Toyota is stepping into the humanoid robot race, and the Financial Times framed the story on October 25, 2026 as Toyota taking on Tesla and China. The headline sets the world’s best-known carmaker against two very different rivals. One is a US electric-vehicle maker that has tied part of its pitch to Optimus. The other is a Chinese supplier base that, by several reports, ships most of the world’s humanoid machines.
The headline does not settle how much hardware Toyota will actually build. Toyota humanoid robots have a documented history that starts with the T-HR3 reveal in 2017 and runs through a research tie-up with Boston Dynamics in 2025. Yet the numbers circulating online this month, including a claimed 150,000 robots for Toyota’s own plants, are not backed by the Toyota sources we reviewed. Tesla’s own timeline is also under pressure, as we covered when we tested Tesla’s 20,000-a-week Optimus claim.
This analysis separates what is confirmed from what is not. It then compares Toyota’s position against Tesla Optimus, Chinese humanoid robots, and the Boston Dynamics Atlas program, and closes with five predictions and a FAQ. Every figure below carries a named source. Where the record is thin, we say so rather than fill the gap.
Don't miss new tech stories on Google
Add Tech Insider once in the Google app and our stories appear in your news suggestions.
What the Financial Times Reported on October 25, 2026
The Financial Times published the piece under the headline “Toyota takes on Tesla and China in humanoid robot race” on October 25, 2026. Other outlets, including driving.ca, DailyVoice.com and Tekedia, picked up the story within hours. An Arabic-language CNBC Arabia version followed on October 26. The core message across all of them is the same: Toyota wants a meaningful share of a market where it has so far been a quiet player.
The FT article sits behind a paywall, and we are not going to paraphrase details we cannot read in full. What we can say is limited to the headline and the date. We found no on-record quote from a named Toyota executive or engineer tied specifically to the FT story. That absence matters, because several secondary write-ups attach figures and product names to Toyota without showing where those came from.
Readers should treat the framing as a competitive narrative, not a product launch notice. A story about a race implies Toyota is catching up, and that is a fair reading of the market. Tesla has made Optimus a centerpiece of its investor pitch, and Chinese makers have been shipping units for more than a year. Toyota’s challenge is to turn decades of factory expertise into a robot that works on a line, at a cost its plants can justify.
What Toyota Has Actually Confirmed About Humanoid Robots
The confirmed record is shorter than the headlines suggest. Toyota Motor Corporation, known as Toyota, publicly documented a humanoid robot called T-HR3 in November 2017. It also announced research with Boston Dynamics involving the Atlas humanoid and a Large Behavior Model on August 20, 2025, through Toyota Research Institute (TRI). Beyond those two anchors, specifics on price, volume and launch timing remain undisclosed.
| Claim | Status | Basis |
|---|---|---|
| Toyota’s documented humanoid is T-HR3 | Confirmed | Toyota sources |
| T-HR3 revealed November 21, 2017 | Confirmed | Toyota sources |
| Toyota and Boston Dynamics research on Atlas and an LBM, August 20, 2025 | Confirmed | TRI announcement |
| FT headline published October 25, 2026 | Confirmed | Financial Times |
| New Toyota robot officially named ELEY | Unconfirmed | Not in Toyota sources reviewed |
| 150,000 robots in Toyota plants, 250,000 in supplier plants | Unconfirmed as a humanoid plan | Circulating reports only |
| Humanoids already running on production lines | Unconfirmed | No Toyota confirmation found |
| Price, production volume, launch date | Unconfirmed | Nothing official published |
The table shows a pattern. The confirmed items are historical or research-stage. The unconfirmed items are the ones that would make Toyota a volume competitor. Until Toyota publishes figures itself, the responsible position is that the company is a serious research participant and a prospective manufacturer, not yet a shipping rival.
T-HR3: Toyota’s Humanoid Starting Point From 2017
Toyota revealed T-HR3 on November 21, 2017, and described it as the company’s third-generation humanoid robot. Its Partner Robot Division built the machine. The design centered on a remote maneuvering system that mirrored the operator’s movements and returned force feedback between the person and the robot. In plain terms, a human wore a control rig, and the robot copied what the human did while the human felt what the robot touched.
That approach differs from today’s pitch for autonomous humanoids. T-HR3 was a teleoperated platform, closer to an avatar than a worker. Yet it taught Toyota a lesson that now sits at the center of modern robot learning. If a person can guide a machine through a task with precise force feedback, that guidance can become training data. Toyota Research Institute has since built much of its program around exactly that idea.
