Tesla Optimus Robot

How much does a Tesla Optimus cost?

There is no official price. Tesla has never published one, and no leaked enterprise price sheet exists. Every figure you have seen is either a Musk aspiration or somebody's estimate.

The Musk numbers, dated: at Tesla's 2022 AI Day he said Optimus would cost "less than a car." After the We, Robot event in October 2024 he gave the figure that has stuck to the robot ever since - $20,000 to $30,000, explicitly contingent on production running into the millions of units per year. He reaffirmed that range publicly in January 2026. Then it went quiet. It appears in no 2026 earnings call, no shareholder deck, and it was not given to JPMorgan analysts during their Fremont site visit around 20 August 2026. Tesla has not retracted $20,000-$30,000; it has simply stopped saying it.

For a defensible estimate you have to look at what a humanoid actually costs to build. Morgan Stanley's Robot Almanac, published 7 December 2025, put a 2025 humanoid bill of materials at roughly $131,800 for non-China manufacture and about $45,500 for China-built. Bank of America's humanoid research note of 12 March 2026 put China-built BOM around $35,000 in 2025, falling below $17,000 by 2030. Optimus is a US-built robot with a US-built cost structure and, so far, no volume at all.

Our estimate, and we are labelling it as an estimate: a first external Optimus realistically prices between $50,000 and $100,000. The basis is the non-China BOM range above, plus the fact that Musk's $20,000-$30,000 assumes a scale Tesla has not achieved by a factor of thousands. As of 30 August 2026, treat $20,000 as a target, not a price.

Can you buy a Tesla Optimus in 2026?

No. Not through Tesla, not through a dealer, not through a waitlist. There is no order page, no deposit, no configurator and no commercial inquiry channel of any kind.

The clearest dated statement on this came from Tesla itself, relayed by JPMorgan analysts after a Fremont site visit around 20 August 2026: external commercial sales begin "as early as H2 2027." Note the hedge in "as early as." The Q2 2026 shareholder update of 22 July 2026 was even blunter - the first units off the line are going to data collection, not customers.

A useful outside read on the odds: a crypto prediction market was pricing the probability of an Optimus commercial launch during 2026 at around 3% as of late August 2026, reported by Benzinga on 28 August. That's roughly the market's confidence, and it matches the evidence.

One warning worth stating plainly. Because search demand for Optimus is enormous and supply is zero, a large number of sites publish confident spec tables, "pre-order" pages and delivery dates for a robot that Tesla has never shown. Anybody taking money for a Tesla Optimus is not selling you a Tesla Optimus. There is no authorised reseller because there is no product to resell.

Why has the Gen 3 Optimus reveal kept slipping?

This is the single most informative thing about the programme, and most coverage skips it.

The current publicly shown generation is still Gen 2, unveiled in December 2023. Gen 3 - variously called V3 - has never been publicly unveiled. Not at an event, not in a spec sheet, not in a photo release. As of 30 August 2026 the general public has still not seen it.

The slip chain, with dates: the original target was Q1 2026. On 31 March 2026 Musk said Optimus 3 was "walking around" but not ready, and pushed the reveal to "probably middle of this year," citing competitors analysing his demos frame by frame. Mid-year passed. It moved to late July or early August 2026. That passed too.

Meanwhile the manufacturing side did move. Model S and Model X production ended in early May 2026 and the Fremont line was torn down in roughly 46 days. The 22 July 2026 shareholder update confirmed Tesla was "installing the first-generation lines for Optimus" with production anticipated later in the year. Several outlets reported at the end of August 2026 that production had actually begun at Fremont, but none of them cited a Tesla announcement, a date or a unit count - the quotes were recycled from the July call. Tesla has not formally announced a production start.

What we can say with confidence is that Tesla is building a line for a robot it has not shown, and that the reveal has slipped three times. A fourth slip past 31 December 2026 would tell you more than any spec sheet.

How many Optimus robots has Tesla actually built?

Tesla has never published a unit count. That silence is itself a data point, because the targets were extremely public.

At the June 2024 shareholder meeting Musk said Tesla would have "a thousand to a couple thousand robots working in its factories" during 2025. On the January 2025 earnings call he raised it to "roughly 10,000 Optimus robots" built in 2025, and said he was confident several thousand of them would be doing useful things by the end of the year. An internal target of 5,000 units for 2025 was widely reported. A 50,000-unit figure for 2026 circulated and was later abandoned, with Musk calling 2026 output impossible to predict.

The best independent anchor for what actually happened comes from CNN's 29 July 2026 reporting on the US import restrictions: of roughly 15,000 humanoid robots shipped worldwide in 2025, Unitree and AGIBOT each shipped more than 5,000, while Tesla and Figure AI each shipped no more than a few hundred. The research firm Omdia put Tesla below 500 units for 2025.

Against a 10,000-unit target that is a miss of well over 90%. Estimates of the total Optimus population in mid-2026 - somewhere around a thousand units across Fremont and Giga Texas - circulate widely but trace back to a single research outfit rather than to Tesla, so treat them loosely.

