Humanoid Deployment Tracker (2026)

The Humanoid Deployment Tracker

This table ranks humanoid robotics companies based on how close they appear to real-world deployment at scale.

We focus on four things:

• Task Usefulness: how useful and practically valuable the robot's work appears to be.
• Autonomy: how independently the robot can perform that work.
• Real-World Robustness: how stable, repeatable, and resilient the robot seems outside a polished demo setting.
• Deployment Maturity: how close the company appears to actual rollout through pilots, customer traction, industrial partnerships, or scaling plans.

The goal is to keep the table easy to read at a glance while showing which companies look strongest across the core questions that matter for real deployment.

How close is each humanoid robot company to large-scale real-world deployment?

Agility Robotics (Digit)

Agility has the clearest evidence of a humanoid doing useful warehouse work in a live customer workflow. Digit has a commercial deployment with GXO (multi-year RaaS agreement, 100,000+ totes moved), plus agreements with Mercado Libre and Toyota Motor Manufacturing Canada. The main gap: public detail on intervention rates and remote support needs remains limited.

Figure AI (Figure 03)

Figure AI has strong numbers from a real automotive environment. The BMW deployment produced 10-hour weekday shifts, 1,250+ runtime hours, 90,000+ parts loaded, and contribution to 30,000+ vehicles. However, the best proof is still tied to Figure 02, not Figure 03. How transferable those gains are across customers, and how far Figure 03 has replaced 02 in production, remains unclear.

Kepler Robotics (K2 Bumblebee)

Kepler has a strong commercialization narrative. K2 targets industrial work with published payload and endurance specs, a 90-day delivery lead time, and outside coverage of mass production and framework agreements for several thousand units. The limitation: the strongest deployment evidence still comes mostly from Kepler's own materials rather than independent validation.

UBTECH (Walker S2)

UBTECH has an aggressive deployment narrative, highlighting autonomous battery swapping for continuous operation and claims of mass-producing and delivering several hundred Walker S2 units. However, the biggest scale claims come from UBTECH's own materials, and public third-party validation of in-field performance is thin.

Fourier (GR-1)

Fourier shows a real commercial product posture. GR-1 is documented as commercially mass-producible with a full developer and support stack, and NVIDIA used it in public GR00T material and released a dataset built on the platform. Most visible work still looks like structured manipulation and institutional usage, not broad real-world labor deployment.

Boston Dynamics (Atlas)

Boston Dynamics unveiled production Atlas at CES January 2026 and began manufacturing immediately. All 2026 deployments go to Hyundai and Google DeepMind, with Hyundai planning a 30,000-unit-per-year factory by 2028. Atlas now looks far more product-directed, but public evidence of repeated, customer-side production work at scale is still thin.

Apptronik (Apollo)

Apptronik has named partners in Mercedes-Benz, GXO, and Jabil, the latter combining factory validation with a credible manufacturing partner. However, most public evidence still describes pilots and preparation to scale rather than operating fleets with clear uptime and throughput numbers.

Tesla (Optimus)

Tesla's filings describe a Fremont pilot production line in 2025 and Gen 3 as the first mass-production design, with production planned before end-2026. External reporting indicated Tesla was well behind earlier unit targets. Optimus is beyond concept stage but not yet robustly deployed at scale.

1X (NEO Gamma)

1X has some of the clearest evidence of a humanoid being pushed into real home environments. The company has discussed testing NEO Gamma in a few hundred to a few thousand homes, and its product page states that early access includes only basic autonomy with scheduled remote supervision. That transparency helps but caps the autonomy score; robust, low-intervention home performance at scale is unproven.

Sanctuary AI (Phoenix)

Sanctuary AI was early to commercial pilots and showed breadth: the Canadian Tire pilot covered 110 retail tasks, and the Magna partnership offers a credible industrial path. The story still looks pilot-heavy, with less proof that Phoenix can sustain high repeatability and low intervention in production environments.