Nine years after the reveal, T-HR3 remains the only humanoid in the Toyota sources we reviewed that carries an official name, date and design description. That is why we anchor this article on it. Any claim that a different Toyota robot is already in factory service needs its own primary evidence, and so far none has surfaced in the public record.
Toyota Research Institute and the Boston Dynamics Atlas Tie-Up
The most concrete recent signal came on August 20, 2025, when Toyota and Boston Dynamics announced research involving Atlas and a Large Behavior Model. A Large Behavior Model, or LBM, trains a robot on demonstrations so it can learn many tasks from a shared model instead of one hand-coded routine per job. The pairing joins Toyota’s learning research with a machine known for athletic whole-body movement.
TRI researchers have been open about why they favor human-shaped designs. Research scientist Alex Alspach told TRI’s news team that “Current robot manipulation evolved from big, strong industrial robots moving car parts and big tools with their end effectors.” He added that “I think it’s wise to take inspiration from the human form – we are strong enough to perform most everyday tasks with our hands, but when a big, heavy object comes around, we need to get creative with how we wrap our arms around it and position our body to lift it.” Both remarks, from the same TRI story, describe the soft-bodied Punyo research humanoid.
Russ Tedrake, TRI’s senior vice president of robotics research, gave the broader logic in a 2023 TechCrunch interview: “Something we’ve observed is that the world is built for humans.” Factories, doorways, shelves and tools all assume a human body. A robot that shares that shape can, in theory, slot into existing spaces without a plant being rebuilt around it.
The ELEY Name and the 400,000 Figure: What Stays Unconfirmed
Two details dominate social posts about the story, and both need care. The first is a robot name, ELEY. The second is a set of deployment numbers. The Toyota sources we reviewed identify T-HR3 and do not confirm either detail as an official humanoid program. We therefore report them as claims in circulation, not as facts.
The ELEY name
Some reports call Toyota’s new learning robot ELEY. An analysis on the CHINA CASE page on note.com, published September 20, 2026, uses the name and describes a wheeled, two-armed machine that learns by watching people, rather than a bipedal walker. We could not match the name to a Toyota primary source, so we flag it as unconfirmed. Readers should expect the label to either firm up or vanish once Toyota speaks on the record.
The 150,000 and 250,000 figures
Reports have claimed Toyota will deploy 150,000 robots in its own factories and 250,000 more at supplier plants. The same CHINA CASE write-up argues the combined total of roughly 400,000 describes a wider upgrade of factory automation, covering conventional industrial robots and logistics systems as well as humanoid types, and not a count of new humanoids. It also says the figure grew into a “400,000 humanoid robots” headline as it spread across languages. We cannot verify the underlying Toyota presentation, so the safe reading is that no official Toyota humanoid production or deployment plan has been established.
The distinction is not academic. A fleet of 400,000 humanoids would reshape component demand overnight. A fleet of 400,000 mixed robots, where humanoids are a small slice, is a routine automation refresh of the kind large manufacturers run every decade. Investors who price the first scenario while the facts support the second will be disappointed.
Tesla Optimus: The Rival Toyota Is Measured Against
Tesla Optimus is the benchmark the FT headline invokes. Reports describe a gap between ambition and output. Coverage of Tesla’s production progress says the company built only a few hundred Optimus units against a stated target of 5,000. The same reports say Optimus production started on Fremont lines in October 2026, and that mass production at Tesla’s Texas facility is targeted for the third quarter of 2027. Tesla has not disclosed how many units are doing useful factory work.
For Toyota, that gap is an opening. If the best-funded Western rival still counts output in the hundreds, a patient entrant with deep manufacturing discipline is not as far behind as the word “race” implies. Tesla also pulls attention across several hardware programs at once. Our look at Tesla’s three hardware bets in robotaxi, Semi and Roadster shows how crowded that agenda has become.
Capital is the other pressure point. Tesla has been trimming some spending lines even as rivals raise theirs, a theme we traced in Meta’s $34.9B quarter versus Tesla’s $4.73B capex split. A company that rations hardware budgets has to choose between robotaxis, energy storage and humanoids. Toyota, with a long record of funding slow, careful engineering, may be better placed to sustain a multi-year robot program without a sudden budget reversal.
