Figure AI

Yes, but the answer depends on what you mean by buy. Five tiers of humanoid availability exist in 2026: consumer-available (1X NEO at $20,000 outright or $499/month subscription, six-month minimum); research-tools-pricing (Unitree G1 at $13,500-$16,000 and R1 at $5,900); enterprise-deployed (Figure, Apptronik, Agility under contract); consumer-promised but not shipping (Tesla Optimus); and engineering-credibility with commercial transition pending (Boston Dynamics Atlas). Which fits your need depends on whether you're a consumer, a developer, an enterprise procurement organization, or waiting on Tesla.

The Unitree R1 is the cheapest walking humanoid robot commercially available in 2026 at $5,900 base (smaller-form mass-market consumer + developer platform launched July 2025). For a full-size bipedal humanoid, the Unitree G1 starts at roughly $13,500 to $16,000 base. Both are made by Unitree Robotics (Hangzhou, China) and represent the most aggressively priced humanoid platforms commercially available; every other publicly-listed humanoid is meaningfully more expensive.

There is no single best humanoid robot in 2026; the right pick depends on which of the five availability tiers matches your use case. For consumer home use, 1X NEO is the only verified-available option ($20,000 outright or $499/month subscription with a six-month minimum; late-2026 US delivery). For research and developer access, the Unitree G1 is the practical choice. For warehouse and factory pilots, Agility Digit, Apptronik Apollo, and Figure 03 are the platforms with documented commercial deployments. For elite R&D, Boston Dynamics Atlas remains the dynamic-motion benchmark. Tesla Optimus remains consumer-promised but not for sale.

By active commercial deployment activity in 2026, the leading humanoid robot makers are Tesla (Optimus, factory pilots), Figure AI (02, BMW pilot), Agility Robotics (Digit, warehouse operators), Apptronik (Apollo, Mercedes-Benz pilot), 1X Technologies (Neo, consumer pre-launch), Boston Dynamics (Atlas, R&D), and Unitree Robotics (G1/H-series, research). A fast-growing Chinese cohort (UBTech, Xiaomi, XPeng, Fourier, EngineAI, and others) is shipping platforms at increasingly competitive price points.

There is no established lifespan benchmark for modern humanoid robots in 2026. The longest-running commercial deployments are under three years old. Component-level data exists: harmonic drives are typically rated for 20,000 to 30,000 hours of industrial duty, lithium battery packs for 1,000 to 3,000 full charge cycles, and brushless DC actuators for several years of regular use. Battery and actuator wear, not advertised lifespan, are the practical limits.

Not at labor-market scale in 2026. Verified enterprise deployments operate at pilot scope: Figure 02 at BMW Spartanburg assembled 30,000 vehicles over 11 months; Agility Digit at GXO Flowery Branch handles 100,000 totes per year; Apptronik Apollo runs 3 Fortune-500 pilots; 1X NEO performs consumer household tasks with explicit teleoperation disclosure. These are meaningful capability demonstrations but not workforce-displacement at scale. The framework reads workforce-replacement claims as long-horizon trajectory rather than near-term reality.

1X Technologies has explicitly disclosed that NEO relies on remote human teleoperators for complex household tasks during the consumer rollout phase. NEO is a teleop-bridged humanoid, not a fully autonomous one. 1X frames the teleoperation as the deliberate strategic path to consumer scale, not a stop-gap.

No. Figure AI is a US company headquartered in Sunnyvale, California, founded in 2022 by American entrepreneur Brett Adcock. The confusion likely reflects the substantial Chinese humanoid-manufacturer presence in the broader category (Unitree, AgiBot, UBTech), but Figure AI is American.

Figure 03 is the third-generation humanoid robot from Figure AI, a US humanoid manufacturer based in Sunnyvale, California. The robot stands 173 cm (about 5'8"), weighs 61 kg, and is currently in active pilot deployment at a Catalyst Brands distribution logistics center. It is not available for consumer purchase as of mid-2026.

Apptronik Apollo is a bipedal humanoid robot from Apptronik, a US humanoid maker based in Austin, Texas with NASA Valkyrie research heritage. Apollo is deployed in enterprise pilots across three Fortune-500 customers (Mercedes-Benz, GXO Logistics, Jabil) and represents the enterprise-breadth strategy in the humanoid commercial deployment landscape.

No, not yet. Boston Dynamics' Atlas humanoid is in research-to-commercial transition following the April 2024 reveal of the new electric Atlas platform. The company's quadruped Spot is the commercially-verified product line; Atlas commercial deployment timeline has not been announced. Boston Dynamics is owned by Hyundai Motor Group, which positions Atlas for industrial/enterprise rather than consumer markets.

