Digit vs Apollo vs Optimus: An Enterprise Buyer's Guide to Choosing the Right Humanoid Robot Platform
- Ahtesham Shaikh
- 3 days ago
- 24 min read
Physical AI Journal Research Team · Independent analysis by Sekason Research Limited, United Kingdom
Published July 2026 · physicalaijournal.org
Executive Summary
Only one of the three leading enterprise humanoid robot platforms — Digit by Agility Robotics — carries confirmed commercial revenue contracts as of mid-2026; Apollo (Apptronik) remains at the pilot stage, and Optimus (Tesla) has no external commercial availability. Automation and Procurement Buyers must therefore make a platform selection across three materially different stages of commercial readiness, not between three equivalent options.
This report evaluates all three through the lens of procurement — comparing deployment maturity, commercial availability, operating model, vendor financial position, integration complexity, and return on investment — to support the specific decision of whether to deploy now, delay adoption, or pursue a phased pilot strategy.
The evidence base is narrow but real. The International Federation of Robotics recorded approximately 410 full-humanoid unit shipments globally in 2025. Agility Robotics reports Digit has accumulated over 65,000 operational hours across nine customer sites (company-claimed), with confirmed live deployments at GXO Logistics and Toyota Motor Manufacturing Canada. Apollo's announced pilot with Mercedes-Benz has not published operational outcomes. The report concludes with a Buyer Decision Matrix, ROI cost framework, and use-case-segmented platform recommendations.
1. Executive Intelligence Synthesis
Quick Answer — Which humanoid robot platform is currently the most defensible enterprise investment?
Digit by Agility Robotics is the only platform with confirmed commercial revenue deployments as of mid-2026, including live operations at GXO Logistics and Toyota Motor Manufacturing Canada. Apollo has announced pilots but published no outcome data. Optimus is in internal Tesla testing with no external commercial availability. For buyers requiring evidence-based procurement, Digit currently carries the lowest deployment risk — though all three platforms carry material operational and financial uncertainties that buyers must assess independently.
Signal 1 — Commercial deployment is now measurable, not hypothetical
Agility Robotics reports Digit has accumulated 65,000+ operational hours across nine customer sites as of mid-2026 (company-claimed). The International Federation of Robotics recorded approximately 410 full-humanoid unit shipments globally in 2025 — a real but small number. Evidence of commercial deployment now exists; evidence of commercial scale does not.
Signal 2 — Deployment maturity differs dramatically across vendors
Digit has moved from pilot to revenue-generating RaaS contracts with named enterprise customers. Apollo has announced manufacturing pilots — with Mercedes-Benz and GXO — but has not published operational outcome data. Optimus has no external commercial deployments; Tesla's production targets for Optimus have been revised repeatedly since the platform was announced (company-claimed timelines). Buyers should treat each platform as occupying a distinct stage of commercial readiness.
Signal 3 — Enterprise procurement risk remains higher than technology risk
Current humanoid deployments operate in tightly controlled, single-task environments. Integration complexity, infrastructure modification costs, and the absence of global safety standards for human-collaborative robots represent procurement risks that are not addressed by platform specification sheets. Industry analysis identifies 95%+ uptime — the standard expected by enterprise operations — as a threshold that has not been demonstrated at scale by any vendor.
Signal 4 — ROI depends more on workflow selection than robot specifications
No independent deployment has demonstrated profitability at enterprise scale as of mid-2026. Platforms currently in commercial operation handle highly repetitive, physically bounded tasks — tote handling, assembly kit delivery — where workflow predictability minimises autonomy failure risk. Buyers who deploy humanoids in complex, variable-task environments before operational frameworks mature face elevated cost-overrun exposure.
Signal 5 — Buyers should evaluate deployment readiness, not prototype capability
Apptronik raised $935M in its Series A (February 2026). Agility Robotics is pursuing a SPAC merger at a $2.5B pre-money valuation to raise over $620M. Capital raised is not the same as commercial readiness. Buyers should assess vendor-supplied evidence of operational deployments — not product launches, prototype demonstrations, or capital announcements — before committing procurement budget.
Directional statement: For warehouse logistics with bounded, repetitive tasks, a structured Digit pilot is supportable on current evidence. For manufacturing environments requiring flexible task handling, a monitor-and-evaluate posture is appropriate for both Apollo and Optimus until outcome data from announced pilots is published.
