Humanoid Robotics Milestone Tracker (2026)
In our humanoid robotics market deck, you will find everything you need to understand the market
Humanoid Robotics Milestone Tracker
This table tracks 33 deployment milestones for nine leading humanoid robots, with each cell rated Yes, No, or Not Really based on tier-one sourced evidence as of April 2026. Columns are ordered from most commercially advanced (left) to earliest stage (right).
The first column lists progressively harder questions, from basic prototyping to scaled commercial deployment, while each robot column shows how far that program has actually gone. If you want to dig deeper into this market, you can check out our humanoid robotics market report.
| Question | Digit (Agility) | Optimus (Tesla) | Atlas (Boston D.) | Apollo (Apptronik) | NEO (1X) | Figure 03 | G1 (Unitree) | IRON (XPENG) | GR-3 (Fourier) |
|---|---|---|---|---|---|---|---|---|---|
| Has the company announced a humanoid program? | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Has it named a specific humanoid product? | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Has it defined a target use case? | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Has it published target specifications? | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Has the company built a humanoid prototype? | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Has it shown the prototype publicly? | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Has it released prototype operating footage? | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Has it performed a live public demo? | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Has it shown autonomous walking? | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Yes |
| Has it shown object manipulation? | Yes | Yes | Yes | Yes | Yes | Yes | Yes | Not Really | Not Really |
| Has it shown multi-step task execution? | Yes | Yes | Yes | Yes | Yes | Yes | Not Really | Not Really | Not Really |
| Has it shown autonomous task completion? | Yes | Not Really | Not Really | Not Really | Not Really | Not Really | Not Really | No | No |
| Has it disclosed dexterity or precision metrics? | Yes | Yes | Yes | Not Really | Yes | Yes | Yes | Yes | Not Really |
| Has it shown repeated task consistency? | Yes | Not Really | Not Really | Not Really | No | Not Really | No | No | No |
| Has it operated outside controlled demos? | Yes | Yes | Not Really | Not Really | No | Not Really | Not Really | Not Really | No |
| Has it worked inside company facilities? | Yes | Yes | Yes | Yes | No | Yes | Not Really | Yes | Not Really |
| Has it been tested in an external workplace? | Yes | No | Not Really | Yes | No | Yes | No | No | No |
| Has it completed non-demo tasks? | Yes | Not Really | Not Really | Not Really | No | Not Really | No | No | No |
| Has it been tested at a customer site? | Yes | No | Not Really | Yes | No | Yes | No | No | No |
| Has it worked safely around humans? | Yes | Yes | Not Really | Yes | Not Really | Not Really | Not Really | No | Not Really |
| Has it worked for multiple days? | Yes | Yes | Not Really | Not Really | No | Not Really | No | No | No |
| Has it required limited teleoperation? | Yes | Not Really | Not Really | Not Really | No | Not Really | Not Really | No | No |
| Has it performed economically useful work? | Yes | No | No | Not Really | No | Not Really | No | No | No |
| Has it repeated useful work reliably? | Yes | No | No | No | No | No | No | No | No |
| Has it achieved reliable field uptime? | Not Really | No | No | No | No | No | No | No | No |
| Has a customer used it daily? | Yes | No | No | No | No | No | No | No | No |
| Has it moved beyond one pilot? | Yes | No | No | Not Really | No | Not Really | No | No | No |
| Can businesses order it today? | Yes | No | No | Not Really | Yes | No | Yes | No | Not Really |
| Has public pricing been announced? | Yes | Not Really | No | Not Really | Yes | Not Really | Yes | No | Not Really |
| Are delivery timelines publicly available? | Yes | Not Really | Not Really | Not Really | Yes | Not Really | Yes | Not Really | No |
| Can customers deploy it as sold? | Yes | No | No | No | Not Really | No | Yes | No | No |
| Has it won multiple paying customers? | Yes | No | No | Not Really | No | Not Really | Yes | No | No |
| Has it scaled across multiple sites? | Not Really | Not Really | No | No | No | No | Not Really | No | No |
This market map, featured in our humanoid robotics market deck, highlights top companies and startups in the humanoid robotics market
Insights
We compared nine leading humanoid robot programs across 33 milestones spanning prototyping, public demos, workplace testing, customer deployment, and commercial readiness: Agility's Digit, Tesla's Optimus, Boston Dynamics' Atlas, Apptronik's Apollo, 1X's NEO, Figure 03, Unitree's G1, XPENG's IRON, and Fourier's GR-3. Each one was scored using tier-one sourced evidence as of April 2026 to separate headline momentum from verified deployment progress. Here is what stood out.
