Comparisons

Tesla Optimus vs Figure AI: Who's Actually Shipping Robots (2026)

By ToolPilot Editors · Updated October 2, 2026

Tesla Optimus vs Figure AI: real deployments, production numbers, AI tech, and pricing. Who is actually shipping humanoid robots in 2026?

Ask "who's winning the humanoid robot race?" and two names come up: Tesla's Optimus and Figure AI. They are the two most famous humanoid robot programs in the world — one run by the world's richest man at a car company, the other a startup valued at $39 billion. But fame and progress are different things. Here is who is actually shipping robots in 2026.

The short answer: Figure AI has real robots doing paid work at a BMW factory right now, and Tesla has production lines ramping up hundreds of units per week. Neither sells to the public. This guide compares their deployments, production numbers, technology, pricing reality, and which approach looks stronger — without the hype.

Side-by-side: Optimus vs Figure

Real deployments: Figure is ahead

This is the category that matters most, and Figure leads it clearly. Its Figure 02 robots worked at BMW's Spartanburg, South Carolina plant through 2025, contributing to the assembly of roughly 30,000 vehicles, and Figure 03 took over on the line in June 2026 doing more complex sequencing work — picking vehicle components from unsorted containers and placing them into trolleys in assembly order. That is a paying customer with robots doing logistics work, not a demo.

Tesla, meanwhile, has not put Optimus to commercial work anywhere. Musk admitted on Tesla's Q4 2025 earnings call that the robot was "not in usage in our factories in a material way." The plan now is to lease Optimus units to external companies where they will collect training data — a stepping stone, not a shipped product. Figure is also the more dramatic storyteller: in late September 2026, the company publicly retired its Figure 02 fleet by having the old robots walk, leap, and dive into an electric arc furnace in Finland — a theatrical end-of-life that underscored how fast the hardware generations turn over.

Production scale: Tesla is ramping hard

Scale is where Tesla's automotive muscle shows. Supply-chain reports from September 2026 indicate roughly 15,000 Optimus units being built in 2026, with production at several hundred per week and a target of 1,000 per week by year-end. Tesla is converting an existing vehicle production line into an Optimus factory, targeting up to a million robots per year eventually.

Figure's BotQ facility hit a milestone of one robot per hour — a 24x ramp in under 120 days — producing 60 to 70 humanoids weekly with several thousand per year as the near-term target. Respectable for a startup, but an order of magnitude behind Tesla's trajectory.

Both programs have real manufacturing problems, though. Reports from The Information describe Tesla struggling with Optimus's hand dexterity — over 100 tiny screws and components requiring manual assembly — plus component reliability issues from suppliers and units needing rework off the line. Growing pains, but growing.

The brains: two different bets

Figure builds its robots around Helix, its own vision-language-action AI model developed in-house after the company ended its OpenAI partnership in early 2025. Helix converts camera feeds and natural-language instructions into joint-level motor commands. Figure recently tested Helix 2.5 across 30 homes the robots had never seen: 56% task success overall (toy tidying, towel folding, bed making), strictly scored — a meaningful generalization test, though independently unverified and still far from unsupervised reliability.

Tesla bets on its existing advantage: the same AI5 chip and vision-based neural network stack that powers its vehicles, plus the Cortex 2.0 supercomputer training backbone shared with its robotaxi program. Optimus reportedly runs semi-autonomously in defined environments but needs supervised operation and takes days of training to learn basic jobs. Tesla has not published equivalent generalization tests, so its AI claims are harder to evaluate from the outside.

Pricing reality: neither number is real yet

Musk has long quoted a $20,000–$30,000 target price for Optimus "at scale," and Figure targets under $20,000 for a future consumer robot. Treat both as aspirations. Current manufacturing costs for Optimus are estimated at $50,000 to $100,000 per unit, and analysts expect initial commercial pricing to reflect that. Figure's valuation ($39B) dwarfs its revenue, which was negligible until its first commercial payments — the company is priced on the future, not the present.

Neither company offers preorders or a public sale date. Estimates put Tesla's enterprise sales in late 2026–2027 and consumer availability toward the end of 2027 at the earliest — and Musk's timelines have slipped repeatedly since the project was announced in 2021.

Who is actually winning?

It depends what game you measure. Figure is winning at doing the job: real paying deployments, an in-house AI stack with published generalization tests, and a factory producing robots by the hour. Tesla is winning at scaling the factory: automotive-grade production capacity, a shared AI platform across robots and robotaxis, and $20 billion of annual investment.

The honest assessment for 2026: neither company has a robot you can buy, and both face serious technical hurdles — dexterity, reliability, and cost. Figure's approach looks more like a company that has to earn its survival with working robots, while Tesla's looks like a bet that manufacturing scale solves everything. For buyers, developers, and researchers, Figure's deployed-at-BMW track record is the one you can verify today. For everyone else, keep watching the factory floors — not the keynote demos.

One final note: the humanoid field is broader than these two names. Agility Robotics has its Digit working in warehouses, and Chinese makers like Unitree already sell robots to developers today. Optimus and Figure AI dominate the headlines, but the real competition in 2026 includes factories where robots — any robots — already clock in for paid shifts.

Frequently asked questions

Can I buy a Tesla Optimus or Figure robot?

No. Neither robot is available for public purchase. Tesla offers no preorders and no public sale date, with consumer sales not expected before late 2027. Figure 03 is not for sale either; home pilot deployments are planned but no consumer timeline is confirmed.

Which company has more robots working in the real world?

Figure AI. Its robots have a paid commercial deployment at BMW's Spartanburg plant with around 740 robots deployed, while Optimus remains internal to Tesla with no external commercial customers as of 2026.

How much will Optimus cost?

Musk targets $20,000–$30,000 at scale, but current production costs are estimated at $50,000–$100,000 per unit, so early commercial pricing will likely be far above the target.

What AI powers each robot?

Figure uses Helix, its own vision-language-action model built in-house. Tesla uses its vehicle-derived vision AI stack on the AI5 chip, sharing training infrastructure with its robotaxi program.

Is Musk's $25 trillion prediction realistic?

At Davos 2026, Musk predicted Optimus could make Tesla a $25 trillion company. That is a long-range speculation based on hypothetical millions of robots sold — not a financial forecast tied to current revenue. Treat it as ambition, not evidence.