Signals Inbox·July 15, 2026·Humanoid Robotics

When will Tesla Optimus be ready?

Tesla Optimus should become useful inside Tesla factories around 2028, reach selected outside companies around 2029 to 2031, and arrive as a genuinely useful household robot closer to 2034 or 2035.

Q1What has Tesla Optimus actually proved so far?

Tesla Optimus has proved that Tesla is willing to spend real money, factory space and engineering capacity on humanoid robots. It has not yet proved that Optimus can work independently enough to justify that investment.

The programme has clearly accelerated over the past year. In its January 2026 financial update, Tesla said it would ramp six new production lines across vehicles, robots, batteries and energy storage. Three months later, its Q1 shareholder deck described Optimus as progressing toward mass production and confirmed that dedicated robot factories were under construction in California and Texas.

The physical commitment became much harder to dismiss in May 2026, when Tesla stopped building the Model S and Model X and began converting their Fremont production area for Optimus. Footage shared in early July showed the old automobile line stripped out and replaced by new robot-production equipment.

That conversion tells us more than another Optimus video. Model S and Model X volumes had become small compared with Tesla’s mass-market cars, but they were still established products generating revenue. Tesla has now given their factory space to a robot that currently generates none.

At the same time, the operational evidence remains surprisingly thin. During Tesla’s January 2026 earnings discussion, Elon Musk acknowledged that Optimus robots were not yet doing useful factory work. Tesla has shown robots sorting battery cells, transporting objects and carrying out basic manipulation, although it has never published the figures needed to judge those deployments properly.

We still do not know how many tasks an Optimus completes per hour, how often a human takes over, how long the robot runs between failures or whether Tesla saves any money by using one.

The gap between manufacturing commitment and operating proof defines where Optimus stands today. Tesla has moved well beyond a research project, yet it has not crossed into proven industrial automation.

What Tesla Optimus has proved so far

Q2Are Tesla Optimus robots already working in Tesla factories?

Tesla Optimus robots are already being tested inside Tesla factories, although the available evidence does not support calling them a productive workforce today.

Figure said its Figure 02 robots worked ten-hour weekday shifts at BMW’s Spartanburg factory. Across the deployment, the company reported more than 1,250 operating hours, 90,000 sheet-metal parts loaded and involvement in the production of more than 30,000 BMW X3 vehicles. Figure has since retired that robot generation and moved BMW toward Figure 03.

Those numbers come from Figure itself and should not be treated like an independent audit. Still, they tell us considerably more than a polished demonstration. We can see the length of the deployment, the approximate workload and the connection to an actual production line.

Agility Robotics has taken a similar approach with Digit. Its flagship GXO warehouse deployment has moved more than 100,000 totes under a multiyear commercial agreement. Agility has also spoken publicly about passing site-level electrical and safety inspections, which is the kind of boring work a robot company must complete before customers can insure and deploy the machine.

Tesla currently gives us none of those operating metrics for Optimus. We have visual proof of robots inside factories, statements about internal deployment and a large manufacturing plan. We do not have a verified task count, operating-hour total or customer-style case study.

That is why we describe today’s Optimus activity as factory testing and data collection. Some robots may already contribute small amounts of useful labour, but Tesla has not demonstrated that they can hold a job in the way Figure has begun to demonstrate at BMW.

Our estimate is that Tesla has a 75% chance of showing repeatable and genuinely useful factory work during 2027 or 2028. The company already controls the building, the production process and the robot, which allows it to simplify early tasks until Optimus becomes reliable.

Q3When will Optimus do real work across Tesla factories?

Tesla Optimus could begin performing economically useful work across several Tesla facilities in 2028 or 2029.

The strongest argument for that timeline comes from Tesla’s ability to build the workplace around the robot. An outside humanoid company has to walk into an existing customer factory and adapt to whatever it finds. Tesla can change the robot, the workstation, the container, the route and the production schedule together.

A difficult manipulation problem can sometimes be reduced by altering how a component arrives. Tesla could add alignment guides, standardise packaging or reserve part of a production area for robots. This would still represent real automation, even though the environment had been made easier.

Tesla also has an enormous internal test market. Its factories contain repetitive material-handling, machine-tending and parts-delivery jobs that fall between traditional automation and human labour. These tasks do not require Optimus to become a universal robot. They require it to repeat a limited set of movements without regularly stopping the line.

Competitor deployments tell us what the first useful Optimus jobs will probably look like. Figure’s BMW work centred on moving and positioning sheet-metal components. Agility’s Digit moves standardised containers between warehouse systems. Apptronik has focused its Mercedes-Benz pilot on parts handling and logistics.

These are narrow jobs with measurable outcomes. They do not involve robots wandering around factories and deciding what needs to be done.

