The Wrappers Got Zero. Follow the Money to See Why.
Q1 funded agent infrastructure, vertical software, and silicon. The thin app over someone else’s model got almost nothing, and there’s a clean reason
.
The most-funded quarter in history had a clear loser. The “ChatGPT-for-X” app, the startup archetype everyone copied in 2024, captured almost none of Q1 2026’s roughly $300 billion. Strip out the four megacaps and the money that was left didn’t chase clever interfaces. It went to three places: agent infrastructure, vertical software that finishes a real job, and the silicon underneath both.
You can see it in the checks. DriveNets pulled $410 million for AI networking. Adaptive Innovations raised a $50 million Series A in home health by owning both the workflow engine and the service line, not just a chatbot bolted on top. Coralogix took growth money for observability in AI-era systems. Oxford Quantum Circuits funded the compute layer. Different sectors, same pattern. Investors paid for the hard part and skipped the part that’s easy to copy.
The mental model: Commoditize Your Complement
Joel Spolsky’s old line still runs the internet: smart companies try to commoditize their complements. A complement is something people buy alongside your product. When complements get cheap and plentiful, demand for your product goes up.
So ask who the complement is in the AI stack. The frontier labs own the scarce thing, the model. They want everything sitting next to it (the interface, the wrapper, the “just add a prompt” app) to be cheap and abundant. Cheap wrappers mean more model usage, which is exactly what the labs want. Build the thin wrapper and you’re building the product your single biggest supplier is actively working to make free.
That’s why the wrapper got nothing. Investors can see who’s commoditizing whom.
The companies that got funded own a complement the labs can’t flatten: proprietary workflow and data, a physical network, a regulated rail, real distribution into a service line. Those things make AI more valuable to the owner while the owner stays scarce. That’s the side of the trade you want to be on.
Spence’s take
The wrong lesson from “wrappers died” is “add more features.” A thicker wrapper is still a wrapper. The real question is simpler and meaner: does your company own something your biggest supplier wants to be free? If yes, you compound. If no, you’re a feature waiting to get absorbed into the next model release.
It’s also why hard, unglamorous layers age well. /mkt builds inside Reg A+ with tZERO’s trading infrastructure, which is the kind of scarce, regulated complement that AI can’t reduce to a free API call. The compliance and market plumbing is the hard-to-copy part, not the UI.
Build the piece that’s hard to copy. Let your suppliers fight to commoditize everything else.
If this was useful, share it with someone who builds things. And if you want the full toolkit of 50 mental models, my book is coming soon.
This post is for informational and educational purposes only. It is not investment advice, an offer to sell, or a solicitation of an offer to buy any security. References to specific companies, funding rounds, or platforms are illustrative and not recommendations. Securities offerings under Regulation A+ involve risk, including possible loss of principal, and are qualified by their official offering documents. Nothing here should be relied on for any investment decision. Consult a licensed professional before acting. Funding figures are drawn from publicly reported sources including Crunchbase and TechCrunch and reflect reporting at the time of writing; they may be revised.