Unitree's G1 humanoid robot starts at roughly $13,500 to $16,000 for a base research-grade configuration, with more advanced versions running higher. The smaller R1 starts at $5,900 for an entry consumer and developer configuration. Both are made by Unitree Robotics, a Chinese manufacturer based in Hangzhou, and represent the most aggressively priced humanoid platforms commercially available in 2026.

Mentee Robotics is an Israeli humanoid robot maker founded in 2022 by Amnon Shashua, the AI researcher who also leads Mobileye and AI21. Mobileye acquired Mentee in January 2026 for approximately $900 million ($612M cash plus Mobileye Class A shares), making Mentee an independent operating unit inside Mobileye. The company's MenteeBot humanoid is positioned as an AI-first general-purpose platform leveraging Mobileye's autonomous-vehicle perception heritage.

If you want to buy a humanoid robot in 2026, the answer depends on which tier of availability matches your need. 1X NEO is the only consumer-available option ($20,000 outright or $499/month subscription with six-month minimum, late-2026 US delivery). Figure 03 is enterprise-deployed only (Catalyst Brands pilot; no consumer commerce). Tesla Optimus is consumer-promised but not shipping (Musk's $20,000-$30,000 target is a forward claim, no order channel). The three products operate at three structurally distinct tiers.

The major American humanoid makers are Figure AI (Sunnyvale CA), Apptronik (Austin TX), Tesla (Palo Alto CA / Austin TX), Boston Dynamics (Waltham MA; Hyundai-owned), and Agility Robotics (Salem OR). The major Chinese makers are Unitree Robotics (Hangzhou), AgiBot (Shanghai), UBTech (Shenzhen), Fourier Intelligence (Shanghai), XPeng Robotics (Guangzhou), and several others. 1X Technologies operates a Norwegian-American structure (Moss Norway HQ plus Hayward California factory). Mentee Robotics is Israeli (acquired by Mobileye January 2026). Sanctuary AI is Canadian (Vancouver).

The major frontier AI research labs are committing institutionally to robotics in 2025-2026 as a distinct industry cluster. OpenAI formally relaunched its robotics division in May 2026 after a 4-to-5-year hiatus. Google DeepMind operates the Gemini Robotics program as part of its broader model-to-embodiment research. Meta has expanded AI robotics work alongside its Reality Labs portfolio. The cluster is not about any one lab making a robot; it is about foundation-model research lines extending toward physical-world embodiment.

Every major humanoid manufacturer uses teleoperation in development and demonstration. The differential across the cohort is the disclosure layer, not the underlying practice. 1X is the most transparent (explicit teleop disclosure on the consumer commerce surface). Tesla operated framing-without-disclosure at We Robot 2024 (autonomy framing; subsequently confirmed teleoperated). Figure deploys with human-in-loop for exception handling at customer facilities. Apptronik has mixed disclosure across enterprise pilots. The framework treats the disclosure differential as the editorial finding.

Sanctuary AI is a Canadian humanoid robotics company headquartered in Vancouver, British Columbia, founded by Geordie Rose and Suzanne Gildert (both Kindred AI alumni). The company's Phoenix platform is a seventh-generation humanoid emphasizing a cognitive architecture that combines symbolic reasoning with neural learning, structurally distinct from the end-to-end foundation-model approach most US humanoid makers pursue. Sanctuary AI is privately held; not publicly traded.

UBTech Robotics is a publicly-traded Chinese humanoid manufacturer headquartered in Shenzhen, listed on the Hong Kong Stock Exchange. The Walker S2 is the company's industrial-focused humanoid platform deployed in factory pilots with BYD, Geely, Foxconn, and other manufacturing customers. UBTech is real, with verified commercial pilots; the company distinguishes itself from the US private humanoid cohort by being publicly listed with disclosed financial state.

Direct equity exposure to humanoid manufacturers in 2026 is mostly limited to a small set of publicly-traded companies: UBTech Robotics (HKEX-listed Chinese maker), Tesla (NASDAQ; Optimus is one product), and Hyundai Motor Group (KRX; Boston Dynamics parent). The major US humanoid pure-plays (Figure AI, 1X Technologies, Apptronik, Agility Robotics, Sanctuary AI) are privately held and accessible only via venture-stage or accredited-investor channels. Mentee Robotics was acquired by Mobileye in January 2026. DEPLOY is not an investment advisor; this guide documents verification posture, not investment advice.

