A humanoid robot cost Morgan Stanley Research about $200,000 in 2024, in a high-income country. That number gets repeated constantly. What gets repeated far less is the number underneath it: once you divide that price by the hours a robot can actually work, some machines already come out cheaper than the person standing next to them on the line.
That is the real question buried inside every "is a humanoid robot worth it" headline. Not the sticker price. The hourly rate.
The Sticker Price: What a Humanoid Robot Costs Right Now
Morgan Stanley Research puts the average price of a humanoid robot in a high-income country at around $200,000 in 2024. The bank's original 2024 report projected that number falling to about $150,000 by 2028 and $50,000 by 2050. In countries that lean on cheaper Chinese supply chains, the bank estimated prices could fall as low as $15,000 by 2050.
Those figures are already out of date on the low end. Unitree, the Hangzhou-based robotics maker, launched its R1 model in July 2025 at $5,900, undercutting the bank's 2050 China-supply-chain estimate by nearly a decade. The R1 weighs 55 pounds, has 26 joints, and can do a passable cartwheel. It cannot fold laundry or load a dishwasher. For a closer look at how cheap humanoid hardware is getting and what corners get cut to hit that price, see our breakdown of the $5,000 robot hand that exposes the industry's real bottleneck.
The rest of the market sits well above that. Unitree's own G1 lists at roughly $13,500 to $16,000 depending on configuration. The flagship H1 runs about $90,000. UBTech's U1, aimed at industrial buyers, prices at $145,000, a figure we covered in detail here. Figure AI's Figure 02, currently handling sheet-metal tasks at BMW's Spartanburg plant, is informally priced around $50,000. Apptronik's Apollo, in trials at Mercedes-Benz facilities, is targeting a sub-$50,000 list price once production scales.
| Model | Maker | Price |
|---|---|---|
| R1 | Unitree | $5,900 |
| G1 | Unitree | $13,500-$16,000 |
| Figure 02 | Figure AI | ~$50,000 (informal) |
| Apollo | Apptronik | Sub-$50,000 (target) |
| H1 | Unitree | ~$90,000 |
| U1 | UBTech | $145,000 |
Robot Cost Per Hour vs a Human Worker
Sticker price is the wrong unit for comparing a robot to a worker. Hourly cost is the right one, and it changes the picture considerably.
Take a $50,000 humanoid running two shifts a day. Amortized over a five-year useful life with roughly $25,000 in annual operating costs, industry cost models put its effective hourly rate around $9 to $10. A fully loaded human warehouse worker in the US, once wages, benefits, payroll tax, and turnover are added in, runs $22 to $45 an hour depending on the role and region. A German industrial comparison found an even starker gap: about €12 an hour for a robot against roughly €61 for a human worker, once training, sick leave, and mandated breaks are priced in.
The catch sits in what that hour of robot labor actually produces. A human warehouse worker moves 60 to 80 totes an hour. Current humanoid pilots manage 30 to 50. The hourly rate looks great until you account for the fact that the robot is doing less work in that hour, and that its uptime, battery life, and error rate still lag a trained person on a good day.
Morgan Stanley's Own 2050 Number Just Moved
Here is a detail that gets lost every time the 2024 Morgan Stanley figures get recycled: the bank has already revised them.
A more recent Morgan Stanley outlook, reported in January 2026, pushes the 2050 forecast for high-income countries up from $50,000 to about $75,000. For middle- and low-income countries, the bank's number moved down slightly, from $50,000 to around $21,000. The direction is not uniform. Rich-world prices are now expected to stay higher for longer than the bank originally thought, while the low end keeps getting cheaper.
This has not been confirmed by any newer public statement narrowing the range further, and Morgan Stanley has not published a detailed explanation for the revision beyond citing production and component costs. Worth checking directly with the bank's research desk if the exact reasoning matters to your use case.
Three Analysts, Three Different Timelines
On Al Jazeera's Counting the Cost, a panel was asked a version of this article's core question: how cheap do humanoid robots need to get before they are worth it. The three answers pulled in different directions, and none of them agreed on a number.