The main uncertainty comes from Tesla’s production history. Tesla frequently moves from a difficult prototype to impressive scale, but its early ramps tend to involve long periods of redesign. Musk acknowledged this directly in January 2026 when he described initial Optimus production as “agonizingly slow” because nearly every major part and manufacturing process was new.

This is why the one-million-unit Fremont target does not drive our forecast. Tesla says the line is being designed for that annual capacity, but factories are often designed for volumes they reach much later or never reach at all. The useful milestone will be Tesla reporting several hundred robots completing paid human work, rather than installing equipment theoretically capable of producing a million.

Q4When will Tesla Optimus work for companies outside Tesla?

Selected outside companies could receive Optimus robots between 2028 and 2030, probably through partnerships rather than ordinary sales.

Tesla has good reasons to keep Optimus inside its own factories at first. Internal deployments allow engineers to watch every failure, modify the workspace overnight and quietly withdraw a robot that is not performing well. An outside customer will expect service guarantees and will notice immediately when the robot interrupts production.

Commercial deployments also create a long list of responsibilities that Tesla can delay internally. Someone must decide who repairs the robot, how quickly replacement parts arrive, how software updates are managed and who pays when an Optimus damages equipment.

Apptronik announced its first Mercedes-Benz agreement in March 2024. By February 2026, it had raised more than $935 million from investors including Mercedes-Benz, Google, John Deere and Jabil-linked backers. The investor list points rather clearly to where Apptronik expects demand: automobiles, manufacturing, agriculture and telecommunications.

Figure’s BMW programme has already moved into a new robot generation after a ten-month deployment. Agility has commercial agreements in logistics, while Boston Dynamics plans to place the electric Atlas robot at Hyundai’s Georgia factory by 2028.

Together, these programmes suggest that external humanoid pilots will be normal by the end of the decade. Tesla would be entering an established industrial buying process rather than creating the category from zero.

Q5When will companies be able to buy Tesla Optimus normally?

Companies will probably not be able to buy Optimus as a standard commercial product before 2029 to 2031.

The key milestone will be repeat orders. A company accepting three robots for a highly publicised pilot tells us that it wants to explore humanoid robotics. Ordering another 100 after using the first fleet would tell us that the economics work.

That second step remains rare across the entire industry today.

Figure has published impressive BMW activity, but the reported deployment involved a small robot fleet and one highly structured task. Agility has crossed into paid warehouse work, although its clearest high-volume case remains concentrated at a single GXO site. Apptronik has attracted large strategic investors and manufacturing partners, yet its public deployments are still described mainly as pilots.

This gives Tesla a little breathing room. Nobody has yet built a broad market with thousands of paying humanoid customers, so Optimus does not need to catch up overnight. The commercial race remains open.

Companies will judge Optimus through total cost per productive hour. The hardware price will only be one part of it. Buyers will add installation, supervision, downtime, remote assistance, electricity, repairs and software fees.

A cheap robot that needs frequent rescue can cost more than an expensive robot that runs reliably. Human intervention is particularly dangerous for the business model. One remote operator supervising ten robots could make economic sense. One operator constantly controlling one robot simply moves the worker away from the factory floor.

We would consider Optimus ready for normal industrial sales once Tesla can show that early customers reordered it, the robots run through long shifts and one technician can support a meaningful fleet.

Q6When will Tesla Optimus become truly autonomous?

Tesla Optimus could become autonomous enough for fixed factory tasks around 2028 to 2030, while flexible autonomy across several jobs will probably take until 2029 to 2032.

A robot does not need human-level intelligence to become commercially useful. It can deliver value by working within a clearly marked area, handling known objects and stopping safely when something unusual happens.

The harder question is how much human support remains behind the scenes.

The consumer robotics company 1X offers a useful clue. Its NEO home robot can learn and repeat tasks, but early customers may use an “Expert Mode” that lets a remote human guide the machine when autonomy fails. 1X openly includes this model in its product strategy instead of pretending the robot can handle every home alone.

Weave Robotics is taking a similar path with its Isaac home robot. The company says its laundry-folding system can work mostly autonomously, although occasional remote intervention remains necessary. Its average folding time improved from roughly four minutes per garment to two minutes, which shows real progress and also how far the industry remains from normal human speed.

Tesla is likely to use remote assistance as well, even if it describes the system differently. During the April 2026 earnings discussion, Musk said Optimus would eventually use local intelligence for most actions while Grok acted more like a manager that stepped in occasionally.

The word “occasionally” will determine the economics. Assistance once per shift would be impressive. Assistance every few minutes would make Optimus difficult to scale.

