Acquisitions are easy to understand. Scientific breakthroughs are hard to understand. And yet, the second will probably change more.

In one week we got two headlines from OpenAI.

One was clear. The acquisition of OpenCLAW's founder. Agent-management infrastructure that became the talk of the day.

The second almost disappeared from the feed. GPT 5.2 helped derive a new formula in quantum field theory. Something that until now was considered zero in all cases.

The first touches engineering. The second touches scientific breakthroughs. One moves the market in the short term. The other changes the boundaries of what is possible.

It is easy for us to react to the first. It is hard to free up attention for the second. And we choose to build on the balance between them.

Because some of the layers that feel strategic today will become commodity faster than we think.

We are not looking only for immediate impact, and not for theoretical research depth. We are looking for the balance: companies that create measurable value now, while building a real layer of depth. Data that accumulates. A control point that forms. Switching costs that are not just convenience.

Not an impressive demo. Not brilliant research without a market.

If you are building something like that, something that balances the short term and the long term, something that looks today like a point solution and tomorrow becomes a control layer, I would be glad to dive together into how today's product becomes tomorrow's infrastructure.

Because in this era, the challenge is not only speed. It is the allocation of attention. And who we choose to build with, for the layers that will hold a decade out.