SemiAnalysis has published a 5-level autonomy framework for robotics, built around commercial viability, not theoretical capability. The framework, developed by Dylan Patel, Reyk Knuhtsen, and four industry co-authors including practitioners Niko Ciminelli, Joe Ryu, and Robert Ghilduta, defines each level by the capability unlocked and the reliability required to justify deployment cost. A robot that reaches Level 3 does not skip Levels 1 and 2. Progress is additive.
The core argument is simple: general-purpose robots are no longer a question of if, but when. Modern AI frameworks reframe prior hardware and software limitations as data problems. As real-world deployment data accumulates, robots gain skills faster and deploy into broader labor markets. The taxonomy is the first industry attempt to quantify where that progression stands today, grounded in current deployment data, use-case economics, and active research.
Read the full piece for the level-by-level breakdown. The methodology section, which separates reliability from throughput as distinct commercial thresholds, is the part most analysts will want to stress-test. SemiAnalysis says robots may target one level of tasks while still benefiting from capabilities developed at others. That claim alone has significant implications for how companies should be sequencing investment right now.
[READ ORIGINAL →]