- Agility's Digit is the only humanoid that has completed the full cycle from pilot to paying commercial customer. Every other program remains in some variant of demo, pilot, or internal testing.
- Agility's paying customers (GXO, Toyota, Mercado Libre, Amazon, Schaeffler) all share one trait: repetitive tote movement. The real entry point for humanoid robots is third-party logistics, not general-purpose tasks.
- The Robot-as-a-Service model is the quiet winner of this cycle. It transfers risk from buyer to vendor, which is critical when the technology is still immature.
- On Tesla's Q4 2025 earnings call, Musk admitted ~1,000 deployed Optimus units aren't doing "useful work." They exist for learning and data collection, not production tasks.
- Tesla is converting its Fremont Model S/X production lines to Optimus manufacturing. It is cannibalizing profitable vehicle lines for a robot its own CEO says isn't useful yet.
- Figure AI is valued at $39 billion with one pilot customer (BMW) and zero consumer sales. Goldman Sachs projects the entire humanoid market at $38 billion by 2035.
- Apptronik has raised nearly $1 billion with zero disclosed revenue from robot sales. Its funding-to-deployment-evidence ratio is the most extreme in the sector.
- XPENG publicly admitted its IRON robot failed in its own factory. Hands wore out in one month, and total robot cost exceeded Chinese factory worker wages.
- Every company's "degrees of freedom" spec is used as marketing, not a measure of utility. XPENG's 82 DoF sounds impressive versus Digit's ~30, but Digit is the one doing actual work.
- Almost no company publishes maintenance or durability data for its humanoid. No mean-time-between-failures figures exist for any robot. This is the real bottleneck nobody talks about.
- Unitree leads globally with 5,500+ shipped humanoid units, confirmed in its IPO prospectus. However, the 127 cm G1 is a research platform, and very few units do economically productive work.
- China dominates humanoid robot volume; the US dominates deployment maturity. Chinese companies ship over 90% of global units, but Agility leads on verified commercial traction.
- There is a stealth bifurcation between home, factory, and service humanoids. These categories face fundamentally different engineering, regulatory, and commercial challenges.
- Multiple CEOs promised humanoid robots in homes by 2026. Not one humanoid has been deployed in a consumer home for sustained daily use. Realistic timelines point to 2028 or 2029.
This chart, featured in our humanoid robotics market deck, compares the main business model options for humanoid robot manufacturers
Which humanoid robot company looks furthest ahead today on real-world performance?
Agility Robotics is the furthest ahead on real-world performance in humanoid robotics, because it is the only company that has completed the full cycle from prototype to pilot to paying commercial customer doing sustained daily work as of April 2026.
At GXO's Flowery Branch, Georgia facility, Agility's Digit robot has moved over 100,000 totes in a live commerce fulfillment operation, a milestone announced on November 20, 2025. This is documented commercial throughput at a third-party logistics site where Digit operates in production workflows alongside automated mobile robots and conveyors.
On February 19, 2026, Toyota Motor Manufacturing Canada signed a commercial Robots-as-a-Service agreement to deploy seven Digit robots at its Woodstock, Ontario plant, the facility that produces the RAV4. This Toyota deal followed a year-long pilot during which three Digit units were evaluated in real factory conditions. Agility's confirmed commercial customers in humanoid robotics now include GXO, Toyota TMMC, Mercado Libre, Schaeffler, and Amazon.
