Figure Built Robot #1,000 and Painted It Gold

Figure painted this robot gold because it is number 1,000, the 1,000th humanoid to roll off a single production line in California. In January, that line built one robot per day. By late April, one robot per hour. No Western company has ever shown a curve like that. And the ANA scoreboard still says China is far ahead. Both things are true. So let's run the The gold unit appeared on July 23rd, front and center, in a team photo outside a factory called BotQ, posted by Figure's founder Brett Adcock. Everyone shared the robot. The number behind it is the actual story, and Figure published that number trail itself. By late April, more than 350 robots built, and a demonstrated cycle time of one robot per hour. A 24 times jump in throughput in under 120 days. Three months later, unit number 1,000. The first line is built to make up to 12,000 robots a year. And Figure talks about a path to 100,000 over four years. How does a demo company learn to do that? By making the robot boring to build. Figure's previous robot was a prototype. Its parts carved from solid metal on CNC machines, some taking more than a week each. Figure 03 was redesigned around tooling, die casting, injection molding, metal stamping. The same class of part now comes out of a steel mold in under 20 seconds. Fewer parts, fewer assembly steps, and the expensive precision moved into the mold instead of into every single robot. Then, Figure wrapped the line in software. 150 networked work stations track every part and every test. More than 50 inspection points check sub-assemblies before they move on. And before any robot leaves, it has to pass more than 80 functional tests, plus a burn-in routine that sounds like a gym membership. Squats, shoulder presses, jogging, thousands of cycles. The results? In Figure's own numbers, more than 80% of robots now pass final test on the first try, and the battery line passes at 99.3%. Now the part the gold paint does not tell you. You cannot buy this robot. There is no price on Figure's website, no pre-order button, no retail channel, and no public map of where all 1,000 robots actually went. Figure says they feed internal research, data collection, its push to crack household chores, and customer sites. The one paid deployment you can actually look at is inside BMW's biggest American factory, in Spartanburg, South Carolina. Figure's previous model worked there for 11 months, feeding sheet metal parts into a line that BMW says supported production of more than 30,000 cars. This summer, Figure 03 took over a harder job, pulling unsorted parts out of containers and loading them into sequencing carts shift after shift. The billing rate is reported at roughly $25 per robot hour. And notice what is still missing from all of it. No public uptime numbers, no intervention rates, no revenue per robot. A factory that publishes its yield is still choosing which numbers to publish. So where does 1,000 robots put Figure on the world scoreboard? That depends on reading the scoreboard correctly, because every company is counting a different thing. Figure's 1,000 is robots manufactured, on one line, one model. In Shanghai, Agibot rolled its 15,000th robot off the line in late June, cumulative production across its whole robot family, and accelerating. The first 5,000 took about a year. The next 5,000 took three months. Unitry, the company behind the viral $13,500 humanoid, shipped around 5,500 humanoids in 2025 alone. Actual deliveries to actual buyers. And the company is profitable. Chinese regulators just cleared it to raise over $600 million on the Shanghai stock market. And Ubitech took more than 13,000 orders for its new consumer humanoid on launch day. Orders, not deliveries. The first units shipped in September. Manufactured, cumulative, shipped, ordered. Four different verbs, and on every one of them, China's numbers are bigger. But look at what each side chooses to publish. Agibot publishes its totals. Figure publishes its yield rates, its cycle times, its inspection counts, the factory homework. One side is proving volume. The other is proving in public that it has learned the process. Those are different kinds of evidence. And for the first time, the West actually has one of them. There is one more player, and it refuses to fit this scoreboard at all. Tesla has published no production count for Optimus. No disclosed number, no customers, no price. What Tesla is building is capacity. It tore out the Fremont line that built the Model S for 14 years and is installing an Optimus line in its place, with production expected, in Tesla's words, later this year. And the first robots going to Tesla's own training academy, not to customers. Meanwhile, at Giga Texas, drone pilot Joe Tegtmeyer flew over the of a second, far larger Optimus factory. About 600 feet of structure so far, seven cranes working. Tesla is betting it can skip the small batch years entirely. So far, that bet has produced spectacular buildings. And no public robots. So three manufacturing bets are now on the table at the same time. China is betting on volume. Build many, sell cheap, iterate fast. And the price floor keeps falling. Figure is betting on yield. Fewer robots, but a documented factory that improves every week, with a paying industrial customer while it learns. Tesla is betting on capacity. Build the biggest robot factories on Earth first, fill them later. Nobody knows which bet wins. But now you know what to watch. And it is not the demo reels. Watch whether figures one per hour becomes a sustained rate instead of a demonstrated one. Watch whether a price ever appears on that website. Watch what actually comes out of Tesla's buildings. And how fast China's 15,000 becomes 30. A demo tells you what one robot can do on its best day. A factory tells you how many times a company can repeat it. And for the first time, an American factory is worth watching.