How the Robot Monopolies Will Rise

Bringing the economics of the web to the physical world

Creating something useful and making money from that it is not as strong a guarantee as intuition leads us to believe. Take the internet: $10s of trillions of company value and consumer surplus and yet no one on the original team made a dime. Capturing dollars has much more to do with business model moats (e.g. network effects, defendable IP, capex) preventing competitors from increasing supply of your good or service than the absolute utility of your offering.

The maturing digital universe and recent advances in machine learning have given the necessary tools for treating the physical world like a modern computer – a system that allows simple commands to be issued and a guaranteed result to be returned. In the same way I can put together modular pieces of software, a web server, a rendering library, and a database, the same will be true for tangible world interaction. I can hook up a material supply drop off, to a robot that performs assembly, to a storage solution. Amazon Web Services illustrates this concretely, by abstracting the complexities of deployment a user can string along hundreds of different modules to build a scalable product and Amazon’s large investment in fixed costs allows them to reap the unit economics of the web. Companies have modularized the digital world and soon they will come for the physical.

The unit economics that made the internet revolution unique, primarily zero marginal distribution cost, will also exist in the new digital-physical paradigm. Hardware costs will trend towards zero, something unbeknown to most already happening in the robotic arm space, leaving the AI layer for robotic control as the obvious focus for the first movers in the space. For those seeking monopoly though, this is the wrong place to start. IP related to these solutions is diffuse in today's ML community. As an example of what a slog this will become, the self-driving car space has 10s of competitors on the roads today all deploying similar techniques with similar success. The techniques, and knowledge of how to make future progress, have become a commodity and it shows in the converging competition.

Will one eventually succeed before the others? Almost certainly, but right behind them will be a fleet of competitors driving down their market power. The winner will be the entity who now has commotized supply (the self driving companies) and has coalesced the demand (the riders). This sounds a lot like Uber. While it’s apparent the weaknesses Uber has it’s not a question about absolute power, but rather, a question about increasing the chance of attaining the most.

Given the same economic system will exist for all physical labor the same distribution of market power is a given. Suppliers, cut to size by a lack of competitive advantages, will be at the mercy of those that control demand and those that control demand will do so by removing the friction of deployment of automated solutions.

We’re in the IBM mainframe of automation today, but tomorrow will see an economy where labor is free and those who organize it reap the rewards.