2. The Enterprise Humanoid Robot Market in 2026
Quick Answer — What are the leading enterprise humanoid robot platforms in 2026?
The three most commercially advanced enterprise humanoid platforms in 2026 are Digit (Agility Robotics), Apollo (Apptronik), and Optimus (Tesla). Digit is the only platform with confirmed commercial revenue contracts and reported operational hours. Apollo has entered pilot agreements with major manufacturers. Optimus remains in internal development with no external commercial availability. The Silicon Valley Robotics Center identifies approximately 12 commercial humanoid platforms (purchase or lease) globally in 2026, up from 3 in 2024.
The International Federation of Robotics reported full-humanoid shipments of approximately 410 units in 2025 — a figure that places the entire sector at early-adoption stage, not at scale. The Silicon Valley Robotics Center's State of Robotics 2026 report documents approximately 12 commercial humanoid platforms available for purchase or lease globally, up from 3 in 2024. The pace of platform creation has outpaced the pace of deployment evidence.
Three platforms dominate enterprise procurement conversations in Western markets: Agility Robotics' Digit, Apptronik's Apollo, and Tesla's Optimus. Their commercial positions are materially different. Agility Robotics has signed multi-year Robots-as-a-Service contracts with named enterprise customers and reports $300M+ in committed Digit v5 orders (company-claimed). Apptronik has announced pilot agreements and closed a $935M Series A in February 2026, but has not disclosed commercial revenue figures. Tesla has made no external commercial availability announcement for Optimus; the platform is currently used in-house for R&D.
The competitive landscape is not limited to these three vendors. China's government has committed over $20B to humanoid robotics development, with domestic producers including AgiBot and Unitree receiving state support alongside aspirations of 1 million units per year by 2030 (company- and government-claimed). Western industrial buyers sourcing from Chinese platforms face a separate risk and supply-chain evaluation not covered in this report. Leading Western adopters span logistics, warehousing, and manufacturing: GXO Logistics, Toyota Motor Manufacturing Canada, Schaeffler, Mercado Libre, and Amazon (R&D) have all been named in connection with humanoid deployments in the research period.
Platform Landscape Overview — Mid-2026 | |||||
Vendor | Platform | Commercial Status | Primary Target Market | Pricing Model | Verification Status |
Agility Robotics | Digit | Commercial RaaS contracts in operation | Warehouse logistics, automotive manufacturing | RaaS ~$8,500/month (company-claimed) | VERIFIED — Revenue contracts confirmed |
Apptronik | Apollo | Pilot agreements announced; no outcome data published | Automotive manufacturing, general industry | Not published | COMPANY-CLAIMED — Pilot announcements only |
Tesla | Optimus | Internal development; no external commercial availability | General manufacturing (stated target) | ~$30,000/unit at launch; ~$20,000 at scale (company-claimed) | COMPANY-CLAIMED — No external deployment |
Platform Specifications
Digit stands at approximately 1.75 m tall, weighs 65–76 kg, and carries a payload of approximately 16 kg per arm at a walking speed of approximately 5 km/h (all specifications company-claimed). Battery runtime for the v4 generation was approximately 4 hours; independent testing has corroborated a category-wide sustained runtime in the 1–4 hour range at current technology maturity. Digit v4 deployments have operated with physical separation or fencing in place; Agility Robotics states that v5 is designed to operate without this requirement (company-claimed). Buyers should confirm site-specific configuration requirements directly with Agility Robotics and their own safety engineers.
Apollo stands at approximately 1.73 m, weighs approximately 72 kg, and carries a payload of approximately 25 kg (all specifications company-claimed). It operates on swappable batteries and is offered in both bipedal and wheeled-base variants; the wheeled variant is positioned to align with existing industrial robot safety frameworks. Apptronik has not published official pricing or lease terms as of this report date.
Optimus is planned at approximately 1.73 m, 57 kg, and a payload of approximately 20 kg (all specifications company-claimed from 2021–2022 announcements). Battery runtime and operational speed have not been disclosed. Tesla has indicated a target launch price of approximately $30,000 per unit, declining toward approximately $20,000 at production scale — both figures are company-claimed projections with no confirmed availability timeline. No independently verified specification sheet has been published.
3. Real Deployments vs Demonstrations: Evaluating the Evidence
Quick Answer — Which humanoid robots are working inside confirmed commercial operations?
