AI Agent UpdateToyota formalises $6.4 billion-a-year factory-automation plan covering 400,000 robots
Toyota is formalising a roughly $6.4 billion-a-year factory-automation plan covering about 400,000 robots from 2028 — robots broadly, not humanoids only — with its own factories first.
The News
Toyota is formalising a factory-automation plan worth roughly $6.4 billion a year, covering about 400,000 robots from 2028, the Financial Times reported on October 25. The plan steps up the commercialisation of two decades of robotics research inside Toyota, turning laboratory work into factory equipment at industrial scale. About 250,000 of the robots are for Toyota group companies and suppliers, with the remainder presumably for wider sale — though the FT summaries do not break that remainder down. One clarification is essential: the 400,000 figure is robots broadly, not a humanoid-only order, and reporting it as humanoids would be wrong.
The organisational change is as important as the money. Toyota is creating a dedicated robotics team under its chief production officer, the executive responsible for how its factories run, and it will prioritise its own factories first. That is a sensible order: Toyota can test robots where it controls the environment, fix problems quickly, and only then sell proven machines to others. It also signals that Toyota sees robotics not as a side project but as a core part of how it will manufacture cars — and potentially other products — in the next decade.
The humanoid strand of the story runs through Toyota Research Institute, known as TRI. Its spinout Walden Robotics raised $300 million in July at a $1.1 billion valuation and already runs wheeled factory humanoids at a Toyota plant in North America — wheeled, not walking, which is an easier and more stable design for factory floors. TRI's foundation model, a large AI model that can be adapted to different tasks, is also used in Boston Dynamics' Atlas humanoid and in Toyota's own 50-kilogram Eley humanoid, which learns factory tasks by observation. One caution: the FT piece itself is paywalled, so the figures here are as reported by FT summaries, not taken from the full FT text, and they should be read in that light.
400,000 robots — broadly, not humanoids only — from a carmaker turning two decades of research into factory floors.
Why It Matters
The first reason this matters is scale. A plan covering 400,000 robots is not a pilot programme or a publicity stunt; it is a manufacturing strategy. Even spread over years, that volume would make Toyota one of the largest deployers and sellers of factory robots in the world, and it would give its suppliers — the companies that make parts for Toyota cars — access to automation they could not develop alone. When the biggest carmaker automates, its whole supply chain tends to follow.
The second reason is the humanoid angle, properly understood. Humanoids attract attention because they look like us, but their real factory value is flexibility: a robot shaped like a person can, in principle, use tools, walkways and workstations designed for people, without rebuilding the factory around it. That is why Toyota's approach — wheeled humanoids now, observation-learning models like Eley in development, and a foundation model shared with Boston Dynamics' Atlas — is interesting even though humanoids are only part of the 400,000. If observation learning works reliably, training a robot for a new task becomes as simple as showing it once.
The third reason is competition. Toyota is explicitly positioning itself against Tesla and Chinese humanoid makers, and Tesla appears here as a public read-through: a humanoid competitor, not a partner in this plan. Tesla's Optimus and China's fast-moving humanoid startups have dominated headlines, while Toyota has worked quietly for twenty years. A $6.4 billion-a-year plan is Toyota's answer — less flashy, more industrial, and backed by factories that actually need the machines.
Analyst's Take
This is a Bullish story for Toyota and for factory robotics generally, because it converts research into a dated, budgeted plan. Many carmakers talk about robots; few commit roughly $6.4 billion a year and a dedicated team under the chief production officer. Starting with its own factories is the right discipline: it lets Toyota prove the machines work where downtime is costly and quality standards are unforgiving, before asking anyone else to buy them. If the plan delivers, Toyota gains both cheaper, more flexible factories and a new product line to sell to other manufacturers.
The honest limits should stay attached to the enthusiasm. First, the 400,000 figure is easily misread, and any version of this story that calls them all humanoids is simply wrong. Second, the FT article is paywalled, so these numbers come via summaries rather than the full text, and they deserve that caveat. Third, Tesla is named here only as a public read-through — a humanoid competitor whose Optimus programme is the obvious comparison — not as a participant in Toyota's plan. With those limits in place, the core signal stands: the world's biggest carmaker is betting billions that robots, humanoid and otherwise, belong on the factory floor. Readers' poll: will Toyota's own factories prove humanoids work at scale before 2030?
By The Numbers
FAQ
Are all 400,000 of these robots humanoids?
No — and that distinction matters. The roughly 400,000 figure covers robots broadly: factory arms, wheeled machines, automated vehicles and other automation, as well as humanoids. About 250,000 are for Toyota group companies and suppliers. Presenting all 400,000 as humanoids would badly overstate the humanoid part of the plan.
What is Walden Robotics?
Walden Robotics is a spinout from Toyota Research Institute, Toyota's research arm — a separate company created from research developed inside Toyota. It raised $300 million in July at a $1.1 billion valuation, and it already runs wheeled factory humanoids at a Toyota plant in North America. A valuation is what investors think a company is worth, based on the price they paid for a stake in it.
What does 'learning by observation' mean for a robot?
It means the robot learns a task by watching a human do it, rather than being programmed step by step for every movement. Toyota's 50-kilogram Eley humanoid is reported to learn factory tasks this way, using a foundation model — a large AI model trained on broad data that can then be adapted to specific tasks — from Toyota Research Institute.
Source: Financial Times via TradersUnion