own something the model doesn’t: proprietary data, a real distribution advantage, or a workflow customers can’t easily rip out.

A laptop showing growth charts and metrics — Photo by Carlos Muza on Unsplash

The wrapper problem

Investors have a name for the risky version: the wrapper. A nice interface around a model, with little underneath. In a boom, wrappers raise just fine, because the demo dazzles and the market is impatient.

The trouble arrives later, when the model providers themselves ship the same feature as a checkbox, or a better-funded rival undercuts on price. Plenty of today’s hot AI apps are one platform update away from irrelevance, and the funding rarely buys protection from that, just runway to find out.

An abstract representation of artificial intelligence — Photo by Growtika on Unsplash

How to tell the two apart

Strip away the pitch and ask one question: what does this company have that a competitor with the same model couldn’t replicate in a month? If the honest answer is “nothing yet,” that’s not a company, it’s a feature waiting to be absorbed.

The real ones have an answer. Unique data that compounds, a distribution channel that’s hard to copy, deep integration into how customers actually work, the kind of edge our startups coverage keeps coming back to. Booms are great at hiding the difference. The correction that follows every gold rush is very good at revealing it, and that reckoning tends to arrive right when the easy money stops.