AgiBot (A2 Ultra)

AgiBot claims commercial deployment across 20+ enterprises but provides little operational detail on autonomy or robustness. Stronger reporting exists around the related A2-W platform in Fulin Precision's factories. For A2 Ultra specifically, the proof trail on repeatability and independence is too thin.

Booster Robotics (Booster T1)

Booster T1 is a real shipped platform, but it reads more as a developer, education, and RoboCup product than a near-term worker humanoid. It's available globally with a notably open software stack, but the public use case isn't yet broad practical work.

NEURA Robotics (4NE1)

NEURA has built an ambitious industrial narrative around humanoids in Europe. The Schaeffler partnership is the strongest signal, pointing to integration goals and supply-chain planning. Most public evidence remains company-led capability framing rather than hard proof of repeated production-floor performance.

Unitree (G1)

Unitree has made humanoids visible and accessible, and G1 is clearly real hardware. But public evidence still leans toward developer platform readiness and internal demos rather than validated work deployments. A recent in-house factory assembly video is a step forward, though reporting notes the controlled setting and open generalization questions.

EngineAI (SE01)

EngineAI has built attention around locomotion quality, focusing on human-like gait and dynamic motion, with CES 2025 adding factory-oriented claims. Public evidence still leans heavily toward motion demos rather than proof of repeatable, useful work under realistic operating constraints.

XPENG (IRON)

XPENG says IRON has been integrated into daily factory and store operations and announced a Baosteel ecosystem partnership. Most of what's public is still launch-stage company material, not field metrics on repeatable work cycles, failure handling, or production-side performance.

Which humanoid robotics company leads in real-world deployment readiness?

Agility Robotics leads in real-world deployment readiness because it has the cleanest combination of formal commercial agreements, named live customers, and verified operating throughput.

Deployment readiness is not about who has the most impressive hardware or the largest funding round. It is about which company can show real evidence that a humanoid is doing repeatable, paid work inside a customer workflow. Agility Robotics is the strongest answer on this question because Digit is not just being tested in controlled pilots. GXO publicly described the relationship as a multi-year commercial deployment under a Robots-as-a-Service agreement, and Agility later reported more than 100,000 totes moved in live operation at GXO's Flowery Branch facility.

Since then, Agility has added commercial agreements with Mercado Libre for a fulfillment facility in San Antonio, Texas, and Toyota Motor Manufacturing Canada for RAV4 production logistics at the Woodstock, Ontario plant. Toyota's agreement followed a year-long pilot across three phases and will deploy seven Digit robots for tote handling. That expansion across multiple named customers is the strongest indicator of commercial replication.

Figure AI remains the closest challenger, with the richest published factory operating data from its 11-month BMW deployment. UBTECH's mass-production claims are also notable. But under a strict deployment-readiness lens, Agility's combination of commercial agreements, live throughput data, and multi-customer expansion is still the strongest publicly visible package.

Which humanoid robotics company has the strongest proof of real-world deployment?

Agility Robotics has the strongest proof of real-world deployment because Digit has the clearest publicly documented evidence of operating in a named customer workflow under a formal commercial agreement with measurable throughput.

The key distinction is not that other companies lack progress. It is that Agility's evidence is the least ambiguous. GXO and Agility both described the relationship as a formal commercial deployment, and the 100,000-tote milestone gives the market a concrete operating number tied to a live facility. Agility also celebrated the one-year anniversary of Digit's full-time deployment at GXO in late 2025, confirming sustained rather than short-term operation.

Figure AI has excellent proof of real industrial usefulness. The BMW deployment ran for 11 months, with Figure 02 loading over 90,000 sheet-metal parts across 1,250+ runtime hours. But this is best read as a high-quality pilot that generated design lessons for Figure 03, not as the clearest multi-customer commercial rollout. Figure's own announcement described it as lessons being rolled into Figure 03 operational readiness.