As of mid-2026, Digit by Agility Robotics is the only platform with confirmed, revenue-generating commercial deployments. These include a live warehouse operation at GXO Logistics in Georgia, USA, and a commercial RaaS agreement with Toyota Motor Manufacturing Canada active from April 2026. Apollo by Apptronik has announced a pilot with Mercedes-Benz and a research programme with GXO, but no operational outcomes have been reported. Optimus has no external commercial deployments. All deployment figures cited by vendors should be treated as company-claimed unless independently confirmed. The most significant analytical gap in current humanoid robotics coverage is the conflation of announcement with deployment, and deployment with commercial operation.
For procurement purposes, four categories are materially distinct:
(1) announcement of intent,
(2) prototype demonstration,
(3) time-limited pilot, and
(4) commercial operation under a revenue-generating contract. Only the fourth category provides evidence relevant to a procurement decision.
Applying these categories to the three platforms under review produces a clear stratification. Digit has reached Category 4 with multiple named customers. Apollo has reached Category 3 — announced pilots — but has not published Category 4 evidence. Optimus remains at Category 2 for external purposes; internal Tesla use does not constitute external commercial deployment.
Deployment Outcome Tracker — Digit, Apollo, Optimus — Mid-2026 | |||||
Vendor / Platform | Customer | Country | Deployment Category | Reported Outcome | Verification Status |
Digit / Agility Robotics | GXO Logistics | USA | Commercial RaaS (live) | 100,000+ totes moved (company-claimed); "fully operational... in a live warehouse environment" (GXO CTO, joint press release) | COMPANY-CLAIMED — Joint press release Jun 2024 (GXO / Agility Robotics); no independent Tier 1/2 source confirms contract terms |
Digit / Agility Robotics | Toyota Motor Mfg Canada | Canada | Commercial RaaS (live from Apr 2026) | 7-unit deployment for repetitive tote handling; outcomes pending | VERIFIED — Contract confirmed Feb 2026; reported by Automotive Manufacturing Solutions (Tier 2) |
Digit / Agility Robotics | Mercado Libre | USA / LatAm | Announced commercial agreement (2025) | Commerce fulfilment tasks; no outcome data published | COMPANY-CLAIMED — Announcement only; no independent confirmation of operation |
Apollo / Apptronik | Mercedes-Benz | Germany | Pilot announced (Mar 2024) | Assembly kit supply and component inspection; no operational outcomes reported | COMPANY-CLAIMED — Announcement confirmed; no outcome data |
Apollo / Apptronik | GXO (R&D programme) | USA | R&D collaboration announced | No usage figures public | COMPANY-CLAIMED — Announcement only |
Optimus / Tesla | Tesla (internal) | USA | Internal R&D use | No external commercial deployment | ESTIMATED — Based on public statements; no independent confirmation |
Sources: GXO / Agility Robotics joint press release Jun 2024; Automotive Manufacturing Solutions Feb 2026; Apptronik press release Mar 2024. All outcome figures from Agility Robotics are company-claimed unless otherwise noted.
Agility Robotics reports that Digit has accumulated over 65,000 operational hours across nine customer sites as of mid-2026 (company-claimed). This figure has not been independently audited. The GXO deployment is the most documented: Adrian Stoch, Chief Automation Officer at GXO, confirmed "fully operational Digit humanoids in a live warehouse environment" in the June 2024 joint press release with
Agility Robotics. The tote-moved figure of 100,000+ originates with Agility Robotics — GXO has not published an independent operational review.
The Toyota Motor Manufacturing Canada deployment is the most commercially significant recent development. Following a year-long pilot with 3 units, TMMC converted to a RaaS agreement for 7 units commencing April 2026, as reported by Automotive Manufacturing Solutions. This represents the first confirmed commercial humanoid deployment in Canadian automotive manufacturing. Operational outcomes from the expanded deployment are not yet public.
For Apollo, the Mercedes-Benz pilot announced in March 2024 — covering assembly kit supply and component inspection — has not published outcome data in the period covered by this report. Apptronik's June 2026 launch of Robot Park in Austin, TX is a data-collection and development facility, not a revenue-generating customer deployment. Buyers evaluating Apollo are making a forward commitment on announced capability, not demonstrated commercial performance.
4. Economics & ROI Analysis: What Does a Humanoid Robot Actually Cost to Operate?
Quick Answer — Is a humanoid robot financially viable for warehouse and manufacturing operations today?
