Tasty Meals

I’ve been running a fun poll amongst AGI-pilled friends when they think humanoid robot will be as prevalent as self-driving cars today (2025).


(Natural Numbers Estimate)


A good prior is self-driving cars. Efforts truly kicked off with DARPA Grand Challenge in 2004.  Since then, over ~1,050 self-driving startups got funded, and ~57 or so had raised more than $100m, and at least 41 companies have licenses to test in CA DMV.


Even with all this venture-backed innovation, most laypersons today would only name Waymo and Tesla FSD as examples of self-driving cars. For context, Waymo has spent ~$30 billion since inception, and Tesla has spent ~$10 billion on computing, storage, and FSD training to scale to what they have today.


So from a pure spend perspective, we are early in the spending curve in robotics and likely the same or more will need to be spent in robotics.


But from a non-financial perspective, here are some other considerations debated: whether progress will exponential here on out, whether data collection for robotics could scale faster than that for self driving cars, or whether supply chain constraints could be disrupted (e.g. alternatives to neodymium and lithium are discovered).


Debate then mostly shifts to defining the milestone (e.g. you can buy Unitree humanoid today; should the benchmark be humanoids emptying a dishwasher at the same ability as a human; or more simply, Kalshi style, when do you think Tesla Optimus will make its first delivery), etc.


We landed on the milestone / definition that "robotic humanoids to be as prevalent as self-driving cars today” should mean that at least two major services / companies allows you to buy or use a humanoid / self-driving car with widespread agreement that the technology is as good as human driven action.


I can’t wait to use two competing humanoid robots to cook me a tasty meal, and running Hells Kitchen type of evals :P