No consumer-deployed humanoid robot can cook autonomously in 2026. Demonstrations and research exist across the cohort; consumer-grade cooking capability is not verified at deployment scale. Purpose-built robotic kitchens (such as Moley Robotics, a non-humanoid installed kitchen) exist at the research-and-experimental tier near $300,000. For a general-purpose humanoid robot that can prepare meals autonomously in a customer's kitchen, the answer is years from verified consumer-deployment.

Yes, with disclosure. 1X NEO is the verified-leader for consumer-deployment laundry capability in 2026: NEO performs laundry tasks (folding, sorting, light loading) in consumer homes with explicit Expert Mode teleoperation disclosure for complex tasks. Other cohort manufacturers (Tesla Optimus, Figure 03, Apptronik Apollo, Boston Dynamics Atlas, Unitree) demonstrate clothes-folding or related manipulation but do not consumer-deploy laundry capability. Laundry is the canonical example of how task capability varies dramatically across the humanoid cohort: 1X delivers verified consumer capability with teleop disclosure; others deliver demonstrations without consumer deployment.

Humanoid robot capability in 2026 sorts into four verification tiers per DEPLOY's framework. Verified consumer-deployed: 1X NEO performs laundry, organizing, and light manipulation in customer homes with explicit teleop disclosure. Verified enterprise-deployed: Figure 03, Apptronik Apollo, Agility Digit, and UBTech Walker S2 perform manufacturing and logistics tasks at Fortune-500 customer facilities. Research and demonstration: Boston Dynamics Atlas, Tesla Optimus, and Unitree platforms show capability footage but do not deploy. Claimed future: cooking, autonomous home assistance, childcare, and general-purpose household work remain claimed across the cohort but not consumer-deployed.

Agility Robotics is a US humanoid robotics company headquartered in Albany, Oregon, with humanoid R&D heritage tracing through the Cassie research platform that preceded Digit. The Digit humanoid is the company's commercial warehouse and logistics platform, deployed in pilots at GXO Logistics (the 100,000-tote scaled-throughput anchor), Amazon Spanx Tennessee, Schaeffler, and other industrial customers. Agility is privately held, with Amazon as a strategic investor and a manufacturing facility (RoboFab) in Salem, Oregon.

Fourier Intelligence is a Chinese humanoid robotics company headquartered in Shanghai with distinctive technical heritage in lower-limb exoskeleton and rehabilitation robotics. The GR-3 is the company's third-generation general-purpose humanoid platform, evolved from the GR-1 and GR-2 predecessors. Fourier occupies a distinctive position in the Chinese humanoid cohort: medical and rehabilitation engineering heritage transitioning to general-purpose humanoid product, with research-platform commercial positioning rather than the factory-deployment focus of UBTech Walker S2 or the research-tools pricing of Unitree G1 and R1.

PAL Robotics is a Spanish humanoid robotics company headquartered in Barcelona, founded in 2004 with extensive European research-consortium history. The TALOS is the company's full-size adult bipedal humanoid platform, positioned for research-institution deployment rather than consumer or scaled-enterprise commercial use. PAL extends DEPLOY's humanoid manufacturer cohort to European context, representing a distinct geographic-and-strategic position from the American, Chinese, and Canadian cohort members.

Humanoid robots are bipedal robots with arms, hands, and roughly human-like proportions designed to operate in human environments and perform general-purpose tasks (Tesla Optimus, 1X NEO, Figure 03, Apptronik Apollo, Boston Dynamics Atlas, Unitree G1). Industrial robots are fixed-base manipulator arms designed for specific factory-automation tasks at high precision and reliability (FANUC, Universal Robots, KUKA, ABB). The categories share the word 'robot' but operate at substantively different scales (industrial robotics is a mature commercial category with hundreds of thousands of installed units; humanoid robotics is an emerging category with consumer-deployment at single-manufacturer scale).

Physical AI refers to AI systems that operate in the physical world rather than purely in digital environments. The category spans autonomous vehicles, humanoid robots, autonomous mobile robots in warehouses, drones, and AI-augmented industrial automation. Physical AI differs from digital-only AI in that the system must perceive, decide, and act under physical-world constraints (sensor noise, latency, mechanical failure modes, regulatory frameworks). DEPLOY tracks physical AI across four subcategories with distinct verification frameworks per category.

A foundation model for robotics extends the large-language-model paradigm to physical-action prediction. Trained on robot demonstrations rather than internet text alone, these models output action sequences (motor commands, manipulator trajectories) rather than text tokens. The category is dominated by vision-language-action (VLA) architectures that take camera images plus optional language instructions as input and produce action tokens as output. Companies building these models constitute the brain-provider tier of the robotics value chain, distinct from the humanoid OEM tier that builds the physical platforms the models run on.