Financial viability depends entirely on workflow selection and cost structure. At Digit's reported RaaS rate of approximately $8,500/month (company-claimed), the annual platform cost is approximately $102,000 per robot. A single unit must replace more than one worker's output, operate across multiple shifts, or address tasks where the cost premium is justified by operational benefit — none of which has been independently demonstrated at scale. No independent deployment has demonstrated profitability at enterprise scale as of mid-2026. Buyers should treat ROI projections as estimates pending operational evidence.
CapEx vs RaaS: Choosing the Right Commercial Structure
The dominant commercial model for humanoid robots in 2026 is Robots-as-a-Service. Agility Robotics has not published a unit sale price for Digit v5; the only known historical sale price was approximately $250,000 per unit in 2020 (company-claimed, now superseded). Its June 2026 investor deck projects a RaaS rate of approximately $8,500 per month per robot, equivalent to approximately $102,000 per year (company-claimed; sourced from SEC filing). The Silicon Valley Robotics Center's State of Robotics 2026 report documents full-humanoid platform prices in the range of approximately $85,000–$245,000 for purchase, or approximately $3,500–$8,000 per month for lease, across the market (verified, Tier 2).
Apptronik has not announced pricing for Apollo. Tesla has stated a target retail price of approximately $30,000 for Optimus at initial commercial launch, declining toward approximately $20,000 at production scale — both figures are company-claimed projections with no confirmed timeline for external availability.
RaaS & Pricing Comparison — Mid-2026 | ||||
Platform | RaaS / Lease Rate | Purchase Price | Annual Platform Cost (RaaS) | Verification Status |
Digit | ~$8,500/month (company-claimed) | Not published (historical: ~$250k in 2020, company-claimed) | ~$102,000/year (company-claimed) | COMPANY-CLAIMED — SEC investor deck, Jun 2026 |
Apollo | Not published | Not published | Not available | ESTIMATED — No pricing disclosed |
Optimus | Not applicable (no external availability) | ~$30,000 at launch; ~$20,000 at scale (company-claimed) | Not calculable | COMPANY-CLAIMED — CEO statements; no confirmed availability |
Market range (SVRC) | ~$3,500–$8,000/month | ~$85,000–$245,000 | ~$42,000–$96,000/year | VERIFIED — Silicon Valley Robotics Center, State of Robotics 2026 |
Labour Economics: Where the ROI Case Is Built
The economic case for humanoid robots depends on labour cost displacement relative to total platform cost. At Digit's reported RaaS rate of approximately $102,000 per year (company-claimed), a single robot must either replace more than one worker, operate across multiple shifts, or address tasks with cost-per-unit economics that make the premium acceptable. Peggy Johnson, CEO of Agility Robotics, has cited a total automatable labour market of approximately $1 trillion (company-claimed) as the addressable opportunity. This figure represents market framing, not a deployment ROI calculation.
The genuine ROI variable for enterprise buyers is: at what utilisation rate, over what number of shifts, across what task set, does a $102,000/year platform cost become cost-neutral or cost-advantaged relative to fully-loaded human labour for that specific task? No published independent study has answered this question for any of the three platforms reviewed. Buyers must build their own business cases using site-specific labour cost data, which will vary materially by geography, role, and operational model.
Total Cost of Ownership
Platform cost is a single line in the total cost of ownership. Deployment budgets should also account for infrastructure modification costs (workspace adaptation, charging infrastructure, safety fencing where required for Digit v4 configurations), integration costs (software, fleet management systems, API connectivity to existing warehouse management systems), training costs (operator and maintenance personnel), downtime cost (current platform uptime falls short of the 95%+ standard expected in continuous operations), and support contract costs. None of these ancillary costs are publicly quantified by any of the three vendors.
Deployment Cost Matrix — Indicative Pilot Budget (Single Unit, 12 Months) | |||
Cost Element | Indicative Range | Notes | Verification Status |
Platform cost (RaaS) | ~$102,000/year | Digit v5 (company-claimed) | COMPANY-CLAIMED |
Infrastructure modification | $20,000–$100,000+ | Charging, fencing, floor markings; varies by site | ESTIMATED — No vendor disclosure |
Integration (WMS/IT) | $15,000–$60,000 | API integration, fleet management software | ESTIMATED — No vendor disclosure |
Operator training | $5,000–$20,000 | Initial and ongoing | ESTIMATED — No vendor disclosure |
Support / maintenance | Included in some RaaS; unknown in others | Clarify with vendor before signing | ESTIMATED — Not publicly disclosed |
Indicative Year 1 Total | $142,000–$282,000+ | Per single unit; excludes downtime cost | ESTIMATED — Composite of above elements |
These figures are estimates for planning purposes. Buyers must obtain itemised quotes from vendors and integration partners.
