Figure 03 robot speeds manufacturing 24x

Humanoid robotics has had a weird gap for years. Companies could build a flashy demo robot, but turning that into a repeatable factory product was the hard part. Figure says that gap just narrowed a lot. On April 29, the company said its BotQ plant ramped Figure 03 production from one robot per day to one robot per hour in under 120 days, with more than 350 third-generation robots delivered so far. (figure.ai) ### What actually changed? The news is not that Figure invented a new humanoid this week. Figure 03 launched in October 2025 as the company’s production-focused redesign — built for affordability and high-volume manufacturing instead of prototype-style fabrication. What changed now is the factory ramp. Figure says BotQ has moved from blue(figure.ai)ot-per-hour cycle time needed for its production targets. (figure.ai) ### Why is 24x a big deal? Because the base rate matters. Going from one robot a day to one robot an hour is not a small optimization — it changes the shape of the business. At one a day, you are still basically feeding pilots and internal testing. At one an hour, you can start building a fleet fast enough to learn from it, break it, fix it, and(figure.ai)is the real point — more robots in the field means more data streams for autonomy. (figure.ai) ### Did this come from AI perception alone? Not from the evidence Figure published. The company’s own writeup ties the ramp mostly to manufacturing execution software, supplier quality work, in-process inspections, testing, and a production-oriented hardware architecture. It says the line runs on custom manufacturing software across more (figure.ai) and 80+ functional verification tests before sign-off. That is a factory story first. (figure.ai) ### So where does the software story fit? It fits in the next step. Figure explicitly links higher production volume to faster progress in autonomy, and it frames the payoff as “perception-conditioned whole-body control.” That language lines up with the company’s Helix work — models that control walking, balance, manipulation, and reasoni(figure.ai)hed as the fuel source for the AI loop, not just a way to ship hardware faster. (figure.ai) ### Why was Figure 03 easier to scale? Because Figure changed the robot itself months ago. BotQ’s March 2025 manufacturing post said Figure 02 still leaned on slow CNC-heavy prototype processes, while Figure 03 was redesigned around tooled manufacturing like injection molding, die casting, metal injection molding, and stamping. Figure sai(figure.ai)e in under 20 seconds with molds. That is the kind of redesign that makes a one-per-hour line plausible. (figure.ai) ### Is the factory healthy yet? Healthy enough to be interesting, but not finished. Figure says end-of-line first-pass yield is now above 80% and improving weekly, while its battery line has reached 99.3% first-pass yield and shipped more than 500 battery packs. Those are real manufacturing metrics, and they suggest the company is moving past pure prototype chao(figure.ai)an active ramp, not a mature consumer-electronics line. (figure.ai) ### Why does this matter beyond Figure? Because the humanoid race has been short on proof that anyone can bridge the last mile from “look what the robot can do” to “look how many we can build.” BotQ was introduced in March 2025 with a first-generation target of up to 12,000 humanoids per year. Figure has not said it is there yet, but this(figure.ai) manufacturing scale itself into a product advantage. (figure.ai) ### Bottom line? The headline is not really “AI made robots 24x faster.” The cleaner read is simpler — Figure says it has finally started to look like a manufacturer, not just a lab. If that holds, the next moat in humanoids may be less about the best demo and more about who can build, test, deploy, and learn the fastest. (figure.ai)