BIG TOKEN AND THE STARTUP ECOSYSTEM With recent tech M&A (Scale - Meta, Windsurf - Google), the attention has fixated on the unusual deal structures, typically founder hiring & IP licensing agreements in lieu of clean cut acquisitions. To me, these recent transactions also signal a shift in the potential exit path for startups and a change to technology market structure itself. The acquirers represent a small set of companies with the adequate resources and competitive positioning to actually pursue AGI. They are OpenAI, Anthropic, Google (DeepMind), xAI, and Meta. I saw the name “Big Token” and found that to be an apt descriptor. They have access to lots of capital (measured in the tens to hundreds of billions), active large model training efforts, and Internet / mobile properties to distribute model-centric products and continue to collect data. There are increasing parallels in technology & AI to the oligopoly structure that the energy and pharmaceutical industries have. The energy industry has supermajors such as Exxon Mobil, Shell, Chevron, BP, TotalEnergies. Big Pharma is companies like Pfizer, J&J, Roche, Novartis, and several others. In these spaces (and other mature industries), startups and smaller companies live more as outsourced R&D functions versus designed to grow quickly and exist as an independent, cash flowing entity one day. Successful small firms in energy and bio are acquired by or enter special licensing or JV deals with one of the few big firms. They are the ones with the existing scale, infrastructure, and assets to commercialize most of the tech. My belief is that part of the technology and startup market will break in a similar way, especially as the orbits of Big Token become larger. This could change, but the barriers to entry feel quite high. There are three major inputs to AGI – data, compute, and talent. And Big Token will scale all of those up in any way they can. Acquiring startups with access to unique tokens as a means of data acquisition, vertically integrating their compute stacks, and picking off talented teams aggregated through equity capital. All of the above have already happened. I don’t see this as the end state for every successful venture-backed startup. There will be plenty that go public and exist independently. Current venture fund sizes need that to happen in order for the math to math. Yet, this could become a real path for founders, employees, and investors (smaller funds could generate meaningful DPI). The value wouldn’t stem from discounted cash flow, but rather, as an input into AGI – licensing, sole supplier, acquisition, or any form of general absorption by Big Token.

Présentation de vmux - des environnements cloud étatful incroyablement rapides pour les agents de codage pour la première fois, vous obtenez des environnements GPU/CPU persistants via Modal/CF soutenus par des Objets Durables pour diffuser des journaux en direct, des URL d'aperçu natif et attacher un véritable shell lancez un notebook ou entraînez nanogpt via codex - avec un environnement Modal lancé en quelques secondes