Break-Even Analysis
At an indicative Year 1 total cost of $142,000–$282,000 per unit (estimated), break-even requires either multi-shift utilisation replacing more than one equivalent headcount, or cost-per-unit economics that favour the platform over human labour for the specific task — neither of which has been independently demonstrated by any vendor. Current operational evidence from the GXO deployment does not include a published throughput-per-hour comparison against human workers. Buyers should not assume break-even within a standard 3-year RaaS term on current evidence.
The ROI case improves as fleet size increases — fixed integration and infrastructure costs spread across more units — and as platform reliability data accumulates. A phased pilot of 1–3 units to establish operational parameters before committing to a fleet contract is the most defensible procurement approach at this stage of market maturity.
5. Named Deployment Case Studies
Quick Answer — Which enterprises have deployed humanoid robots in confirmed operational settings?
Three enterprise deployments have progressed beyond announcement to confirmed commercial or pilot operation as of mid-2026.
GXO Logistics in the USA operates Digit robots for warehouse tote handling under a commercial RaaS contract (confirmed by joint press release).
Toyota Motor Manufacturing Canada has converted a 3-unit pilot to a commercial deployment of 7 Digit units from April 2026 (independently reported).
Mercedes-Benz is piloting Apollo in German factories for assembly kit supply — the first known pilot for that platform — but has not published operational outcomes. In all three cases, performance figures originate with vendors, not independent audits.
Case Study 1 — GXO Logistics: Warehouse Tote Handling, USA
GXO Logistics, one of the world's largest pure-play contract logistics providers, signed what was described as the industry's first commercial humanoid robot deployment agreement with Agility Robotics in June 2024. The deployment is located at GXO's Flowery Branch, Georgia fulfilment facility. A late-2023 pilot integrated Digit with existing Automated Mobile Robots on the warehouse floor; the primary task is loading and unloading totes between collaborative robots and conveyors.
Adrian Stoch, Chief Automation Officer at GXO, stated in the joint press release:
"We're building on the success of last year's groundbreaking pilot with Agility by deploying fully operational Digit humanoids into a live warehouse environment... Digit is the perfect addition to work alongside our people in our fulfillment center."
Agility Robotics claims that Digit has moved over 100,000 totes at the GXO facility (company-claimed, June 2025). GXO has not published an independent operational review or throughput comparison.
Procurement relevance: This deployment establishes that Digit can operate in a live logistics environment alongside humans and AMRs under a commercial contract. The absence of published operational performance data means buyers cannot extract specific throughput or uptime benchmarks from this case for their own business cases.
Case Study 2 — Toyota Motor Manufacturing Canada: Automotive Assembly, Canada
Toyota Motor Manufacturing Canada operates an automotive assembly plant in Woodstock, Ontario. Following a year-long evaluation pilot with 3 Digit robots, TMMC announced in February 2026 a commercial RaaS agreement for 7 units, with deployment commencing April 2026. Automotive Manufacturing Solutions reported this as the first commercial humanoid robot deployment in Canadian automotive production.
The operational role is empty container management within the assembly process — a highly repetitive, bounded task.
Tim Hollander, President of Toyota Motor Manufacturing Canada, stated:
"After evaluating a number of robots, we are excited to deploy Digit to improve the team member experience and further increase operational efficiency."
Operational outcomes from the expanded 7-unit deployment are not yet public.
Procurement relevance: The conversion from a 3-unit pilot to a 7-unit commercial agreement is a meaningful signal of deployment viability in automotive assembly. The 12-month pilot period before commercial commitment is a reference data point for buyers planning evaluation timelines. The task scope is narrowly defined; broader manufacturing task sets are not validated by this case.
Case Study 3 — Mercedes-Benz Group: Manufacturing Pilot, Germany
Mercedes-Benz Group announced a commercial pilot agreement with Apptronik in March 2024, making it the first known commercial deployment announcement for the Apollo platform. Planned use cases include supplying assembly kits to production lines and inspecting components — tasks aligned with the bounded-task profile that has proven most accessible to current humanoid capability.
