The silver bullet for democratization of robotics development is not AI. Although the robotics software (intelligence + sim) stack is radically changing (Zeromatter (Reactor), Skild, PI, Genesis) and becoming more open (NVIDIA's Isaac Sim), the real choke point for democratization of robotics is in production of critical components.

Until production is radically improved, any company building robot hardware is bottle-necked by their relationships and available line capacity with tier-1 suppliers. Supplier capacity is finite and is already being consumed by the companies who’ve been building purpose-built robot platforms for years (a few but mighty) or have the brand power to swing at general purpose (Figure, OpenAI, Tesla, etc.) and command capital to capture all supply on demand.

If next-gen foundation models progress as people think they will, general-purpose players will be first in line to translate SW leaps into real hardware by absorbing as much capital and supplier capacity as possible to lock in production. New entrants and existing purpose-built robot platforms, even the well-funded ones, won’t be able to compete if they can’t get capacity. Even if they co-develop with tier-2 or tier-3 suppliers, the first wave of winners may be growing so fast that there just may not be room and tier-2/3's might be eaten as well. The purpose-built robot companies will die in some n year period of ambiguity surrounding model capabilities, despite being good companies because capital directly drives supply.

Assume both Tesla and Figure form a duopoly in humanoid robots, creating, funding, and allowing a system to exists where two companies have the ability to displace and/or control 65% of the global workforce is really bad. Open sourcing, let alone increasing competition within, the humanoid software and hardware (K-scale) stack is simply not enough to mitigate this problem.

The world is headed in a truly dystopian track until we have x-China and decentralized matter stack: extraction (mining, drilling, etc.), processing (smelting, refining, etc.), and production (manufacturing). Matter suppliers across the stack shouldn't have the power to decide what companies get to produce physical goods.

We need to fund more suppliers. More mines, more steel, more solar, more nuclear, more high mix low volume shops, more everything.