The other humanoid robot contenders are not close on deployment maturity. Boston Dynamics began production of its all-electric Atlas in January 2026, but all 2026 Atlas units are committed to Hyundai and Google DeepMind as R&D partners, with factory deployment at Hyundai's Georgia Metaplant not expected until 2028. Tesla confirmed on its Q4 2025 earnings call that Optimus units are "primarily for learning and data collection rather than performing productive tasks," with zero external customers. Figure AI completed an 11-month pilot at BMW Spartanburg loading over 90,000 parts, but BMW remains a single pilot customer, and Figure 03's home demonstrations in March 2026 were staged environments.
Which humanoid robot companies have real commercial traction today, versus strong demos and fundraising but limited real-world deployment?
Agility Robotics and Unitree Robotics are the only humanoid robot companies with real commercial traction today. Figure AI, Apptronik, Boston Dynamics (Atlas), 1X Technologies, and Tesla (Optimus) all have strong demos, significant funding, or both, but limited or zero real-world deployment generating revenue in humanoid robotics.
Agility Robotics has multiple paying RaaS customers for its Digit humanoid robot (GXO, Toyota TMMC, Mercado Libre, Schaeffler, Amazon), over 100 Digit units sold, and a manufacturing facility in Salem, Oregon with capacity for 10,000+ units per year. Unitree Robotics shipped 5,500+ humanoid units by end of 2025, confirmed in its IPO prospectus filed March 2026, with revenue of 1.7 billion RMB (~$235M) and 335% year-over-year growth. However, most of Unitree's humanoid customers are universities and research labs; very few G1 units are doing economically productive work.
Figure AI reached a $39 billion valuation in September 2025 but has one pilot customer (BMW) and zero consumer sales of its humanoid robot. Apptronik raised nearly $1 billion in total funding but TechCrunch confirmed as of February 2025 that its Mercedes-Benz humanoid robot partnership "has yet to graduate beyond the pilot stage." Boston Dynamics Atlas generates zero commercial revenue in humanoid robotics, with all 2026 units committed to R&D partners. 1X Technologies opened NEO humanoid robot pre-orders at $20,000 but has shipped zero units. Tesla Optimus has zero external customers and zero commercially useful output by Musk's own admission on the Q4 2025 earnings call.
Which humanoid robot company looks the most overvalued relative to actual deployment evidence today?
Figure AI is the most overvalued humanoid robot company today, at a $39 billion valuation that is heavily disconnected from its deployment evidence.
Figure AI's deployment record consists of one pilot customer (BMW Spartanburg), an 11-month deployment of Figure 02 that loaded over 90,000 parts across 1,250+ runtime hours, a BotQ facility in California with 12,000 units/year capacity, and $1.9 billion in total funding. Figure AI does not have multiple paying customers, consumer sales, confirmed mass-production pricing, or revenue at meaningful scale for its humanoid robot.
Goldman Sachs projects the entire global humanoid robot market at $38 billion by 2035, meaning Figure AI's current valuation already exceeds the projected total addressable market nine years out. Agility Robotics, which is 18 to 24 months ahead of Figure on commercial humanoid robot deployment with multiple paying customers, raised $400 million in its Series C, roughly 1/100th of Figure's valuation. Unitree, which shipped 5,500+ humanoid units and generated $235 million in actual revenue, is seeking roughly $580 million in its IPO.
Figure AI's former head of product safety also sued the company in November 2025 for being fired after raising concerns that the robots were strong enough to fracture a human skull. Figure AI's $39 billion valuation reflects investor enthusiasm for the humanoid robotics narrative and Figure's demo capabilities, not deployment evidence or proven product-market fit.
This chart, featured in our humanoid robotics market deck, shows how Agility Robotics is capturing share in humanoid robotics
Why has Tesla not become the clear leader in humanoid robotics with Optimus, despite its financial scale?
Tesla has not become the clear leader in humanoid robotics because financial scale does not compensate for a late start in a different engineering discipline, divided organizational attention, and a pattern of missed timelines that has eroded credibility with customers and partners.