A Mercedes-Benz executive framed the programme as addressing labour gaps on repetitive tasks. No operational outcome data — throughput figures, uptime rates, or cost comparisons — has been published by either party as of this report's research period. This deployment remains at the announcement and early-pilot stage for procurement purposes.
Procurement relevance: The Mercedes-Benz announcement establishes that Apollo is under evaluation at a credible enterprise level. It does not provide the operational evidence required to support a procurement decision for Apollo in comparable environments. Buyers should seek direct vendor disclosure of any outcome data from this programme before proceeding to evaluation.
6. Friction, Risk & Unresolved Issues
Quick Answer — What risks should companies assess before committing to a humanoid robot procurement?
The most significant current deployment barriers are operational uptime, integration complexity, and the absence of global standards for human-collaborative humanoids. Platforms in the category currently sustain runtime of approximately 1–4 hours before recharge — well below the 95%+ uptime expected in enterprise operations. No globally harmonised standard exists for humanoid robots working alongside humans; each deployment requires site-specific validation. Integration with existing warehouse management and manufacturing execution systems is not standardised across vendors. Site-specific technical validation before procurement commitment is essential for all three platforms.
Technical Limitations
Digit, Apollo, and Optimus — alongside the broader humanoid category — are constrained by battery endurance, autonomy reliability, and perception accuracy in real-world environments. Independent analysis identifies typical continuous operating time of 30–90 minutes before recharge and a sustained runtime of 1–4 hours under controlled conditions. Enterprise operations expecting 95%+ uptime — the standard in most logistics and manufacturing contracts — cannot currently achieve that threshold from any of the three platforms reviewed. This reflects the maturity stage of the entire category, not a platform-specific deficiency.
AI perception and behaviour models require retraining when deployed in new environments, and edge-case failures — unexpected object placement, floor surface variation, human behaviour variation — require operational fallbacks including teleoperation in some configurations. Buyers deploying humanoids in environments more complex than single-task, bounded-space operations face elevated failure-rate risk.
Safety and Regulatory Gap
No global harmonised standard exists for humanoid robots operating in close proximity to human workers. Digit v4 deployments have operated with physical separation or fencing in place; Agility Robotics states that v5 is designed to operate without this requirement (company-claimed). Apptronik offers a wheeled-base Apollo variant positioned to align with existing industrial robot safety frameworks. Pending completion of ISO and national certification processes, every humanoid deployment requires site-specific safety validation. This adds time and cost to any deployment programme.
Physical AI Journal does not evaluate, certify, or comment on the safety conformity of specific platforms. Buyers must consult qualified safety engineers and the relevant certification bodies for their jurisdictions before any deployment.
Economic and Vendor Risks
No independent deployment has demonstrated profitability at enterprise scale as of mid-2026. Cost-overrun risk is elevated for buyers who deploy into multi-task or variable-environment applications before operational frameworks for humanoids in those environments are established. Tesla's Optimus programme has attracted external scrutiny regarding timeline credibility; Musk's original mass-production target for 2023 has shifted to 2026–27 statements with production line aspirations of 1,000,000 units/year at Fremont and 10,000,000 units/year in Texas (company-claimed). The gap between these projections and current external availability is a vendor risk signal that buyers should account for in platform selection.
Enterprise Risk Matrix — Humanoid Robot Deployment 2026 | |||||
Risk Category | Risk Description | Likelihood | Business Impact | Mitigation | Verification |
Technical | Uptime below 95% threshold; recharge downtime | High | High — production disruption | Phased pilot; define uptime KPIs in contract | VERIFIED — Industry analysis |
Safety / Regulatory | No global standard for human-collaborative humanoids | Certain — gap exists | High — site-specific validation required | Engage safety engineers pre-deployment; site audit | VERIFIED — Industry analysis |
Financial | ROI unproven at enterprise scale | High | High — capital commitment without return evidence | Single-unit pilot before fleet commitment; open-book cost modelling | VERIFIED — No independent profitability study published |
Integration | WMS/MES integration complexity and cost | Medium–High | Medium — delays and overruns on integration | Pre-procurement technical assessment; vendor API documentation review | ESTIMATED — No vendor disclosure of integration complexity |
Vendor viability | Platform discontinuation or pivot pre-scale | Medium | High — stranded assets; retraining; loss of support | Assess vendor funding runway; SPAC/IPO status; contract exit clauses | VERIFIED — Agility SPAC filing; Apptronik Series A |
Supply chain | Component availability for fleet scaling | Medium | Medium — deployment delays | Confirm production capacity commitments before fleet orders | ESTIMATED — No public production capacity data |
Policy / Labour | Workforce displacement regulatory or union response | Low–Medium (current) | Medium — operational or reputational | Engage HR and labour relations before deployment announcement | VERIFIED — Documented in global policy reporting |
7. Competitive Platform Comparison
Quick Answer — How do Digit, Apollo, and Optimus compare for enterprise procurement?
