Musk Says No One Is 'Even Close' to Optimus. The Same Week, Figure's Robot Walked 13 Days Straight and Agility Signed Another Factory. Here's What Each Receipt Proves.

Honest fence (read first): Nobody can tell you who's "winning" the humanoid race from the outside. No company publishes comparable benchmarks, and nearly every number below comes from the company itself. This post doesn't crown anyone. It puts Musk's claim next to the rivals' dated, checkable receipts from the same week, so you can see what each one does and doesn't prove.

Editorial Thesis

On Oct 21, during Tesla's Q3 call, Musk said: "to the best of my knowledge, there is no one that is even close to Tesla with respect to a general purpose humanoid robot." Later he added, "I'd be stunned if anyone has a robot that is remotely competitive with Optimus when we ... start delivering it to customers, which is likely to be towards the end of next year." That's a strong claim with a date on it. The useful question isn't whether it's true today. Nobody outside can check that. It's what evidence each company is putting on the table, and what kind of evidence it is.

Honest Verdict

Believe that Musk said it, and that it's a forecast about late 2027, not a measurement. Believe the rivals' receipts for exactly what they are: Figure's is an endurance test it ran and reported itself, Agility's is a signed pilot that doesn't start until 2027, and Boston Dynamics' is a production start with a customer list made of its owner and a research partner. Don't believe anyone, Tesla included, who says the race is already decided.

The claim, in full context

Musk made the remark after describing what Optimus still needs: a better hand, longer usage life and better AI, all of which he expects to come together "around the end of next year." So he's not saying Optimus is ahead of everyone at doing useful work today. He's saying no rival will match it when Tesla starts selling. That's a prediction about a product that hasn't shipped, compared against products that are still changing.

Receipt 1: Figure F.03, the 13-day treadmill run

• What happened: On Oct 19, Figure CEO Brett Adcock posted on X that an F.03 had "just crossed 9 straight days walking on this treadmill: 348 miles and 1.2 million steps," a test that "started as a side project to stress the robot for any failures" (via heise, Oct 20). By Oct 23, his updated count was 13 days, 422 miles and 1.5 million steps (Robotics Weekly).
• What it proves: One F.03 kept walking, on its own legs, for nearly two weeks. That's a real durability data point, and durability is exactly the "can't wear out after a month" problem Musk named on Tesla's call.
• What it doesn't prove: heise notes the robot ran on wired external power, and Figure didn't say whether anything failed or got swapped during the run. Walking on a treadmill isn't factory work. Figure's earlier claim, three F.03s sorting 249,560 packages over roughly 200 hours in May, is closer to real work, but it's still Figure's own count.

Receipt 2: Agility Digit, a signed factory pilot

• What happened: On Oct 20, AGCO, the farm-equipment maker behind Fendt and Massey Ferguson, announced a commercial agreement to pilot Agility's Digit at its Hesston, Kansas plant. The pilot is "expected to begin in 2027," starting with parts sorting and stacking and placing totes for storage. AGCO joins Toyota, GXO and Schaeffler on Agility's named-customer list.
• Same week: On Oct 22, Agility announced it's expanding its NVIDIA work as it prepares Digit 5 for commercial scale. Earlier that week it named an audit committee chair for its planned public-company board.
• What it proves: A named manufacturer signed a paper agreement to put a humanoid on a real line, for specific, boring tasks. That's the kind of receipt Tesla hasn't shown for outside customers yet.
• What it doesn't prove: A pilot isn't a fleet. It starts next year, and the release gives no robot count, price or success target.

Receipt 3: Boston Dynamics Atlas, a production start with a sold-out year

• What happened (context, January): At CES on Jan 5, 2026, Boston Dynamics said it would begin making the product version of Atlas immediately, and that "all Atlas deployments are already fully committed for 2026," to Hyundai's Robotics Metaplant Application Center and Google DeepMind. More customers are planned for early 2027. Electrek reported Oct 22 that Hyundai opened an Atlas training center at its Georgia Metaplant in September.
• What it proves: Atlas is being built and shipped to named sites this year.
• What it doesn't prove: Both 2026 customers are close partners: Hyundai owns Boston Dynamics, and DeepMind supplies its AI. That's a lot like Tesla putting Optimus in its own factories first.

One more for context: XPeng IRON

XPeng commissioned an IRON production line in Guangzhou on Sep 8, saying more than 80% of core processes are automated, with mass production by the end of 2026 and deliveries in 2027. That's older than this week's news, but it's the closest match to Tesla's own schedule. Electrek's Fred Lambert argued Oct 22 that XPeng and Boston Dynamics "look at least as far along."

Claim vs proof

Why the receipts matter

Every company here faces the three problems Musk listed on his own call: hands that survive real work, robots that last, and AI that copes with new tasks. This week's useful receipts aimed at those: a durability run, a named factory job, a production line. Tesla's answer is a promised unscripted video "in a month or two." Judge it the same way when it lands.

What to watch next

• Tesla's promised video: unscripted, by Musk's description. Check for cuts and teleoperation notes.
• Figure: whether the treadmill run reports a failure, and whether any customer deployment gets named numbers.
• AGCO and Agility: the 2027 pilot start, and whether Digit 5 ships on schedule.
• Boston Dynamics: the "early 2027" customer list beyond Hyundai and DeepMind.

Video note

Ryan Shaw's breakdown (gS1iOU_JnkA, uploaded Sat Oct 24 at 12:00 PM ET, 37:43) walks through Musk's "no one is even close" line and compares Figure, Agility, Boston Dynamics, 1X and Chinese makers. Ignore the "Leaked Optimus 3" title: Tesla hasn't revealed V3. Alternate: Optimus Wire's 90-second explainer on Figure's 348-mile run (DfH3GJk-xUc, Oct 19), a small channel. The facts here come from the transcripts, press releases and written reports, not the videos.