The headline has changed: from “can it walk?” to “who pays?”
Tesla’s Optimus program has crossed an important rhetorical line. The question is no longer whether a humanoid robot can dance on stage, sort objects in a video, or wave from a trade-show booth. The question is whether Tesla can turn Optimus into a repeatable product, manufacture it at car-industry scale, and find enough customers to absorb that scale.
Fresh reporting this weekend frames the tension sharply. Tesla has begun early Optimus production at Fremont, on high-volume lines created from factory space once associated with Model S and Model X production, and Elon Musk has described a path toward roughly 20,000 robots per week by the end of 2027 . That run rate is about one million units per year, a number that sounds normal only if Optimus becomes less like today’s humanoid robots and more like a mass-market appliance, industrial tool, or labor platform.
Yet the buying market has not proven anything close to that appetite. On October 31, IoT News concluded that Optimus still cannot be bought or reserved, that Tesla has published no price, and that Musk’s late-2027 timing refers to deliveries to companies outside Tesla, not home consumers . In other words: Tesla may be building the factory logic before the market logic is fully visible.
Fremont is the bet
The physical symbolism matters. Reusing Fremont capacity for Optimus tells investors that Tesla wants the humanoid to be treated not as a research project, but as a manufactured product. The San Francisco Standard reported that Tesla has converted parts of its Fremont factory to produce humanoids and that Musk confirmed on the October earnings call that production had started there . Thomasnet’s fresh manufacturing coverage, echoed by other robotics outlets, says the Fremont line is designed for up to a million robots a year, although the ramp is expected to take at least a year and the company is still refining the manufacturing process .
That distinction is critical. A line designed for a million units is not the same thing as a line producing a million units. Robot factories, like car factories, pass through tooling, commissioning, yield improvement, supplier stabilization, quality control and rework phases. With humanoids, the challenge is even more severe because the product contains actuators, hands, batteries, sensors, compute, software and mechanical systems that must all survive real-world contact.
The latest reports also indicate that Tesla is not positioning the first wave as a consumer gadget. Humanoids Daily says Musk’s recent guidance puts deliveries to companies outside Tesla toward the end of 2027, with industrial work expected first, and no consumer delivery date has been provided . That makes strategic sense. A factory, warehouse or logistics site can measure whether a robot saves time, reduces injury risk, or fills shifts that are hard to staff. A household asks messier questions: Can it cook safely? Can it handle pets and children? Can it understand clutter? Can it fail gracefully?
The reservation confusion shows the demand problem
The past 72 hours also produced a mini-drama around reservations. The San Francisco Standard reported on October 30 that Tesla was expected to open customer reservations in early December, citing robotics investor and Humanoids Summit founder Modar Alaoui . The same article was later updated after Musk appeared to deny December waitlist plans on X with a one-word reply: “No” .
Humanoids Daily treated the December claim more cautiously, noting that the report pointed to a possible consumer waitlist rather than imminent home deliveries, and that reservation fees, final terms and consumer timing remained unresolved . IoT News went further: as of October 31, it found no Tesla order page, no official reservation process and no published price for Optimus . Musk’s archived post on October 30 is the cleanest public response so far to the December-ordering claim, but it still does not establish a replacement date .
That ambiguity matters because a waitlist can be mistaken for demand. A reservation page can generate headlines, deposits and social proof, but it does not prove that buyers will accept a robot once pricing, liability, maintenance, capabilities and delivery dates become concrete. Tesla has lived through that gap before with delayed products. Optimus will face a harder version of it because a humanoid robot is not only expensive hardware; it is also a promise about autonomy in human spaces.
The market is tiny compared with Tesla’s target
The most important context is the size of the current humanoid market. MACCUS reported on October 30 that China’s industry ministry says domestic factories built more than 40,000 humanoid robots in the first half of 2026 and forecasts more than 100,000 for the full year, but independent research firms counting shipments put global first-half deliveries lower, around 19,100 to about 25,000 units . The same report notes that Chinese makers accounted for more than 95% of the robots counted by IDC worldwide in the first half .
That means Tesla’s Fremont target, if achieved, would not merely compete with today’s market. It would overwhelm it. A million Optimus units per year would be roughly twenty times a 50,000-unit global annual market, using the lower commercial-shipment frame, and still far beyond China’s official 2026 production ambition. That is why the title question matters: “billions of robots” may be a long-term vision, but the near-term issue is whether the world has enough paying work for the first million.
There are reasons to believe demand could appear. Logistics centers, warehouses, factories, construction sites, elder care, hospitals and hazardous environments all contain tasks designed around the human body. If a robot can walk through human spaces, use human tools, work long shifts and improve through fleet learning, the addressable market becomes enormous. But “addressable” is not the same as “available.” Buyers will ask for uptime, speed, safety certification, maintenance cost, integration cost, insurance coverage and clear responsibility when something goes wrong.
China is proving both opportunity and warning
China offers the most useful comparison because it is already shipping or building humanoids at meaningful volumes. MACCUS found a sharp gap between Chinese official production figures and independent shipment estimates, suggesting that some robots may be staying in labs, training centers, demonstrations or inventory rather than reaching paying end users . It also reported that much early demand appears connected to state procurement, training centers, research and demonstrations, while commercial use remains uneven .
That should worry and encourage Tesla at the same time. It is encouraging because costs fall when a supply chain scales. It is worrying because supply can run ahead of useful demand. If a humanoid works at 20% of a human’s pace, needs constant supervision, or succeeds only in constrained demos, customers may delay purchases no matter how impressive the hardware looks.
The Researcher’s Guide to Physical AI summarized the week’s challenge bluntly: Tesla gave no unit count, no customer and no description of what the first robots will do, even as Musk set a 20,000-per-week target for late 2027 . That missing information is not a footnote. It is the core investment question.
What buyers will need to see
For Optimus to justify Fremont-scale production, Tesla must prove at least four things.
First, it must prove useful autonomy. A robot that needs frequent human rescue may still be valuable for data collection, but it is hard to sell at scale as labor.
Second, it must prove dexterity. Walking is difficult; hands are brutal. Factories and homes are full of deformable objects, odd angles, slippery packaging, cables, doors, drawers and tools. Manipulation, not stage movement, is where general-purpose robots become useful.
Third, Tesla must publish credible economics. A $20,000 target is often discussed, but IoT News emphasizes that Tesla has not published an actual price and that the quoted figure remains a forecast, not a product page . Buyers will care less about sticker price than total cost per useful hour.
Fourth, Tesla must show a deployment path. The first customers will likely be companies, not families, because companies can standardize environments and measure productivity. The home market may be larger emotionally, but it is also less forgiving technically.
The sudo moment
The joke in the French prompt is perfect: in a world full of robots, perhaps we will need to learn sudo to command them. But the more serious version is this: who gets permission to deploy physical AI into the real world, and who pays for that permission?
Tesla has made the manufacturing claim. The market now needs proof of work. If Optimus can perform boring, repetitive, dangerous tasks for long hours at a measurable cost advantage, Fremont’s million-unit ambition could look visionary. If not, the line risks becoming a symbol of a familiar tech problem: production capacity arriving before product-market fit.
For now, the robots may be coming. The buyers are still the mystery.
Comments
Be the first to comment.