The important companies of this decade will not stay on a screen. They will make chips, move energy, sequence biology, fly machines, and run the loops that used to take a ministry. That is already visible. The open question is when you get involved.
Most of the capital shows up when the system is already state-scale: a fab, a reactor fuel line, a defense program, a lab with a waitlist. That is a real job. It is not our job.
Baus lives in the years before the machine is obvious. When the insight is still uncomfortable. When the team still fits around one table. When the product is an ugly demo that changes the economics of a real workflow. Zero, then one. Not ten.
The same map, earlier
The next industrial stack has a shape, even if the logos are not finished.
Compute. Models are hungry. The constraint is silicon, power, and the software that makes new hardware usable. We care about teams who can move that constraint, not decks that admire it.
Intelligence. Agents that do the work, not chat that describes it. Data loops that get better because they are in the critical path. Tools that give a three-person company the leverage of thirty.
Biology. Measurement, design, and intervention as software problems. Sequencing, assays, closed-loop experiments. Not a slide about “AI plus biotech.”
Energy and matter. Power, materials, logistics, the physical plant. Intelligence that changes how atoms are made, stored, or moved.
Control. Robots, vehicles, autonomy. The last mile between a model and the world. If it cannot act, it is a report.
This is not a shopping list. It is a filter. If the product can stay a thin UI forever, it is not this thesis. If the product has to leave the screen to matter, it might be.
Why a studio belongs here
State-scale systems are assembled by people who started earlier than the press release. Someone formed the company, found the first customer, and refused to wait for a perfect syndicate.
That is the studio job. Originate the idea. Partner with a founder who already cannot stop thinking about it. Make a real thing before the category has a name. Then, when there is velocity, back it like a fund.
We do not need to own a reactor to care about energy. We need to be in the room when a small team finds a non-obvious way to make power, compute, or biology cheaper, faster, or newly possible.
A pre-seed check does not underwrite a gigafactory. It underwrites the team that can make the next plant, chip, or assay obvious. A wrapper around a public model is a feature. Neither is this thesis.
Who this is for
If you are building something that will one day look like infrastructure, and it does not look like that yet, we should talk. Send the sharpest version of the edge: what is true that the market has not priced, what you have made real, and what you need next.