This is a list of programs we would love to start.

Low activation jacket materials for fusion reactors.

Inverse design from structure → process.

Silicon-Protein Interfaces.  Proteins are one of the few ways we know to design nanoscale structures with atomic precision. Silicon lithography enables us to interface between nanoscale (though not-quite atomically precise) structures and our entire technology stack (through digital and analog electronics). It should be possible to combining these two normally separate technological paradigms by embedding proteins in lithographic silicon features which could enable everything from novel sensors to adaptive materials to applications we can’t yet think of.

Artificial Cells as chemical plants. Currently, chemical reactions are done in bulk and depend on everything in the tank coming to a new equilibrium. Living cells can be coaxed into synthesizing a subset of chemicals that don't kill them, but that subset is relatively small and opens the doors to runaway evolution. Abstractly, cells are systems that maintain an out-of-equilibrium state separate from the environment -- artificial, non-self-replicating cells could enable multiple reactions to happen in parallel and more efficiently, potentially entirely new processes.

General Purpose Telerobotics. Robotic autonomy is shockingly hard.  The ability to seamlessly act in the world through a robot anywhere in the world could enable everything from space manufacturing to vastly better utilized labs to improved eldercare.

Rethinking the supply chain with modern machines. The modern manufacturing tool stack has changed drastically since the majority of our industrial base was created, but the majority of products are still designed to be made en masse using older tools. It might be possible create new manufacturing processes based on modern tooling that enables things to be created closer to point of use, reduce costs, and enable designs that would otherwise be impossible.

Seamlessly Scalable mammalian cell cultures. The ability to seamlessly scale up successful small-scale mammalian cell cultures could dramatically reduce the cost of everything from viral vectors to lab-grown animal products. Currently scaling involves essentially redoing almost all the small-scale experiments to find correct parameters at larger scales with no guarantee of success.

Autonomous Design. It may be possible to use modern machine learning to design physical tools/systems in areas where human intuition (and thus our ability to design things) is severely limited. Additionally, many designs were created around manufacturing constraints, but drastically improved CNC and additive manufacturing technology has made it so that we can actually build bizarre designs that computers create. This has already been demonstrated for some engine parts and radar antenna.

Low-cost compliant robot arms. In order to achieve human-level robotic performance, we need arms that are all of: cheap, precise, strong, and don’t punch a hole in anything they bump into. Currently, you can choose ~2 of those.