Tesla does have unusually strong ingredients for this problem. It operates large AI-training systems, designs inference chips and has years of experience collecting video for vehicle autonomy. The company can also generate robot training data inside its own factories.

Manipulating physical objects remains less forgiving than recognising roads. A car can adjust its path before touching an obstacle. A hand often learns that its grasp was wrong only after an object slips, bends or breaks.

Optimus will probably become good at a small library of factory actions before it becomes good at learning unfamiliar work. The practical breakthrough will arrive when the robot notices a failed grasp, repositions itself and finishes the task without calling a human.

Q7When will Tesla Optimus be safe enough to work beside people?

Optimus could work near trained factory employees under controlled conditions around 2028 to 2030. Safe unsupervised use around children, elderly people and pets will require much more evidence.

Industrial deployments can reduce risk by controlling speed, force, routes and access. A robot may work behind a marked boundary, slow down when someone approaches and stop completely whenever its vision system becomes uncertain.

Agility Robotics has shown how much work sits behind that process. The company has promoted Digit’s successful Nationally Recognized Testing Laboratory field evaluations because customers need electrical and safety approval before deploying robots at operational sites. These evaluations are less exciting than a dexterity video, but they influence whether a warehouse can insure the machine.

Tesla has not yet described an equivalent Optimus certification programme publicly.

The safety challenge becomes harder as the AI improves. Traditional industrial robots are predictable because engineers define their movements. A general robot chooses actions according to perception and learned policies, so its behaviour can vary between situations.

Recent robotics research has focused on separating intelligent decision-making from a simpler certified safety layer. The AI may choose how to complete a job, while a second system limits speed, force and movement whenever a person enters a dangerous area. Optimus will probably need that kind of architecture before regulators and customers accept it.

Q8When can you buy Tesla Optimus for your home in the U.S.?

A limited Tesla Optimus home release in the United States looks most plausible around 2031 to 2033, with a 45% probability of happening during that period.

This estimate refers to an early product, rather than the polished robot assistant shown in science-fiction films. The first owners may accept strict usage rules, regular software updates and remote support when Optimus encounters a task it cannot complete.

The surprising clue comes from Tesla’s smaller competitors. Consumer humanoids are already reaching the order stage.

1X currently offers NEO for $20,000 or $499 per month and says U.S. deliveries begin in 2026. The company states clearly that early robots arrive with basic autonomy and improve over time. It had reportedly collected around 10,000 preorders by July 2026.

Weave Robotics plans to begin shipping its $7,999 Isaac 1 in California during autumn 2026. Its initial focus is deliberately narrow: folding laundry, tidying and moving common household items. The company wants to place 1,000 units in homes by 2027.

These products do not show that useful general-purpose home robots are already solved. They show that customers may accept an incomplete robot when the price, task and support model are clearly explained.

Tesla could follow the same route earlier than our estimate if it wanted to release Optimus as a premium beta product. We think it will wait because Tesla’s largest economic opportunity currently sits in industrial labour. A factory robot can repeat one valuable task all day, while a home robot must understand hundreds of objects and routines to justify its price.

Q9When will Tesla Optimus become genuinely useful at home?

Tesla Optimus is unlikely to become a broadly useful household robot before 2033 to 2036.

The first consumer release may be able to carry objects, bring known items, monitor rooms and perform a handful of prepared chores. Real household usefulness starts when the same robot handles several jobs across a changing environment.

Laundry reveals why this takes time. A robot may learn to fold one shirt on an empty table, but a normal laundry load contains towels, socks, dresses, tangled sleeves and fabrics that behave differently every time.

Weave Robotics’ current two-minute folding time per garment gives us a useful order of magnitude. A load containing 25 items would require roughly 50 minutes if everything went correctly. That could still save the owner time because the robot works alone, although repeated assistance requests would quickly reduce the value.

The kitchen is harder. Cooking combines liquids, heat, knives, food safety, deformable ingredients and many actions that cannot fail safely. Tesla will probably avoid these tasks during the early years.

Optimus may eventually gain better hardware than these competitors. Tesla still has to combine dexterity, mobility, autonomy, battery life and repairability in one affordable machine.

Q10When will Tesla Optimus be sold around the world?

Industrial Optimus deployments could spread internationally around 2030, while broad consumer availability looks closer to 2034 to 2038.

Tesla already has factories, engineering teams and service infrastructure in North America, Europe and China. That gives it several natural locations for controlled industrial deployments. Shanghai is particularly important because Tesla executives have highlighted the factory’s supplier network and manufacturing expertise as potential assets for scaling Optimus.

The Shanghai operation delivered roughly 851,000 vehicles in 2025, which puts its manufacturing base in a completely different category from any current humanoid startup. If Tesla needs millions of actuators, gears, sensors and cast parts, its Chinese supply chain could become a major advantage.