China’s Lead in Humanoid Shipments
The Chinese side of the race is where the data looks strongest. StartUp Fortune and Ground News both reported that Chinese manufacturers accounted for more than 97% of global humanoid shipments in the first half of 2026. Unitree, UBTech and AgiBot were among the leading suppliers, and together with others they helped ship about 19,100 humanoid robots in that period. A figure that lopsided means the question for Toyota is less about Tesla and more about whether it can compete on cost against a mature domestic supply chain in China.
Early commercial traction backs up the volume. The Physical AI Lab’s 2026 deployment tracker reports that UBTech had delivered more than 500 Walker S robots to Chinese automakers and electronics makers by December 2025. It also notes AgiBot held an order for 100 A2-W wheeled humanoids from auto-parts maker Fulin Precision. Those are modest numbers next to a 400,000-unit headline, but they are real orders from real factories, which is more than most Western programs can show.
Chinese makers are also looking abroad. A Worldnews.com item dated October 23, 2026 described Chinese humanoids hunting for work in South Korea as Seoul bets on physical AI. Compute supply is part of the same story, as in our report on 32 Nvidia B300 GPUs spotted at China Telecom. Cheap robots need cheap, abundant training compute behind them.
Competitive Comparison: Toyota, Tesla, China and Boston Dynamics
The table below lines up the four camps using only figures that appear in the sources cited in this article. Where no verified number exists, we write “Not disclosed” rather than estimate.
| Player | Documented platform | Reported scale | Production timing |
|---|---|---|---|
| Toyota | T-HR3 (2017), Atlas research via TRI (2025) | Not disclosed | Not disclosed |
| Tesla | Optimus | A few hundred units vs a 5,000 target (reported) | Fremont start Oct. 2026, Texas mass production targeted Q3 2027 (reported) |
| Unitree, UBTech, AgiBot | Multiple humanoid lines | About 19,100 humanoids shipped, H1 2026 (reported) | Shipping now |
| UBTech | Walker S | 500+ delivered by Dec. 2025 (reported) | Shipping now |
| AgiBot | A2-W wheeled humanoid | Order for 100 units from Fulin Precision (reported) | Order stage |
| Boston Dynamics | Atlas | Not disclosed | Research with Toyota announced Aug. 20, 2025 |
Two lessons come out of the table. First, China is the only camp with shipment data at scale. Second, Toyota’s cell is empty on production, which is exactly why the FT’s framing reads as aspiration. Toyota’s advantage lies elsewhere, in plant know-how, supplier relationships and the willingness to test a robot against a real takt time.
Why a Carmaker’s Factory Floor Is an Advantage
Robot makers need three things to succeed: capable hardware, good training data and a customer willing to run pilots. Startups can build the first. Few can supply the other two at scale. Toyota owns assembly plants across several continents and a supplier network that touches thousands of tasks. Every human demonstration on those lines is a training sample, and every line is a test bed where success is measured in seconds per cycle.
Data collection is the real bottleneck for learning-based robots, and its cost is falling. We saw one side of that trend in our report on Nvidia Flora labeling robot video at $0.72 an hour. Cheaper labeling helps every player, but it helps most the companies that already hold large libraries of workplace video. A carmaker with decades of line footage and a research arm focused on behavior models fits that profile.
The counterweight is cost discipline. A humanoid has to beat a conventional industrial arm or a human worker on total cost per task. Fixed-base arms are cheap, fast and proven. A humanoid wins only where the task is awkward, the layout is fixed by older design, or the work shifts often. Expect Toyota to start with narrow, high-value tasks and widen the scope slowly.
Historical Context: From Industrial Arms to Learning Robots
Toyota’s robotics story predates the humanoid craze. The company built its production system on automation with a human touch, a principle it calls Jidoka. TRI chief Gill Pratt has described the institute’s aim as amplifying human ability rather than replacing it. That philosophy shapes how Toyota talks about robots, and it explains why the company avoids the replace-the-worker language that some rivals use.
Leadership has long pointed past cars. Then-president Akio Toyoda told a CES audience in January 2018, as recorded in Toyota’s global newsroom, “It’s my goal to transition Toyota from an automobile company to a mobility company, and the possibilities of what we can build, in my mind, are endless.” Humanoid work grew from that same stretch. T-HR3 arrived in late 2017, the Atlas research followed in 2025, and the FT headline lands in 2026.
