The Stack — Ep. 01: Owning markets, money, and intelligence

Welcome to The Stack — FTG's series on building the infrastructure of the digital economy. We made this show because the interesting conversations about this stuff tend to happen privately, between people who are actually building it, and we thought it was worth having them out loud. Episode one lays out the whole thesis: why markets, money, and intelligence are converging into a single stack, and why we chose to build and operate that stack rather than fund it from a distance. These are the notes; if you read one section, read the takeaways.

The argument, in brief

We start from a single observation and follow it to an uncomfortable conclusion. The observation: money is becoming machine-readable — programmable, composable, settleable by software — at the exact moment that software is becoming an agent that can hold intent and act. The conclusion: value that software can read, software will eventually move, which turns the interface between money and machine intelligence into one of the most important pieces of infrastructure of the decade. Then we get concrete about what it takes to actually build that interface, which is where most of the episode lives.

What we get into

• The two curves. Tokenization of value and agentic intelligence, and why their meeting point is the thing worth owning rather than the individual curves.
• The seams problem. Why the value at this convergence sits between the layers — in shared identity, memory, and settlement — and why a market of rival single-layer companies structurally cannot build it.
• Three positions, one system. Markets, money, and intelligence, connected by one backbone, and why owning one of each is the price of admission to building the connective tissue.
• The compounding loop. How liquidity, the wallet, and the AI make each other more valuable, and why loops are so much harder to copy than features.
• Operator-investor, in practice. What it actually means to ship an exchange, a wallet, and an AI arm — the custody discipline, the key security, the model orchestration — instead of advising from a board seat.

Takeaways

1. The prize is the seams, not any single layer. Owning markets, or money, or intelligence, is a business. Owning the connective tissue between them is a compounding one, and it is the only part a fragmented market will leave on the table.
2. Build-and-operate is an edge, not a temperament. In infrastructure, the teams that run the thing learn what is real faster and can originate the companies the thesis requires. Speed to truth compounds.
3. Loops beat features. A feature is copied in a quarter. A loop across three owned layers has to be reproduced all at once, seams included, or it does not compound at all.

We are not building features. We are assembling the operating stack for how value will move, be secured, and be acted upon.

Later episodes go deeper on each layer, and on the harder questions we are still working through. Next up: sovereign AI, and the moment "where does the model run" became a question buyers actually ask.