Tesla held $41.6 billion in cash and investments as of Q3 2025, generated record free cash flow of nearly $4 billion that quarter, and committed $20 billion in capex for 2026. Capital is definitively not the constraint for Tesla's humanoid robotics program.
Agility Robotics is further ahead than Tesla Optimus on every humanoid robot deployment metric: multiple paying customers (GXO, Toyota TMMC, Mercado Libre, Schaeffler, Amazon), over 100,000 totes moved in live operations, and commercial RaaS agreements at multiple sites. Boston Dynamics Atlas has stronger hardware than Optimus (56 degrees of freedom versus roughly 30, 50 kg payload versus 20 kg, autonomous battery swapping), backed by 30+ years of bipedal robotics experience.
Tesla Optimus has been deployed internally since mid-2024 but exclusively for data collection, not productive work. Musk confirmed on January 28, 2026 that Optimus robots are not doing useful work yet. Tesla's 2025 target of 5,000 Optimus units yielded several hundred to roughly 1,000, and the Optimus Gen 3 reveal planned for Q1 2026 was missed.
Tesla announced Optimus in August 2021, entering a field where Agility Robotics (founded 2015) and Boston Dynamics (bipedal robots since 2009) had 5 to 15 years of accumulated know-how in bipedal dynamics and manipulation. Tesla's FSD advantage transfers less to humanoid robotics than commonly assumed: FSD solves 2D navigation on structured roads, while humanoid robotics requires 3D manipulation with force sensing, sub-millimeter precision, and whole-body balance, as the ongoing Optimus Gen 3 hand redesign confirms.
Tesla Optimus also lacks deployment data from external customer environments, unlike Agility which trains on data from GXO, Toyota, and Mercado Libre facilities with different layouts and conditions. China's rare earth export restrictions constrained materials for Optimus actuators, and Tesla's vertical integration becomes a bottleneck when the humanoid robot design is still rapidly iterating.
Tesla's organizational attention is split across the refreshed Model Y, Cybercab robotaxi, FSD in Europe, Megapack, Tesla Semi, and Optimus simultaneously. In June 2025, Optimus program head Milan Kovac resigned after three years. Musk's timeline commitments for Optimus have been systematically missed (production-ready by 2023, 5,000 units in 2025, Gen 3 in Q1 2026), causing external customers and partners to increasingly discount Tesla's humanoid robotics announcements.
What is the probability Tesla ships genuinely useful humanoid robots before Figure does?
The probability Tesla ships genuinely useful humanoid robots before Figure does is currently 35 to 50%, closer to a coin flip than most observers expect, because Tesla's vast resource advantage is offset by its pattern of missed Optimus timelines and harder-than-expected hardware iteration.
Tesla has $41.6 billion in cash, $20 billion committed in 2026 capex, and 1,000+ Optimus units generating internal training data. If Tesla delivers even modestly useful Optimus robots to its own factories for repetitive tasks like battery sorting, it would technically ship "useful" humanoid robots before Figure ships to consumer homes. Tesla is targeting external Optimus sales by late 2026 or 2027.
Figure AI has already completed an 11-month humanoid robot deployment at BMW with documented throughput: 90,000+ parts loaded across 1,250+ runtime hours. Figure 03 targets home deployment at $20,000 with select placements planned for late 2026. Figure AI's Helix AI system is purpose-built for embodied intelligence, and the company has shipped three humanoid robot generations in under three years, a faster hardware iteration cycle than Tesla Optimus.
The critical distinction in this Tesla-versus-Figure comparison is that "useful" for Tesla Optimus means repetitive industrial tasks in Tesla's own controlled factories, a lower bar, while "useful" for Figure means household tasks in consumer homes, a vastly harder problem. Neither Tesla nor Figure has achieved externally deployed, economically productive humanoid robot work at a paying customer site. Agility Robotics remains the only humanoid robot company that has.