The timeline shows a company that moves in deliberate steps. Eight years separate the T-HR3 reveal from the Atlas announcement. That pace frustrates observers who want a Toyota product tomorrow. It also suggests that any commercial launch, when it comes, will have gone through years of validation. Toyota’s reputation for reliability is its brand asset, and a failed humanoid rollout would put it at risk.
Market Impact for Suppliers, Investors and Component Makers
Even without a confirmed volume plan, the Toyota story moves the conversation. Component makers read each large-OEM commitment as a demand signal for actuators, reducers, sensors and compute modules. Toyota’s supplier network, which includes some of the world’s biggest parts makers, can turn research interest into purchase orders faster than a startup could. If Toyota signals a real program, those suppliers will race to qualify humanoid-grade parts.
Investors should separate three layers. The first is research credibility, where Toyota already scores well. The second is pilot deployment, which no source we reviewed has confirmed. The third is volume manufacturing, where China holds a clear lead on the available data. Pricing a stock on the third layer while only the first is verified invites a correction when details emerge.
The Chinese 97% share also carries trade-policy risk. Western automakers that depend on Chinese-built humanoids may face tariffs, export rules or security reviews. A Toyota-built or Toyota-sourced robot could appeal to buyers who want a non-Chinese supply chain. That is a plausible niche, though no source we found shows Toyota targeting it. More hardware coverage sits in our hardware section.
5 Predictions for the Humanoid Robot Race
These are our analytical judgments, not reported facts. Each one depends on how the unconfirmed items above resolve.
- Toyota stays quiet on price and volume through the first half of 2027. Its history of slow, validated rollouts suggests demos and research updates will come before any commercial figures.
- Chinese makers keep a clear majority of global humanoid shipments through the end of 2026. Their existing orders and supply chains give them momentum that no Western program can erase in a few months.
- Tesla’s Texas target of the third quarter of 2027 becomes the next test. If the Fremont ramp stays in the hundreds, expect questions about that date to grow louder.
- The 400,000 headline gets corrected in more outlets. As readers compare it against Toyota’s own materials, the claim that it describes humanoids alone will fade.
- Behavior-model software becomes the main differentiator. Hardware is converging, so the contest shifts to who trains the best general-purpose models from demonstration data, which favors Toyota’s TRI work.
What to Watch Next
Four signals will tell readers whether Toyota is a real contender. Watch for an on-record statement from a named Toyota executive that confirms a robot name and deployment scope. Watch for any pilot on a Toyota line with measurable cycle-time results. Watch for supplier announcements tied to humanoid-grade components. And watch for fresh shipment tallies from Unitree, UBTech and AgiBot for the second half of 2026.
Toyota Europe’s robotics page is a sensible place to check for official updates, since corporate pages tend to carry confirmed language before aggregators do. Until Toyota speaks, the story stays a research-stage entrant facing a leader that ships and a rival that struggles to scale. That is still a notable position, and it explains why the Financial Times chose the framing it did.
Frequently Asked Questions
Is Toyota really building humanoid robots? Toyota has documented humanoid work for years. It revealed T-HR3 in November 2017 and announced Atlas and Large Behavior Model research with Boston Dynamics on August 20, 2025. It has not published a price, production volume or commercial launch date for a humanoid.
What did the Financial Times report about Toyota? The FT published a piece titled “Toyota takes on Tesla and China in humanoid robot race” on October 25, 2026. Its full text is paywalled, so this article relies only on the headline and date.
Is the Toyota robot called ELEY? Some reports use that name. The Toyota sources we reviewed identify T-HR3 and do not confirm ELEY, so we treat the name as unconfirmed.
Will Toyota deploy 400,000 humanoid robots? No official Toyota humanoid plan with those numbers has been established. An analysis on note.com argues the figure refers to a broader factory-robot upgrade that includes non-humanoid machines.
How does Toyota compare with Tesla Optimus? Reports say Tesla built a few hundred Optimus units against a 5,000 target and aims for Texas mass production in the third quarter of 2027. Toyota has disclosed no production numbers, so a direct volume comparison is not possible yet.
Who leads humanoid shipments today? Chinese manufacturers. StartUp Fortune and Ground News reported they accounted for more than 97% of global humanoid shipments in the first half of 2026, with about 19,100 units shipped.
What is Toyota Research Institute’s role? TRI is Toyota’s research arm for AI and robotics. It leads the Large Behavior Model work, and its researchers study how robots can learn manipulation from human demonstration.
