Brad Porter backs Eidon's read on the robot data market
The former Scale AI CTO says he lived the data collection treadmill first hand, and it ends the way Eidon's founder described.
Brad Porter, who was CTO at Scale AI, has weighed in on Eidon AI's shutdown post and says the founder got the economics right. He quoted the line about data providers being a bootstrapping force, and added that he lived this first hand.
Sam Padilla announced the shutdown after two-plus years in robotics data, saying the thesis was right but the business is brutally hard. Eidon is open-sourcing everything it built and publishing lessons for anyone else going into the space. We covered the announcement and the first reactions earlier.
The thesis was right. But the business is brutally hard.
The flywheel argument
The passage Porter pulled out is the core of it. Once a lab can fill its own data needs, either with an internal collection team or with data coming off its own deployed fleet, the need for third-party collectors goes away. Porter says collection and labeling can be lucrative in the short run, but only if you keep chasing the next data frontier, and that is hard. His read on his old employer is that Scale AI is very good at realigning to that next frontier, and that survival depends on doing it.
That sits next to Ted Xiao's take on the same post. Xiao called it the best honest account of the market and said there is a brief arbitrage window open right now, but that the market is saturated, entry costs are higher than people expect, and the number of real buyers is in the single digits.
the market is saturated, entry costs are surprisingly non-trivial, and there are a single digit number of real buyers
What this does not tell you
None of the three say demand for robot data is falling. The argument is about who captures it. If the labs building the models also own the fleets generating the trajectories, the value stays inside the labs and the vendors sitting between them get squeezed. Nobody in these posts puts a number on how many data startups are in the same position as Eidon.
the need for third-party data collectors will evaporate
Sound like you were at the demo
One email a day. Under three minutes. The demos, the deployments and the part the press release left out.