External sales to date: zero. That one is not an estimate.

Is the Tesla Optimus autonomous, or is a human driving it?

Both, depending on which part of the robot you are watching, and the distinction is where most of the public confusion lives.

Locomotion is autonomous. The walking, the balance recovery, the gait over uneven ground - that is neural-network driven, inherited from the same end-to-end architecture Tesla uses for driving. Nobody is puppeteering the legs.

Manipulation is where the humans come in. The January 2024 shirt-folding video that went around the world was teleoperated, and Musk said so himself at the time, noting Optimus could not yet do it autonomously. A telepresence glove is visible in the frame. At the We, Robot event on 10 October 2024 the robots that poured drinks, played rock-paper-scissors and held conversations with attendees were human-assisted; this was established by attendees on the night rather than disclosed by Tesla on stage. A 2025 hand-dexterity demo was confirmed teleoperated by Tesla.

The Tesla Diner in Hollywood, which opened 21 July 2025, has a single Optimus serving popcorn. Staff confirmed a human operator, positioned within about nine metres. On 23 July 2025 the operator lost connection and the robot stopped working mid-service, on camera.

This isn't really a scandal so much as a category error that Tesla has been happy to leave uncorrected. If you see an Optimus doing something dexterous and impressive, the default assumption should be that a person is doing it.

Can a Tesla robot clean your house or cook?

This question appears in the People Also Ask box on virtually every Optimus search, so it deserves a direct answer: no, and not soon.

Household work is the hardest category in robotics, not the easiest. A kitchen is unstructured, cluttered, full of deformable objects and unforgiving of a 57 kg machine getting it wrong. Tesla has never demonstrated an Optimus completing a household task autonomously. The closest thing to a domestic demo was the folded shirt in January 2024, and that was hand-driven.

Tesla's own roadmap does not point at your house first either. The 22 July 2026 shareholder update has first production units doing data collection. Tesla's briefing to JPMorgan in August 2026 pointed at external commercial sales - industrial buyers - from the second half of 2027 at the earliest. A consumer household product sits somewhere beyond that, with no date attached by anyone at Tesla.

If you want a robot in your home this decade, the honest comparison set is 1X NEO, which does take consumer orders and openly relies on remote human operators for chores, and the growing category of cheaper single-purpose machines. Optimus is not competing in that race yet, whatever the rendering on your feed suggests.

Is there an Optimus SDK or developer program?

No. There is no Optimus SDK, no API, no developer documentation, no simulator release, no accessory ecosystem and no third-party module program. We checked, and the absence is complete.

Tesla does run a developer program - the Fleet API, launched in October 2023 - but that is for vehicles and has nothing to do with the robot. No public interface to Optimus exists in any form.

You will find claims of an "Enterprise SDK" arriving in Q4 2026 and a consumer module marketplace in 2027. Those originate entirely from AI-generated content sites and have no Tesla source behind them. They contradict Tesla's own 22 July 2026 filing, which describes lines still being installed for a robot with no announced customers.

For developers this matters more than the price. Even if Optimus ships in 2027, nothing published so far suggests Tesla intends to let anyone else write code for it. The Optimus stack is being built as an extension of Tesla's own vertically integrated AI programme - the same silicon, the same training pipeline, the same closed model. If you need a humanoid you can actually program this year, that market belongs to Unitree, and the entry point is the G1 EDU.

What would it actually cost to own and run one?

Nothing can be said here honestly, and the reason for that is instructive.

Nobody outside Tesla owns an Optimus. Tesla has published no battery capacity officially confirmed for the shipping configuration, no runtime per charge, no charging specification, no service plan, no warranty terms, no spare-parts pricing and no consumables cost. There is no dealer network, no repair channel and no published maintenance interval.

The specification figures that circulate - a 2.3 kWh pack at roughly 52 V, about 5 hours of runtime, a top speed near 2.2 m/s - are drawn from Gen 2 teardowns, presentations and unofficial estimates, and Gen 3 has not been shown at all. Applying them to an ownership model would be arithmetic on top of guesswork.

So the correct entry in a total-cost-of-ownership table for Optimus, as of 30 August 2026, is "not applicable." Any site presenting you with a five-year running cost for a Tesla Optimus has invented every number in it. That is not a rhetorical flourish - there is no source material from which those numbers could be derived.

For comparison, the reason we can publish real running costs for the Unitree G1 is that batteries, chargers, hands, LiDAR modules and even a repair deposit all have live listed prices. Optimus has none.

What are the security, supply and regulatory exposures?

Three worth knowing about, and one of them has a hard deadline.

Rare earths. China imposed export controls on seven rare-earth elements and magnets on 4 April 2025. On Tesla's Q1 2025 earnings call, 22 April 2025, Musk said Optimus production was affected and that Tesla was seeking an export licence, describing Beijing's condition as assurance the components would not go into weapons systems. Tesla has never publicly confirmed the licence was granted. Two of the toughest Chinese measures are paused only until 10 November 2026. Optimus needs neodymium and dysprosium in every actuator; this is the most concrete near-term risk to the programme.