Agility's evidence is narrower in task scope but cleaner in deployment classification. Figure's evidence is broader and impressive but still somewhat more transitional.

Which humanoid robotics company looks most commercially advanced for large-scale rollout?

Agility Robotics looks most commercially advanced for large-scale rollout because it now combines a proven live deployment with multiple additional commercial agreements that suggest repeatability across customers.

The decisive point is that Agility is no longer a one-site story. Beyond GXO, Agility announced a commercial agreement with Mercado Libre in December 2025 and another with Toyota Motor Manufacturing Canada in February 2026. Both follow the same logic of embedding Digit into logistics and manufacturing workflows after earlier validation. Toyota's agreement, for example, followed a year-long evaluation across three phases and resulted in a plan to deploy seven Digit robots. Agility also operates a 70,000-square-foot manufacturing facility in Salem, Oregon, designed for up to 10,000 Digit units per year.

Figure AI may yet become the scale leader, and Boston Dynamics now looks much more product-directed, with all 2026 Atlas deployments committed to Hyundai and Google DeepMind. But those cases still look earlier in commercial rollout maturity than Agility's sequence of live deployment plus signed commercial expansion.

UBTECH is the main wildcard. If its several-hundred-unit delivery claims and 800 million yuan order book hold up under independent operational validation, it could narrow the gap quickly. Today, Agility still has the stronger evidence mix because more of its rollout story is tied to named customer workflow adoption.

Which humanoid robotics company shows the strongest mix of useful work, robustness, and deployment maturity?

Agility Robotics shows the strongest mix of useful work, robustness, and deployment maturity because Digit has public evidence of doing economically relevant repetitive work with enough operating continuity to support a real deployment story.

This combined lens matters more than any single metric. Agility's task set is narrower than the "general humanoid" narrative many founders prefer, but that is exactly why the evidence is stronger. Narrow, repetitive warehouse and factory logistics work in a live facility is easier to evaluate honestly. Digit has accumulated the best public proof on that basis, including the 100,000-tote throughput milestone at GXO, confirmed multi-shift operation, and successful navigation in confined, dynamic warehouse spaces.

Figure AI is very close on useful work and may have the stronger public runtime data per deployment. But Figure still sits a step behind Agility on deployment maturity because its strongest proof belongs to a recently retired prior-generation robot. UBTECH is credible on industrial posture and nonstop-operation design, especially with autonomous battery swapping. But until there is stronger third-party proof of how Walker S2 performs after delivery, it remains harder to rate above Agility on robustness.

Kepler Robotics scores well on this combined lens too, particularly on the deployment maturity side. But its autonomy and robustness evidence still relies more on company-led reporting than on independently confirmed field data.

Which data proves that Agility Robotics is closest to deployment?

The data that most strongly proves Agility Robotics is closest to deployment are the formal GXO commercial agreement, the live 100,000-tote operating milestone, and the follow-on commercial agreements with Mercado Libre and Toyota Motor Manufacturing Canada.

The most important single proof point is the 100,000-tote figure. It converts the debate from "can the robot do a task?" to "has the robot sustained useful output inside an operating facility?" The formal GXO deployment language matters almost as much because it distinguishes commercial adoption from a loose proof-of-concept. Agility also confirmed the one-year anniversary of Digit's full-time deployment at GXO, and described how Digit handles workflow fluctuations by intelligently stacking totes to the side when downstream stations are backed up.

The newer Toyota and Mercado Libre agreements strengthen the case further. Toyota Motor Manufacturing Canada signed a commercial agreement in February 2026 to deploy seven Digit robots for tote handling at its RAV4 plant in Woodstock, Ontario, after a year-long pilot. Mercado Libre signed a commercial agreement in December 2025 for fulfillment work in San Antonio, Texas. Even if their operational metrics are not yet as detailed as GXO's, they show that Agility's deployment story is beginning to replicate across customers and geographies.