Skild AI is a US foundation-model-for-robotics company headquartered in Pittsburgh, Pennsylvania, founded by Deepak Pathak and Abhinav Gupta (both Carnegie Mellon University robotics-research alumni). The company's Skild Brain product is a vision-language-action (VLA) foundation model architected around a cross-platform general-purpose thesis: a single model trained to operate across multiple robot platforms rather than platform-specific brains. Skild has raised substantial 2024-2025 funding rounds; the company is privately held and represents a distinctive position in the brain-provider tier of the robotics value chain.

The brain-provider tier of robotics in 2026 includes several distinct strategic theses. Skild AI pursues cross-platform general-purpose brain deployment. Physical Intelligence (Pi-0; Pi-0.5) emphasizes transformer-based VLA research publications. Covariant specializes in warehouse-automation foundation models. Google DeepMind operates Gemini Robotics and RT-2 across AV and humanoid research. OpenAI Robotics relaunched in May 2026 after a 2021 hiatus. NVIDIA Project GR00T pursues cross-platform humanoid integration aligned with NVIDIA's broader stack. Meta operates research-publication-emphasizing work via FAIR and Reality Labs. The cohort is at research-and-demonstration verification depth; commercial-scale deployment lags behind humanoid OEM commercial deployment substantially.

Robotics value chain operates across three structural tiers. Brain-provider tier companies (Skild AI, Physical Intelligence, Covariant, Google DeepMind, OpenAI Robotics, NVIDIA Project GR00T) build foundation models for robotics without making hardware. OEM-platform tier companies (Figure AI, Apptronik, 1X Technologies, Tesla, Agility Robotics, Boston Dynamics, Unitree, UBTech) build robot hardware platforms with integrated brains. Deployment tier represents real-world operation at customer facilities (BMW Spartanburg, GXO Flowery Branch, Mercedes-Benz pilots). The three tiers operate complementarily; understanding which tier a company occupies is essential for evaluating its competitive position and verification posture.

The Figure 02 humanoid robot has operated in production at BMW Group's Spartanburg, South Carolina plant since August 2024, contributing to the assembly of more than 30,000 BMW X3 vehicles over an 11-month deployment that accumulated 1,250 hours of runtime handling more than 90,000 parts in chassis-assembly tasks. The deployment is the canonical commercial humanoid manufacturing reference for the cohort, with end-product OEM acceptance verification distinguishing it from peer humanoid deployments. BMW Group has announced expansion to its Plant Leipzig facility as the second humanoid production site. Figure's current-generation Figure 03 is deployed at Catalyst Brands Reno, not at BMW; BMW Spartanburg remains the Figure 02 deployment anchor as of mid-2026.

The Figure 03 humanoid is deployed at Catalyst Brands' Reno, Nevada distribution logistics center in a pilot operation that represents Figure AI's verified non-manufacturing deployment context. Catalyst Brands is the corporate parent of Forever 21, Brooks Brothers, Aeropostale, Lucky Brand, Nautica, and other retail brands under the SPARC Group restructuring. The Reno deployment is Figure's logistics-and-distribution operational reference, complementing the Figure 02 BMW Spartanburg manufacturing reference and demonstrating Figure's dual-vertical commercial strategy across two structurally distinct deployment contexts.

Apptronik's Apollo humanoid is deployed at Mercedes-Benz's Berlin-Marienfelde manufacturing facility in a pilot operation that represents Apollo's premium-segment automotive verification reference. The Mercedes-Benz partnership is one of three Apptronik enterprise customer relationships (alongside GXO Logistics and Jabil) that together establish Apollo's three-customer enterprise-deployment breadth strategy, structurally distinct from Figure AI's dual-vertical positioning and Agility's single-vertical depth specialization.

Boston Dynamics' Atlas humanoid is deployed at Hyundai Metaplant America's Savannah, Georgia EV manufacturing facility in a pilot operation that represents Atlas's verified enterprise customer relationship. The deployment is structurally distinguished by Hyundai's corporate-parent relationship with Boston Dynamics (Hyundai Motor Group acquired Boston Dynamics from SoftBank in 2021), creating a maker-customer relationship that differs from the arms-length customer relationships of cohort peers Figure-BMW, Apptronik-Mercedes, and Agility-GXO.

