Tesla Optimus: Programme Profile

The Programme

Tesla was founded in 2003 and makes cars, batteries and solar equipment; Optimus is one programme within it. The robot was announced as “Tesla Bot” at AI Day on 19 August 2021 — an announcement made without a robot, using a dancer in a bodysuit. Elon Musk said at the time that it would become “the biggest product ever, not just Tesla's biggest product ever.”

The underlying argument is Tesla's real one, and it is not silly: a company that already designs its own actuators, battery packs, inference chips and vision neural networks, and that builds several thousand complex machines a day, has more of a humanoid robot's bill of materials in-house than any startup does. Optimus is a bet that manufacturing and autonomy at automotive volume transfer to a different body.

Country and Headquarters

Tesla is a United States company headquartered in Austin, Texas. Optimus itself is being built at the Fremont, California plant, on the line that until May 2026 produced the Model S and Model X. A second line is planned at Gigafactory Texas.

Leadership

Milan Kovac led Optimus from 2022 and is generally credited with the hardware — he resigned in June 2025. He was replaced by Ashok Elluswamy, who runs Tesla's Autopilot and AI teams.

That succession is the clearest statement Tesla has made about its own strategy: the programme moved from a hardware lead to the person responsible for Full Self-Driving. Tesla's claimed advantage is not the body but the vision neural network and the training infrastructure behind it, reused from millions of cars.

Scale and Resources

The funding question that matters for every other company in this section does not apply here. Optimus is financed from the balance sheet of one of the world's most valuable companies, with no round to raise, no valuation to defend and no runway to run out of. That is a structural advantage none of its rivals has.

It cuts the other way too. A programme inside a public company competes for capital, factory space and engineering attention with the products that actually earn the money, and it is accountable to quarterly expectations rather than to a patient private cap table.

Development Directions and Products

• 2021 — announced at AI Day. A concept and a costume.
• 2022 — semi-functional prototypes at the second AI Day in September, one walking on stage. Planned as 173 cm and about 57 kg, carrying 20 kg.
• Gen 2 (December 2023) — the real engineering step. Tesla-designed actuators throughout, a lighter and more anthropomorphic frame, faster walking, and hands with 11 degrees of freedom and tactile fingertips. Shown poaching an egg and, later, doing factory tasks.
• Gen 3 / V3 — still not revealed as of September 2026. Tesla has described it publicly — a redesigned forearm with 22 degrees of freedom per hand driven by around 50 actuators relocated into the forearm, running on Tesla's AI5 chip — but there has been no launch, no published specification and no price. Musk's stated reason for withholding it is that competitors do frame-by-frame analysis of anything Tesla shows.

Treat any V3 figure, including the ones above, as a company statement rather than a specification. That distinction matters more here than on any other page in this section.

Deployment, and the Record on Promises

As of September 2026 there is no documented Optimus deployment outside Tesla's own facilities, and no customer. Everything so far is internal factory testing. Set against the rest of this section, that is the central fact: Figure has BMW and a production figure, Agility has Schaeffler, GXO and Toyota with 65,000 logged operating hours, Boston Dynamics has Hyundai. Tesla has intentions.

The intentions are enormous. Tesla says the Fremont lines are designed for one million robots a year, and that a second-generation line at Gigafactory Texas is being designed for ten million a year from 2027. Those numbers are roughly three orders of magnitude beyond anything anyone has built.

They sit against a documented record of targets not met:

Two further criticisms are on the record. Tesla's 2024 demonstration videos were shown to have relied on teleoperation for some tasks without saying so, which is the difference between a robot doing something and a person doing it through a robot. And Chris Walti, who previously led Optimus, has said publicly that engineering humanoids is “multiple orders of magnitude” harder than self-driving cars — a pointed remark given that the programme's current leader came from self-driving.

None of this means Optimus fails. Musk himself has been unusually candid that this ramp is the hardest Tesla has attempted, with more than 10,000 unique parts, and that output “will move as fast as the least lucky, slowest, dumbest part in the entire 10,000.” Tesla has converted scepticism into volume before, with the Model 3. But a reader comparing this programme with the others in this section should weigh demonstrated deployment against announced capacity, and note that Optimus currently has a great deal of the second and none of the first.