The main limitation is that public data on intervention rates, exception handling, and the level of remaining remote support is still sparse. That gap does not undermine the deployment conclusion, but it does limit the precision of the autonomy assessment.

Which data shows that Agility Robotics is closer to deployment than Figure AI?

The data showing that Agility Robotics is closer to deployment than Figure AI is that Agility already has a formal commercial rollout with GXO and subsequent additional commercial agreements, while Figure AI's best public evidence still centers on a highly successful BMW pilot tied to Figure 02.

Agility's proof is deployment-class because the evidence includes a commercial agreement, a live throughput milestone, and expanding commercial adoption across multiple named enterprises. Figure AI's proof is production-relevant because its robots worked 10-hour weekday shifts for 1,250+ hours and loaded 90,000+ parts at BMW Spartanburg, but the public framing still emphasizes learning, retirement of Figure 02, and transition into Figure 03 readiness. Figure's own post described the deployment as generating lessons to be "rolled into Figure 03 operational readiness."

BMW and Figure AI are now evaluating additional use cases for Figure 03, and BMW has also begun a separate pilot with Hexagon's AEON robot at its Leipzig plant. That confirms BMW's broader commitment to humanoids, but it does not yet give Figure 03 the same depth of deployment evidence that Digit has accumulated across GXO, Toyota, and Mercado Libre.

Figure may have the stronger public pilot dataset. Agility has the stronger public commercialization dataset. Under a deployment-readiness framework, commercialization evidence carries more weight.

Why is UBTECH not as close to deployment as Agility Robotics?

UBTECH is not as close to deployment as Agility Robotics because its rollout story is currently more scale-claim-heavy and less independently validated at the level of live customer operating proof.

UBTECH has real strengths. Walker S2 is clearly designed for industrial use, its autonomous battery-swapping capability supports continuous operation, and the company has publicly claimed mass production and first-batch delivery of several hundred units with orders exceeding 800 million yuan. These are not trivial signals.

What UBTECH still lacks, relative to Agility, is an equally clean public proof trail showing a named customer workflow, measurable operational cadence, and widely corroborated outcome metrics in live deployment. Agility has that with GXO and the 100,000-tote milestone, plus the follow-on agreements with Toyota and Mercado Libre that show the deployment model replicating across customers. UBTECH's biggest claims still come through company or press-release distribution rather than clear third-party field validation.

This is not a claim that UBTECH is behind technically. It is a claim that the public evidence base is weaker on the exact question that matters here: proven field deployment maturity. If UBTECH can independently validate post-delivery performance at customer sites, that gap could close quickly.

Which humanoid robotics company has made the most progress in recent months?

Agility Robotics appears to have made the most progress in recent months because it added new commercial agreements after already proving live operation, rather than merely upgrading the product narrative.

The strongest recent evidence is sequential. In late 2025, Agility crossed the 100,000-tote milestone at GXO. In December 2025, it signed a commercial agreement with Mercado Libre. In February 2026, it signed another with Toyota Motor Manufacturing Canada, which expanded from a year-long pilot of three Digit robots to a commercial deployment of seven, with more robots planned if successful. That pattern suggests commercial replication, which matters more than flashy capability updates.

Boston Dynamics deserves mention because January 2026 marked a major change in Atlas's posture. Boston Dynamics unveiled the production version at CES, began manufacturing immediately, and committed all 2026 deployments to Hyundai and Google DeepMind. Hyundai also announced plans for a 30,000-unit-per-year robotics factory by 2028. That is major progress, but it is still earlier in public field proof than Agility's live commercial record.

Agility's recent progress strengthens an existing lead rather than changing the overall ranking. The company was already first, and the new agreements confirm that its deployment model is beginning to scale across customers and industries.

Which humanoid robotics company looks strongest in factory use cases?

Figure AI looks strongest in factory use cases because it has the richest publicly disclosed production-floor metrics from a real automotive environment.