Digit leads on commercial deployment evidence, vendor financial structure, and available operating model (RaaS with a published indicative rate). Apollo leads on stated payload capability (25 kg, company-claimed) and has secured substantial Series A funding, but carries no published commercial outcome data. Optimus has no external commercial availability, the lowest stated price target (~$20,000 at scale, company-claimed), and the highest timeline uncertainty of the three platforms. For any procurement decision requiring evidence of operational performance, only Digit provides a verifiable commercial reference base as of mid-2026.
Competitive Platform Comparison — Digit vs Apollo vs Optimus — Mid-2026 | |||
Criterion | Digit (Agility Robotics) | Apollo (Apptronik) | Optimus (Tesla) |
Height | ~1.75 m (company-claimed) | ~1.73 m (company-claimed) | ~1.73 m (company-claimed) |
Weight | 65–76 kg (company-claimed) | ~72 kg (company-claimed) | ~57 kg (company-claimed) |
Payload | ~16 kg per arm (company-claimed) | ~25 kg (company-claimed) | ~20 kg (company-claimed) |
Battery Runtime | ~4 hr (v4); v5 not disclosed (category range: 1–4 hr, industry analysis) | ~1–4 hr, swappable batteries (company-claimed) | Not disclosed |
Walking Speed | ~5 km/h (company-claimed) | Not disclosed | Not disclosed |
Commercial Availability | Yes — RaaS contracts live | Pilots only — no commercial revenue disclosed | No — internal use only as of mid-2026 |
Pricing Model | RaaS ~$8,500/month (company-claimed) | Not published | ~$30,000 unit / ~$20,000 at scale (company-claimed; no external availability) |
Confirmed Customer Deployments | GXO Logistics, Toyota Motor Mfg Canada; others company-claimed (9 sites) | Mercedes-Benz (pilot announced; no outcome data) | None (external) |
Physical Separation / Fencing | v4 deployments have used fencing; v5 claims fence-free (company-claimed) | Wheeled variant positioned to align with existing standards | Not disclosed |
Vendor Financial Position | SPAC merger in progress; $2.5B pre-money valuation; $620M+ expected proceeds (company-claimed) | $935M Series A closed Feb 2026 | Tesla (NASDAQ: TSLA) — parent balance sheet |
Key Investors / Partners | NVIDIA, Amazon, SoftBank, Schaeffler | Mercedes-Benz, Google DeepMind | Tesla internal |
Overall Verification Status | VERIFIED — Commercial deployment confirmed | COMPANY-CLAIMED — Pilot announced; no outcomes | COMPANY-CLAIMED — No external deployment |
Vendor Claim Audit
Vendor Claim Audit — Key Figures Under Scrutiny | |||
Claim | Vendor | Verified? | Evidence / Notes |
65,000+ operational hours across 9 sites | Agility Robotics (Digit) | COMPANY-CLAIMED | Reported in Agility investor materials; not independently audited |
100,000+ totes moved at GXO | Agility Robotics (Digit) | COMPANY-CLAIMED | Agility press release Jun 2025; GXO CTO confirmed operational deployment in the same release; tote figure not independently audited |
$300M+ in committed Digit v5 orders; 30+ enterprise customers in pipeline | Agility Robotics | COMPANY-CLAIMED | SEC investor deck, Jun 2026; not independently audited |
Apollo payload: 25 kg | Apptronik | COMPANY-CLAIMED | Press release only; no third-party verification |
Optimus target price: ~$30,000 / ~$20,000 at scale | Tesla | COMPANY-CLAIMED | CEO statements; no confirmed product availability or timeline |
Optimus production targets: 1M units/year (Fremont) by 2027 | Tesla | COMPANY-CLAIMED | CEO statements Apr 2026; prior production targets have not been met on stated timelines |
RoboFab capacity: 10,000 units/year | Agility Robotics | COMPANY-CLAIMED | Investor materials; facility opened; production rate not independently confirmed |
Procurement Readiness Scorecard
Procurement Readiness Scorecard — Indicative Assessment, Mid-2026 (1 = Low, 5 = High) | ||||
Criterion | Weight | Digit | Apollo | Optimus |
Deployment maturity | 15% | 5 — Commercial revenue contracts live | 2 — Pilots announced | 1 — Internal only |