Trade policy, which cuts Tesla's way - though not for the reason most coverage gives. FCC Public Notice DA 26-786, released 28 July 2026, added foreign-produced advanced robotic devices to the Covered List. The text is country-neutral rather than China-specific: 'foreign-produced' means a device fails the Buy American domestic-end-product test, which requires US manufacture plus more than 65% domestic component cost, so UK, EU, Japanese, Korean and Canadian robots are caught alongside Chinese ones, and US assembly of imported parts does not clear the bar. The mechanism is also narrower than the headlines suggest: it blocks new FCC equipment authorisation, which then blocks import, marketing and sale. It is not a customs seizure order and not a recall. Models authorised before 28 July 2026 are grandfathered - import, sale, ownership and resale all continue - and a waiver lets them keep receiving firmware and security updates until at least 1 January 2029. Optimus is designed and built in the United States, so it is unaffected, and Tesla gains a protected runway into a market it does not yet sell into.

Safety regulation, which is a vacuum. No federal humanoid-robot safety standard exists in the US. ISO 25785, the humanoid-specific standard, is still in development. OSHA's General Duty Clause technically applies but there is no accepted compliance methodology and no incident dataset. Tesla does internal testing; no independent third-party safety validation of Optimus has been reported. One correction while we are here: the widely shared "Tesla robot attacks engineer" story refers to a Kuka industrial arm at Giga Texas in November 2021, not to Optimus.

Where are Optimus robots actually working right now?

Three named sites — and that is the complete verified list as of 30 August 2026.

Tesla Fremont Factory, Fremont, California. R&D units plus the first Optimus assembly line, built on the floor the Model S and Model X vacated in May 2026.

Gigafactory Texas, Austin. R&D and data-collection units, with a dedicated Optimus factory under construction on the North Campus - steel was visible in drone footage on 27 May 2026. Higher-volume production there is expected around summer 2027.

Tesla Diner, Hollywood, Los Angeles. One teleoperated unit serving popcorn since 21 July 2025. This is the only place a member of the public can reliably see an Optimus in person.

What is not on the list matters more: no customer sites, no partner factories, no pilot deployments, no third-party installations anywhere in the world. Compare that with Figure, which has BMW's Spartanburg plant and a signed agreement with Catalyst Brands, or with Unitree, whose robots sit in hundreds of university labs. Optimus has the largest search demand in the category and the smallest real-world footprint of any robot on this list.

Tesla Optimus vs Figure 03 vs 1X NEO

These three get compared constantly, and they are at genuinely different stages.

On ordering, 1X leads and nobody is second. NEO has an open consumer preorder, a published price - $20,000 outright or $499 a month - a $200 deposit and reportedly more than 10,000 orders in its first week. It is also the only one of the three that has publicly promised delivery and not yet delivered.

On demonstrated autonomy, Figure leads. In May 2026 Figure ran a package-sorting livestream that its CEO said involved no teleoperation whatsoever, reported at around 101,000 packages over 81 continuous hours. That is a controlled demo in Figure's own facility, and observers noted barcode-orientation errors, but it is still the strongest public autonomy evidence any Western humanoid has produced.

On deployment, Figure also leads. BMW's own press release of 25 June 2026 places Figure 03 in Assembly Hall 52 at Plant Spartanburg on logistics sequencing, described as a production use case. Tesla has no external deployment at all.

On manufacturing, the picture is more even than the share prices suggest. Figure claims one robot per hour off its BotQ line as of April 2026; Tesla is installing its first line. On 2025 shipments, both were in the low hundreds and both were roughly ten times behind Unitree.

Where Optimus wins is capital, silicon and vertical integration. Tesla designs its own AI chips, runs its own training clusters and owns its factories. That is a real long-run advantage. It is just not a 2026 advantage.

Who should watch Optimus, and who should look elsewhere?

Watch it if you are an investor, an industry analyst, a manufacturing strategist, or a competitor. The programme is genuinely consequential - Tesla is the only Western company with the silicon, the capital and the factory footprint to attempt humanoids at automotive volume, and the AI5 chip that taped out on 15 April 2026 is slated to reach Optimus early.

Look elsewhere if you actually need a robot. If you are a university lab, a research group or an AI team that needs a bipedal platform this year, buy a Unitree G1 EDU or an R1 EDU and start work; they cost real money and they exist. If you want a humanoid in a home, 1X NEO is the only one taking consumer orders, with all the caveats that come with it. If you want an expressive interaction robot for a museum or an event, Ameca has been shipping since 2021.

And if you are a journalist, a procurement officer or anyone writing a plan around Optimus availability, anchor on two dates rather than on the narrative. Tesla told JPMorgan external sales start no earlier than H2 2027. Tesla's CEO said in January 2026 that the robot is not doing material work in Tesla's own factories. Everything else is forecast.