Machines Have Won. That is the starting point, not the argument.
The argument ended years ago, when it became clear that machines had crossed the thresholds we once thought uniquely human: understanding language, interpreting images, generating code, reasoning across complex systems. The bet that machines would win is now consensus. A decade ago it was a contrarian position. Today it is the baseline assumption of every fund with “AI” in its mandate.
What’s interesting now is what comes next, and who builds it. This summer, we let the three-vector framework go. That framework was the right instrument for an earlier period; it helped us distinguish the kinds of machine intelligence we were backing and why. Then, we replaced it.
What took its place is not a new framework in the structural sense. It is a set of beliefs: specific, ranked, debated, and tested against the portfolio. Over the course of many sessions, our team surfaced proposals, scored them, argued them down, let them sit, and brought them back changed. The beliefs that survived were the ones that wouldn’t leave. And when we mapped them against what we had already built, the pattern was already there: sub-portfolios of companies clustering around each conviction.
We call them The Futures We Believe In.
A future where the enterprise can both “hire” and manage any digital specialist while maintaining visibility and control across these AI Agents and their outputs.
This is the agentic enterprise: not AI as a tool you call, but AI as a workforce you manage. WARP, Wayfound, Superintelligent, and Orita are each building toward a world where orchestration, accountability, and oversight are first-class products.
A future where the enterprise securely leverages best-in-class general models, tuned to their needs and their proprietary data, in a seamless and unified experience.
The next wave of trillion-dollar companies will be built by combining foundation models with proprietary data. Not fine-tuning alone; full-stack trust. Okareo, Infactory, TensorStax, and Atym are building the evaluation, enrichment, and integration layer that makes that real.
A future where hardware takes mere months to design, prototype, test and then manufacture at scale.
Product-market fit in deeptech should now be measured in months, not years. Unfair access to data paired with earned expertise unlocks what used to require decades. C-Infinity, Drafter, Reshape, and Zenode are compressing that timeline.
A future where the boundary of humanity’s current understanding of the laws of physics are not the limiter to new product design.
Heat resistance. Weight. Durability. Radiation tolerance. The materials constraints that define what is possible in aerospace, defense, energy, and infrastructure. N-ERGY and Besxar are working at the frontier of what matter can do.
The Four Futures are predictions. They are also where we are already invested. Each one has a sub-portfolio behind it, founders who walked through brick walls to get there, and a team that debated the conviction until only the real ones remained.