Aggregator coverage frequently frames the deployment narrative as 'Figure 03 deployed at BMW Plant Spartanburg + Leipzig.' Per primary-source verification: the BMW Spartanburg deployment was Figure 02 (the previous-generation humanoid; completed 30,000 BMW X3 vehicles + 1,250 hours runtime over 11-month deployment from August 2024). BMW Leipzig went to Hexagon AEON, NOT Figure. The current-generation Figure 03 is deployed at Catalyst Brands Reno Distribution Logistics Center per Figure's May 2026 partnership announcement. The aggregator framing collapses three distinct facts (Figure 02 at Spartanburg; Hexagon AEON at Leipzig; Figure 03 at Catalyst Brands Reno) into one wrong claim attributing both BMW deployments to Figure 03. Per [DEPLOY's framework discipline](/explainers/how-deploy-verifies), audit-first verification + aggregator-drift rejection produced the corrected attribution. This piece documents the catch as framework-in-action worked example: deployment-attribution discipline + generation-distinction discipline + cross-customer-conflation rejection operating at editorial-anchor depth.

DEPLOY thinks about robot insurance as a four-dimension actuarial framework operating recursively across the verified-vs-claimed throughline: deployment-incident-recall actuarial depth (61 verified incidents at primary-source-anchored severity + root-cause + regulatory-action depth; exposure denominators absent at most deployments); manufacturer financial-state / counterparty risk (114 investors + 58 funding rounds + 29 acquisitions verified; financial state vs relationship state distinction); supply-chain component failure analysis (absent as structured substrate; bounded to safety-critical components when authored); regulatory clearance per jurisdiction (34 verified filings lopsided 94% US-FDA-only; jurisdictional completeness is the load-bearing gating layer for insurability per region). The discipline that distinguishes DEPLOY's framework from the broader insurance-discourse cohort: honest 'insurability unknown for this region / no exposure data' is more valuable than a fabricated rate. Cap-flag-as-trust-signal operates recursively on actuarial framing.

DEPLOY tracks partnership lifecycle state as a four-state framework operating at relationship-graph granularity: announced (verified-from-press-release; not-yet-active) → active (verified-deployed or verified-shipped; current_status=true) → dissolved (terminated; current_status=false with end_date populated) → unverified-current-state (announced-but-no-update-since; cap-flag honest-absence). Per Agent A's Arc C substrate, 18 partnerships + 32 parties + 14 external counterparties (via XOR pattern) + NVIDIA 4-counterparty multi-party-partnership node verified at primary-source-anchored depth. The canonical lifecycle worked example: Figure × OpenAI announced 2024 → dissolved February 2025 (status=dissolved + endDate populated). The external-name XOR pattern (partnership_parties.company_id when counterparty is tracked entity; external_name when counterparty is genuinely not-in-registry; OpenAI + Uber + Nissan + Microsoft as canonical worked examples) operates as verification-posture discipline at relationship-graph granularity. Cap-flag-as-trust-signal operates recursively on partnership framing.

DEPLOY tracks Phase 3 Dim 2 manufacturer counterparty risk substrate at primary-source-anchored verification depth per Agent A's mfs=7 manufacturer_financial_states baseline backfill: Intuitive Surgical FY2024 $8.35B (sec_10k; mature; counterparty=low) + Stryker nine-month 2024 $16.16B (sec_10q; mature; counterparty=low; SEC-verified-figure-with-narrower-precision-cap-flag CANONICAL worked example) + Zimmer Biomet FY2024 $7.68B (sec_8k; mature; counterparty=low) + Boston Dynamics private_reported NULL revenue (mature; counterparty=low via Hyundai ~80% parent-backing CANONICAL worked example for counterparty-risk-by-parent-backing classification) + Apptronik private_reported NULL (growth; counterparty=moderate; recent $350M Series A Feb 2025) + Figure AI private_reported NULL (growth; counterparty=moderate; ~$1B Series C 2025 at ~$39B reported valuation) + 1X Technologies private_reported NULL (growth; counterparty=moderate; 2024 Series B ~$100M EQT-led + subsequent 2025 round). Cap-flag-as-trust-signal recursive at three substantive layers: SEC-verified-figure-with-narrower-precision-cap-flag pattern (Stryker Q3 nine-month $16,159M used instead of unverified FY extrapolation; verified-vs-claimed at sub-fiscal-period granularity) + honest-absence on private-financials USD fields (Apptronik + Figure + 1X all NULL revenue at honest-absence cap-flag pending primary-source disclosure) + cash_runway_basis NULL when cash_runway_months NULL across the board (validator-aware discipline; cash_runway_months not publicly-disclosed for any of these makers; cash_runway_basis stays NULL preserving validator integrity). Counterparty-risk-by-parent-backing classification discipline as canonical: Boston Dynamics counterparty=low via Hyundai ~80% subsidiary parent-backing vs Apptronik/Figure/1X counterparty=moderate as growth-stage privates without parent-backing.