Tesla really wants you to believe the future of the company walks on two legs.

In its latest earnings update, the automaker said the “Gen 3” version of its Optimus humanoid robot — the first one it claims is truly meant for mass production — will be unveiled in the first quarter of 2026, with a dedicated production line planned before the end of the year. The company is openly talking about an eventual capacity of 1 million robots per year, a number that sounds less like an incremental product roadmap and more like a bet on what comes after the electric car.

If you’ve lost track of Optimus, that’s understandable. Tesla first announced the robot back in 2021 with a stunt that involved a human in a spandex suit, and then followed up with a series of carefully edited demo videos and on-stage prototypes that could walk, wave, and fold a shirt — usually very slowly and next to a safety harness. Internally, Tesla has been testing early versions as factory workers of sorts, doing simple, repetitive tasks around its plants. Elon Musk now says those internal pilots will scale up in 2025, with “several thousand” robots working inside Tesla before the company attempts real mass production.

The Gen 3 version is meant to be the turning point. Tesla says it will have “major upgrades” over the current 2.5 design, including a new hand system, better dexterity, and more capable onboard AI so it can learn from observation, video, and voice commands rather than just being remote-controlled or hard-coded. That evolution matters because so far, much of Optimus’ public demo magic has turned out to be closer to teleoperation than autonomy, with operators guiding its movements behind the scenes — something Tesla has been quietly criticized for by roboticists.

Musk has been unusually explicit about where he thinks this is going. He has told investors and the World Economic Forum crowd that Optimus will end up everywhere: on Tesla’s own assembly lines, in other companies’ factories, in warehouses, as home assistants, and even as surgeons once the technology matures. He has floated the idea of selling Optimus to outside companies in the second half of 2026 and, if safety and reliability checks out, to the general public by late 2026. He’s even tossed out a long-term target price of around $30,000 per robot, arguing that at scale, Tesla could get the cost of goods near $20,000 and still make healthy margins.

That financial angle is important because Tesla is not pitching Optimus as a fun side project. The company is retooling itself around what Musk calls “embodied intelligence” — in other words, AI that doesn’t just generate text or images but can move through the world and act on it. Capital spending is set to spike to roughly $20 billion in 2026, with a huge chunk earmarked for robotics rather than new car platforms. Fremont, the California plant that built Tesla’s high-end Model S and Model X, is being converted into an Optimus factory instead, trading something like 50,000 premium cars per year for a stated goal of up to 1 million robots from the same floor space.

There’s also a lot personally riding on these metal shoulders. Musk’s eye-watering compensation package — regularly estimated in the trillion-dollar range once you account for the equity upside — is now tied, in part, to building at least 1 million Optimus units. That creates an obvious incentive to talk big about humanoid robots and their impact on Tesla’s future, especially at a moment when the core EV business is under pressure from slowing demand, tougher competition, and the loss of some government incentives.

At the same time, Tesla is hardly alone in this new humanoid gold rush. The broader market for human-shaped robots is tiny today but projected to grow fast — from under $3 billion in the mid-2020s to over $15 billion by 2030, with compound growth north of 39 percent. Companies like Agility Robotics (with its Digit robot), Figure AI, Apptronik, and a growing cluster of Chinese players are all chasing the same basic vision: a general-purpose worker that can slot into environments designed around people. Agility, for example, has already built a dedicated factory that can turn out around 10,000 Digit robots a year, and Chinese manufacturers like BYD and Agibot are plotting their own volume ramps into the tens of thousands by 2026.

That’s the opportunity. The skepticism comes from people who actually build robots for a living. Humanoid machines are still fragile, expensive, and prone to falling over in environments that are even slightly messy or unpredictable. They can be incredibly impressive in a choreographed demo, but getting one to reliably work an eight-hour shift on a noisy factory floor — or unload your dishwasher without destroying half your plates — is a different class of problem. Several roboticists have openly questioned whether the industry is jumping ahead of what’s practically deployable outside tightly controlled industrial settings.​

Tesla’s own track record doesn’t help. Optimus has had a rocky rollout marked by awkward teleoperation revelations and the high-profile departure of the company’s head of robotics, Milan Kovac, which raised questions about internal turmoil just as the program was supposed to be hitting its stride. Musk also has a long history of aggressive timelines that age badly. He once predicted 5,000 Optimus units would be built in 2025; even Tesla’s own updated guidance now makes that look optimistic, and analysts tend to treat his dates as directional wishes rather than hard schedules.

Still, if you zoom out, you can see why Tesla keeps hammering this story. In a world where AI models are getting better at reasoning and perception, packaging that intelligence into a body that can move, grasp, and navigate is the logical next step. A robot worker that costs the equivalent of an entry-level car but can operate around the clock would be transformative for manufacturing, logistics, and certain types of service work. Even a limited, boring version of Optimus that can safely do a handful of repetitive jobs could be extremely lucrative at scale.

The risk is that the hype gets so far ahead of the tech that it starts to distort everything else. When you shut down iconic car lines and pour tens of billions into a product that doesn’t really exist yet, you’re not just betting on the future — you’re burning some of the present. If Optimus Gen 3 arrives in early 2026 looking polished, autonomous enough, and capable of things existing robots struggle with, Tesla’s pivot will look bold and maybe even obvious in hindsight. If it shows up as yet another carefully edited demo with more questions than answers, investors and customers may start asking whether the company built on electric cars got a little too distracted trying to build the perfect humanoid.