On the Ground: Real-World Humanoid Deployments That Moved Past Pilots in 2025–26

On the Ground: Real-World Humanoid Deployments That Moved Past Pilots in 2025–26 Humanoid robots crossed a threshold in 2025–26: several vendors moved beyond staged demos into recurring pilots and the first limited commercial rollouts. This survey synthesizes publicly announced deployments, vendor disclosures and interviews with customers and integrators to document where humanoids actually delivered value — and where they stalled. The emphasis is on warehouses, factories, hospitals and service environments where return-on-investment is closest to realization. Use cases and industries Humanoids are being positioned primarily for tasks that require human-like dexterity, legged mobility, and straightforward integration into human workflows. The sectors seeing the most traction are: Logistics and e-commerce warehouses — piece picking, parcel handoffs, and aisle-to-station transport. Manufacturing and light assembly — kitting, screw-driving, and collaborative tasks alongside human assemblers. Healthcare and eldercare — non-clinical assistance, monitoring, mobility support and patient transport within constrained indoor spaces. Retail, hospitality and customer service — concierge, guided shopping, check-in and simple food & beverage service in public-facing environments. Notable deployments and pilot outcomes Agility Robotics — logistics trials Agility’s Digit has been the most visible logistics humanoid. Publicly announced trials in 2024–26 focused on piece-picking, tote handoffs and trailer unloading. Vendors and two anonymous logistics customers told The Humanoid Race that early pilots produced mixed but improving results: typical sustained pick-and-place rates during trials ranged between 20–60 picks per hour depending on SKU variability, with platform uptime starting near 60–70% in week-one and rising to the high 80s after software tuning and local-network hardening. Integrators (notably warehouse automation teams at systems integrators such as Honeywell Intelligrated and regional partners) were essential for WMS/robot orchestration and safety fencing adjustments. Apptronik and Figure AI — manufacturing & assembly pilots Apptronik and Figure pursued collaborative assembly in light manufacturing: screw-driving, part transfer and quality inspection. Company-funded pilots at two automotive and consumer-electronics suppliers reported cycle-time reductions in specific handoffs of 10–30% , with humanoids handling the repetitive ergonomic tasks that previously limited human throughput. Customers emphasized that the robots were most effective when used to augment fixed automation (robot arms) in flexible workcells rather than replace high-speed specialized machines. 1X Technologies and eldercare/healthcare trials Startups and healthcare-oriented vendors (including 1X Technologies and several regional integrators) staged trials in long-term care and hospital support roles: escorting patients, basic monitoring and medicine delivery within wards. Trials commonly ran under nurse supervision; reported uptime in these contexts was constrained by battery-swap procedures but regularly hit 8–12 hours of operational availability per shift after process refinement. A director of operations at a Midwestern assisted-living chain (who asked to remain anonymous) described acceptance by staff as pragmatic: “Residents like the novelty, but caregivers value reduced transfer injuries.” UBTECH / Xiaomi — retail and hospitality service rollouts UBTECH and Xiaomi focused on lower-cost humanoid/service robots for retail check-in, concierge duties and hotel room service in China and select international sites. These deployments emphasize recurring service revenue (SaaS + maintenance) and are priced to compete against stationary kiosks and wheeled robots. Retail rollouts in 2025 included multi-site pilots where units handled greeting, basic inquiry routing and low-complexity delivery between back-of-house and front desk, with vendors reporting transaction completion rates above 85% for scripted interactions. Tesla / Optimus — factory proof-of-concept integration Tesla’s Optimus demonstrations remained in the proof-of-concept stage but were notable for showing integration with an existing automotive production line. Internal demos highlighted repetitive transfers and tool handing as early tasks; Tesla did not disclose recurring billing or commercial availability in 2026. Several OEM suppliers monitored closely to judge whether the robot could meaningfully reduce ergonomic injuries in pre-assembly zones. KPI patterns, economics and TCO Across deployments, three quantitative patterns recur: Uptime improves sharply after site integration: initial uptime commonly 60–75%, rising to 85–95% with custom charging workflows, battery swaps and network reliability fixes. Task-specific KPIs matter: humanoids tend to beat humans or wheeled AMRs on tasks that require upright mobility and two-arm manipulation across variable fixtures; pick rates are lower than high-speed pick-to-light lines but higher than manual-only workflows for some SKUs. Integration costs dominate early TCO: hardware may be priced from under $50k (service models in retail) to $150k–$300k for advanced bipeds; integration and SI services frequently exceed hardware costs for first deployments. Vendor Typical use Pilot KPIs (reported/estimated) Commercial model Agility Robotics Warehouse picking & trailer tasks 20–60 picks/hr; uptime 60→90% Lease + per-robot software fees; SI integration Apptronik / Figure Collaborative assembly Cycle-time ↓10–30% on specific tasks Capex or subscription + service contracts 1X Technologies Eldercare assistance 8–12 hr availability; high staff acceptance Recurring service + monitoring fees UBTECH / Xiaomi Retail/hospitality service 85%+ scripted interaction completion Unit sales + SaaS/maintenance Tesla / Optimus Factory POC tasks Early demos; no public commercial KPIs Internal deployment roadmap Obstacles, labor relations and integration lessons Common obstacles slowed pilots or prevented scale: Safety and compliance: Certifying legged robots within existing safety frameworks required site-by-site risk assessments and new SOPs. This added weeks to pilot timelines. WMS/ERP integration: Humanoids are most valuable when tightly orchestrated with WMS; many pilots stalled until system integrators built middleware to handle tasking, retries and exception handling. Labor and union concerns: Deployments that lacked transparent ROI communication provoked labor pushback. Pilots that emphasized ergonomic injury reduction and upskilling saw fewer relations issues. Staffing and training: Successful pilots invested in 4–8 weeks of on-site training for technicians and process owners; one logistics manager told us this was “the single biggest determinant” of moving to recurring deployments. Vendors and integrators consistently said: scale comes when pilot partners accept a multi-disciplinary rollout team — production engineers, controls integrators and union liaisons — not when robots are dropped onto the floor as standalone demos. Why some pilots failed while others scaled Pilots that transitioned to recurring billing or partial headcount replacement shared common traits: clear, narrowly defined tasks; mature integration partners; and a contractual model that aligned vendor uptime/service SLAs with customer KPIs. Pilots that stalled typically tried to stretch humanoids beyond their demonstrated envelope (e.g., replacing fast mechanical pickers in high-SKU cases) or underestimated integration costs and safety-certification timelines. Competitive positioning and final assessment Vendors diverge by go-to-market strategy. Agility and Apptronik are positioned toward industrial customers with SI-heavy models; UBTECH and Xiaomi aim for high-volume, lower-margin service use-cases; Figure and Tesla emphasize long-term platform roadmaps tied to broader control stacks. For buyers, the pragmatic path to ROI is targeted substitution of repetitive, ergonomically risky human tasks rather than wholesale headcount replacement. The first wave of scaled humanoid value in 2025–26 was operational continuity, reduced injury costs and new flexible automation options — not mass labor elimination. As deployments mature, the decisive factors will be demonstrated durability in multi-shift operations, standardized integration middleware for WMS/ERP, and transparent economic models that include integration and ongoing services. Vendors and customers who learned these lessons in 2025–26 are the ones converting pilots into repeatable business.