Consumer expansion will move more slowly. Tesla would need local safety approvals, language support, spare-parts stocks, trained repair teams and insurance arrangements. A robot operating inside private homes also creates tougher privacy questions than a car driving on public roads.

The physical world itself changes between countries. Appliances, doors, kitchens, power outlets and household layouts are different. Optimus could understand a new spoken language fairly quickly while still struggling with unfamiliar objects.

Tesla’s automotive expansion offers a warning here. Even with established cars and factories, software functions such as supervised driving have launched unevenly because regulations and local validation differ. Optimus adds machinery rules, workplace law and home liability on top of those difficulties.

Tesla may announce availability in many countries relatively early. Widespread delivery and proper service will arrive later.

Q11When will Tesla Optimus cost $20,000?

An Optimus purchase price close to $20,000 looks unlikely before 2035, although Tesla could advertise a lower upfront figure through leasing or subscriptions.

Musk has repeatedly argued that Optimus should eventually cost less than a car because it uses fewer raw materials. That logic becomes plausible at very high volume. A humanoid weighs far less than a vehicle and needs a much smaller battery.

Early robot costs come from complexity rather than steel and aluminium. Custom actuators, precision hands, sensors, computing equipment, assembly labour and low manufacturing yields can make a relatively light machine extremely expensive.

Competitor pricing helps establish today’s range. 1X charges $20,000 for NEO but also offers a $499 monthly subscription. Weave Robotics prices Isaac 1 at $7,999 or $449 per month. Agility’s industrial Digit has historically been estimated far higher because it targets professional customers and includes industrial-grade hardware and support.

The cheaper consumer robots are also more constrained. Isaac uses wheels instead of legs, focuses on selected household chores and occasionally relies on remote help. That design is sensible. Removing bipedal walking lowers cost and reduces the chance of falling.

Tesla wants Optimus to combine legs, dexterous hands, onboard intelligence and useful physical strength. Reaching $20,000 with all four will require genuinely large production volumes and a supply chain that no humanoid company has built yet.

The advertised purchase price may also give an incomplete picture. Tesla could charge separately for advanced autonomy, remote assistance, repairs, connectivity or task packages. A $20,000 robot with a $500 monthly service plan costs $50,000 over five years before maintenance.

We would track the total five-year customer cost rather than the headline price.

Q12Could Tesla Optimus arrive earlier than we expect?

Tesla could beat our timeline if the new Fremont line produces useful operating data quickly rather than simply producing hardware.

The company has several advantages no humanoid startup can easily copy. Tesla ended 2025 with $44.1 billion in cash and investments. It plans to spend around $25 billion in capital expenditure during 2026 across AI, vehicles, batteries, robots and manufacturing infrastructure. Apptronik’s record $935 million funding round is enormous for robotics, yet it represents less than 4% of Tesla’s planned annual investment.

Tesla can also manufacture many robot components internally, train AI models on its own computing infrastructure and test Optimus without negotiating access to another company’s facilities.

Fremont may reveal Tesla’s speed quite soon. The Model S and Model X line was reportedly dismantled in less than seven weeks. Low-volume Optimus production is expected to begin around late July or August 2026, although the exact ramp remains uncertain.

The strongest bullish case would involve Tesla producing hundreds of Gen 3 robots during the second half of 2026, running them continuously through 2027 and publishing clear improvements in human interventions.

Q13So when will Tesla Optimus actually be ready?

Tesla Optimus should become ready for useful factory work around 2028, for selected commercial customers around 2029 to 2031, and for normal households around 2034 or 2035.

That conclusion comes from combining three very different sets of evidence.

Tesla’s manufacturing commitment is now real. The company has removed two car models from Fremont, installed a dedicated robot line, started building a much larger Texas operation and allocated substantial capital to physical AI.

Tesla’s operating proof still trails that commitment. As of July 2026, the company has not published a meaningful Optimus workload, autonomous-hours figure, intervention rate or customer deployment.

Competitors show that industrial humanoids can already complete useful work, but they also reveal how narrow the current jobs remain. Figure has logged more than 1,250 hours at BMW. Agility has moved over 100,000 totes at GXO. Apptronik has attracted major industrial investors and pilots. None has yet created a large, repeatable humanoid market.

Our Tesla Optimus readiness timeline

The next twelve to eighteen months should tell us much more than the previous four years. Fremont is moving from robot prototypes toward a real production line, which means Tesla will soon have enough hardware to expose the weaknesses that small fleets can hide.

The decisive evidence will be rather mundane: hours worked, objects moved, interventions avoided, repairs completed and money saved.

Today, Tesla knows how to build an increasingly capable humanoid. By 2028, we should know whether it has built one that can hold a job.