This chart, featured in our humanoid robotics market deck, illustrates yearly funding for humanoid robotics startups
What is the probability that China will win the humanoid robotics race?
The probability China leads globally in humanoid robotics on manufacturing dominance, deployment scale, and cost position by 2030 is 45 to 60%, because China already dominates on volume, cost, and supply chain while the U.S. leads on AI software, and the outcome depends on which layer captures more value.
"Winning" the humanoid robotics race means achieving dominance on at least three of five indicators: largest market share by units deployed, lowest per-unit cost at scale, deepest supply chain control, largest installed base doing productive work, and most advanced AI/autonomy in deployed robots. China is currently ahead on the first three, roughly even on the fourth, and behind on the fifth.
China installs 54% of the world's industrial robots, Chinese humanoid companies shipped over 90% of global humanoid units by volume in 2025, and China's humanoid robot cost advantage is structural, built on the same supply chain that produces EVs and batteries at global-beating cost. The Chinese government committed $138 billion to a robotics and AI fund, and China's 15th Five-Year Plan embeds humanoid robotics as a national strategic priority. The U.S. leads on AI software for robotics through Physical Intelligence, Google DeepMind, and Skild.ai, and if U.S. AI models become the standard for robot intelligence, Chinese hardware manufacturers risk becoming commodity suppliers.
The probability should not be materially lower because China's advantages in manufacturing scale and government coordination compound over time, and the U.S. has no national robotics strategy comparable to China's Five-Year Plan. The probability should not be materially higher because China remains dependent on Japanese suppliers for critical precision components like harmonic reducers, and the U.S. semiconductor advantage via TSMC and NVIDIA constrains Chinese AI training capacity. The outcome hinges on how fast China localizes Japanese precision components, whether general-purpose AI models can run on any hardware (letting China win on cost), and whether China's faster regulatory environment creates a permanent data advantage over Western competitors.
Where is China doing better than Western companies in robotics?
China is ahead of the West in three areas of robotics: industrial robot deployment scale and density, humanoid robot production volume and cost, and speed of ecosystem development from prototype to market.
China installed approximately 295,000 industrial robots in 2024, representing 54% of global total installations according to the International Federation of Robotics. China now operates over 2 million robots in its factories, the largest installed base globally, with robot density at 470 per 10,000 manufacturing workers in 2023, surpassing the United States at approximately 295. Chinese companies like Xiaomi and BYD operate fully automated dark factories around Shanghai and Shenzhen, and Chinese industrial robot exports grew by roughly 65% per year from 2022 to 2024.
In humanoid robot production, Chinese companies dominate global volume. Unitree alone shipped 5,500+ humanoid units in 2025, more than all Western humanoid robot companies combined, and Chinese companies capture over 90% of global humanoid robot sales by unit volume according to Omdia. The humanoid robot pricing gap between China and the West is dramatic: Unitree's G1 starts at roughly $17,990 and the R1 at roughly $5,900, compared to Agility's Digit at roughly $250,000 and Boston Dynamics Atlas at roughly $150,000.
China's robotics ecosystem development speed is accelerating through policy. China's 15th Five-Year Plan (2026-2030) elevated robotics and "embodied intelligence" into a national strategic priority with its own dedicated track. The Chinese government approved a $138 billion state venture fund for robotics and AI, and at least five Chinese embodied AI companies have each raised more than $210 million. Unitree open-sourced its UnifoLM-VLA-0 vision-language-action model in March 2026, accelerating the broader Chinese robotics ecosystem.
Where China is not ahead in robotics: AI software generalization (Physical Intelligence, Google Gemini Robotics, and Skild.ai in the U.S. still lead), enterprise humanoid robot deployment maturity (Agility Robotics leads globally), and high-end precision components (Japan's Harmonic Drive Systems and German suppliers still dominate critical reducers and servos).
What is the probability that humanoid robots will be in homes?
The probability of meaningful humanoid robot adoption in consumer homes is 2 to 5% by end of 2026, 8 to 15% by end of 2027, 20 to 35% by end of 2028, and 50 to 65% by 2030. "In homes" means at least 1,000 humanoid robots performing daily domestic tasks in consumer households through a commercial channel, not pilots or press events.