Factory use cases create a distinct comparison frame because they require tolerance for demanding industrial conditions, strict cycle times, and consistent accuracy on a live production line. Figure AI's BMW deployment meets that standard more concretely than any competitor. The Figure 02 robots ran 10-hour weekday shifts on an active assembly line at BMW Spartanburg, accumulating 1,250+ runtime hours. They loaded over 90,000 sheet-metal parts into welding fixtures with a target cycle time of 84 seconds per task and a 99% accuracy target per shift.

Agility Robotics is the overall deployment-readiness leader, but its best proof is still logistics and warehouse handling rather than factory-floor assembly. Boston Dynamics and UBTECH are both factory-oriented, and Apptronik has credible factory partners, but none currently match Figure's level of publicly disclosed factory operating data. BMW has since started a new pilot at its Leipzig plant, this time with Hexagon's AEON robot for battery assembly, confirming BMW's broader commitment to humanoids on the production floor.

This is one of the few comparison lenses where Figure has the cleanest edge. It does not make Figure the overall deployment leader, but it does make Figure the strongest current answer for factory-specific proof.

Which humanoid robotics company looks strongest in domestic use cases?

1X looks strongest in domestic use cases because it has the clearest current evidence of actually pushing a humanoid into real household environments, even while openly admitting autonomy limits.

Domestic use cases require a very different comparison frame than factory work. Home environments are unstructured, variable, and demand safe operation around people. 1X is the most credible answer here because NEO Gamma was positioned specifically for home testing. The company discussed plans to test the robot in a few hundred to a few thousand homes in 2025, and its product page openly states that early access includes only basic autonomy, with scheduled remote expert supervision available for harder tasks. That transparency helps, because it sets realistic expectations rather than overpromising.

NEO Gamma shows useful household-oriented behaviors like object pickup, navigation, door interaction, self-charging, and app-mediated task scheduling. Figure AI has also shown domestic demos with Figure 03, including dishwasher loading and towel folding, but its primary deployment focus remains industrial. Tesla, Agility, and Boston Dynamics are all far more factory and logistics centered.

The whole domestic humanoid category still has very limited real-world proof. 1X leads this lens, but the bar itself remains low. No company has yet demonstrated robust, autonomous, scaled home deployment.

Which humanoid robotics company is the most underrated outsider today?

Kepler Robotics looks like the most underrated outsider today because its commercialization posture appears stronger than its level of market attention would suggest, especially on industrial rollout signals.

Kepler is not the overall leader, and its public evidence still relies too much on company-linked and secondary reporting to outrank Agility or Figure. But K2 Bumblebee has several features that usually show up later in the maturity curve: an explicit industrial product configuration, published endurance and payload details, a 90-day delivery lead time, reported mass production, and reported framework agreements for several thousand units. That is more than a demo narrative. It is an actual commercialization narrative.

What makes Kepler "underrated outsider" rather than simply "interesting" is the gap between narrative share and commercialization signals. Kepler gets far less attention than Tesla, Boston Dynamics, or Figure, yet its public story is unusually focused on delivery configuration and deployment economics rather than spectacle. That does not prove large-scale field success yet, but it suggests the company's real position may be stronger than the attention it receives.

1X is the runner-up for this label because it is more advanced in home-testing realism than many assume, but its own autonomy disclosures cap the case. Kepler is the stronger outsider call because its evidence points more directly toward near-term industrial commercialization.

Who is the author of this content?

NEW MARKET PITCH TEAM

We track new markets so founders and investors can move faster

We build living “market pitch” documents for emerging markets: from AI to synthetic biology and new proteins. Instead of digging through outdated PDFs, random blog posts, and hallucinated LLM answers, our clients get a clean, visual, always-updated view of what’s really happening. We map the key players, deals, regulations, metrics and signals that matter so you can decide faster whether a market is worth your time. Want to know more? Check out our about page.