One panelist, Allen, argued the calculation is not purely about price. Robots do not get sick, do not go on strike, and do not complain, and on that basis he expects adoption to move faster than bank forecasts suggest. A second panelist, Z Chen, leaned on China's manufacturing scale and supply chain strength as the more likely driver of falling costs, a dynamic we've written about separately in the context of how Chinese humanoid makers ship functional robots without chasing Boston Dynamics-level polish.
The third panelist, Yianja, split the timeline in two. Industrial and commercial deployment, she argued, is already underway and will keep expanding over the next three to five years. Home robots are a different story entirely, at least 10 to 12 years out in her view, largely because the data and AI needed to handle general household tasks reliably is not there yet, even at a lower price point.
That last point matters more than the price debate. A cheap robot that cannot reliably load a dishwasher without breaking a glass is not a bargain. It is a toy with good marketing.
What $5,900 Actually Buys You Today
The Unitree R1 is the clearest real-world data point available for how far prices have already fallen, and it is worth being specific about what that price includes. The base R1 ships with a fixed hand, one camera on the cheapest configuration, and roughly an hour of battery life. It can walk, run, recover from a fall, and perform demo-friendly moves like cartwheels and handstands. It is marketed at individual developers and small research teams, not households.
The version that can actually grip and manipulate objects, the EDU tier, costs significantly more, and Unitree has not published a fixed price for it publicly. Reviewers who have tested the R1 describe it plainly as a research and hobbyist platform rather than a functioning household assistant. If leasing rather than owning ends up being the more realistic path to affordability, that model already exists elsewhere in healthcare robotics, where hospitals are renting robots on subscription rather than buying them outright.
My Take
The hourly-cost framing is the only honest way to compare a robot to a worker, and most coverage skips it entirely in favor of the sticker price, because $200,000 is a better headline than $9 an hour. Run the actual math and the picture splits cleanly into two markets that get talked about as one. Narrow, repetitive, well-lit industrial tasks are close to break-even now, sometimes already past it. Anything resembling a general household helper is nowhere close, and no amount of price cutting on the hardware fixes a robot that cannot reliably pick up a glass. I think the industrial case closes faster than most bank forecasts assume. The home case takes longer than any of them are willing to say out loud.
- Morgan Stanley's 2024 estimate put humanoid robots at $200,000 in high-income countries, but the bank quietly revised its own 2050 forecast upward, to $75,000, in a January 2026 report.
- The Unitree R1 already sells for $5,900, undercutting the original 2050 forecast for cheap-supply-chain markets by nearly a decade, though it cannot yet do useful household chores.
- Hourly cost, not purchase price, is the number that determines whether a robot is actually cheaper than a human worker, and today that only clearly favors the robot in narrow industrial tasks.
- Industry experts disagree sharply on timeline: one sees near-term adoption driven by labor economics, another points to China's manufacturing scale, and a third puts general home-use robots at least a decade away due to unsolved data and reliability gaps.
FAQ
How much does a humanoid robot cost in 2026?
Prices range from about $5,900 for an entry-level developer model like the Unitree R1 up to $145,000 or more for industrial-grade units like UBTech's U1. Most business-ready models cluster between $50,000 and $100,000.
Is a humanoid robot cheaper than hiring a worker?
On an hourly basis, some low-cost humanoids already come in under $10 an hour once amortized, compared to $22 to $45 an hour for a fully loaded human worker in the US. That gap narrows or disappears once you account for the robot doing less work per hour and needing more downtime.
When will humanoid robots be affordable for home use?
Hardware prices are already low enough for early adopters, but reliability is the real barrier. Industry estimates for general-purpose home robots, capable of handling everyday chores without constant supervision, range from roughly 2030 to over a decade out, depending on who you ask.
What is the cheapest humanoid robot you can currently buy?
The Unitree R1, starting at $5,900, is currently the cheapest full-size humanoid robot with public pricing. Cheaper compact models exist, but the R1 is the lowest-priced bipedal humanoid sold at scale as of mid-2026.
Conclusion
The $200,000 number that keeps getting repeated is real, but it is not the number that matters for deciding whether a humanoid robot is worth buying. The hourly rate is. On that measure, the answer already depends entirely on the job. For a warehouse task running two shifts a day, some robots already pencil out. For anything happening inside an actual home, the honest answer from most of the people building these machines is: not yet, and probably not for a while.
0 Comments