No humanoid robot has ever been deployed in a consumer's home for sustained daily use as of April 2026. Figure 03 demonstrated eight autonomous cleaning skills in March 2026 but in staged environments. 1X Technologies targets US NEO humanoid robot deliveries in Q3-Q4 2026 with acknowledged 60 to 70% autonomy. Even if both ship to early-access homes in late 2026, the total will be dozens to low hundreds. By 2028, two to three full iteration cycles will have completed and humanoid robot prices could fall below $15,000. By 2030, AI models will have had four to five additional years of real-world home data, and prices at $10,000 to $20,000 become accessible to upper-middle-income households.
Home humanoid robot adoption would most likely start in Japan and South Korea due to severe demographic pressure and cultural acceptance, then China due to government support. The companies most likely to deploy humanoid robots in homes first: 1X (NEO) has teleoperation-augmented autonomy and open pre-orders at $20,000; Figure (Figure 03) has the strongest home-focused AI; Unitree (R1) has the lowest price at $5,900 but is a research platform; Tesla has manufacturing scale but consumer Optimus sales are not expected before late 2027.
Six non-obvious factors will determine the timeline for humanoid robots in homes. Liability and insurance frameworks do not exist for robot-caused injury in homes. Consumer Wi-Fi is unreliable for robots depending on cloud AI. Consumer tolerance for imperfect humanoid robot performance is untested at scale. Home environments are adversarial in ways factories are not: pets, children, wet floors, fragile objects. Charging and maintenance burden on consumers will drive high abandonment if not near-zero-effort. Cultural acceptance of humanoid robots in homes varies by 3 to 5x across markets, making adoption fragmented rather than uniform.
This chart, featured in our humanoid robotics market deck, shows how warehouse automation has driven growth in the humanoid robotics market over time
This tracker is a live research tool we maintain to assess which humanoid robotics programs appear closest to real-world deployment at scale. It is updated as new evidence emerges.
The tracker is focused on one question: how far has each humanoid robot actually progressed from announcement to commercial deployment? It is not designed to score every dimension of a company's potential.
To be included, a company must have a publicly announced full-body humanoid robot program with observable evidence of progress. Upper-body systems, wheeled service robots, exoskeletons, and robot heads are excluded.
The 33 questions are ordered by increasing difficulty. Early rows (prototype, demo, walking) are easy to achieve. Later rows (reliable uptime, multiple paying customers, multi-site scaling) are hard. This structure makes it visually obvious where each program's evidence runs out.
Each cell is rated Yes, No, or Not Really. "Yes" means we found credible, tier-one evidence. "No" means we found no such evidence. "Not Really" means partial, ambiguous, or heavily caveated evidence exists.
We rely on direct and credible sources: company press releases, customer announcements, SEC filings, IPO prospectuses, executive earnings call statements, and reporting from publications like TechCrunch, Bloomberg, and IEEE Spectrum. We do not treat demo videos, social media posts, or unverified claims as strong evidence on their own.
When evidence is partial or hard to verify, we score conservatively. The tracker should be read as directional, not absolute. The goal is a disciplined view of which programs seem closest to practical deployment, based on what can be shown today. You can find more details and additional analysis in our humanoid robotics market report.
As this chart shows, and as featured in our humanoid robotics market deck, search interest in where to buy robots has been rising steadily
Related blog posts
- How strong is fundraising in the humanoid robotics market right now?
- The most highly valued startups in the humanoid robotics market
- Who is currently leading the humanoid race?
- How close is Tesla Optimus to being deployed at scale?
- Which humanoid robotics companies seem closest to scale?
- Which humanoid robot is the best as of today?
- How close is Agility Digit to large-scale deployment?
- Can we believe the impressive humanoid robot demos?
- What is some real-world evidence of humanoid robots at work today?
- Which of Musk’s Optimus promises have been fulfilled so far?