Verified customer evidence | 15% | 4 — GXO, TMMC confirmed | 2 — Mercedes announced, no outcomes | 1 — None external |
Commercial availability | 10% | 5 — RaaS contracts offered | 2 — Pre-commercial | 1 — Not available |
Pricing transparency | 10% | 3 — RaaS rate in investor deck (company-claimed) | 1 — Not published | 2 — Target stated (company-claimed); not available |
Vendor financial stability | 10% | 4 — SPAC merger; strong backers | 4 — $935M Series A | 5 — Tesla parent balance sheet |
Integration complexity | 10% | 3 — WMS integration demonstrated at GXO; not standardised | 2 — Limited evidence | 1 — No data |
Warehouse suitability | 10% | 5 — Confirmed in GXO deployment | 3 — Stated; not yet confirmed | 1 — No evidence |
Manufacturing suitability | 5% | 4 — TMMC deployment confirmed | 3 — Mercedes pilot announced | 2 — Internal use claimed |
ROI outlook | 5% | 3 — Cost structure partially known; profitability unproven | 1 — Insufficient data | 1 — Insufficient data |
Procurement risk | 10% | 3 — Moderate; active RaaS structure; SPAC completion risk | 2 — High; pre-commercial; limited data | 1 — Very high; no external availability |
This scorecard reflects current deployment evidence only. It is not an endorsement or a technical evaluation. Scores represent Physical AI Journal's analytical assessment of publicly available information; buyers should validate against their specific requirements.
8. Strategic Recommendations for Automation and Procurement Buyers
Quick Answer — Should we deploy a humanoid robot now, run a pilot, or wait?
The decision depends on use case, risk appetite, and operational context. For warehouse logistics involving bounded, repetitive tote-handling tasks, a structured Digit pilot of 1–3 units is supportable on current evidence — provided it is structured with defined KPIs and exit clauses before fleet commitment. For manufacturing environments requiring flexible task handling, a monitor posture for Apollo and Optimus is appropriate until outcome data from current pilots is published. No buyer should commit to fleet-scale deployment of any humanoid platform without a completed single-unit pilot demonstrating performance against their specific task requirements.
Buy–Pilot–Monitor–Wait Matrix — By Use Case, Mid-2026 | |||
Use Case | Recommended Platform | Decision | Rationale |
Warehouse logistics — bounded tote handling | Digit | PILOT | GXO deployment evidence supports structured pilot; define uptime KPIs before signing |
Automotive assembly — empty tote / kit supply | Digit | PILOT | TMMC deployment directly supports this use case; 12-month evaluation before fleet recommended |
General manufacturing — multi-task | Apollo / Digit | MONITOR | No platform has demonstrated multi-task manufacturing performance at commercial scale |
Distribution / e-commerce fulfilment | Digit | MONITOR | Mercado Libre agreement announced but no operational outcomes published; reassess when outcome data is available |
General enterprise — variable-task environments | Any | WAIT | No platform has operational evidence supporting variable-task deployment; cost-overrun risk is high |
Any use case — Optimus | Optimus | WAIT | No external commercial availability; timeline credibility risk documented; revisit from 2027 if external deployment is announced |
Structuring a Defensible Pilot
Buyers who proceed with a Digit pilot should negotiate the following contract terms before signing: defined minimum uptime commitments with financial recourse if unmet; clarity on whether infrastructure modification, integration, and training costs are included or additional; exit clauses tied to operational performance KPIs at 6 and 12 months; data rights over operational performance metrics generated during the pilot period; and clarity on fleet expansion pricing to prevent uncertainty if the pilot succeeds.
A 1–3 unit pilot over 12 months is the minimum evaluation period supported by the Toyota Motor Manufacturing Canada reference case. Compressing this timeline to meet internal project deadlines increases post-commitment risk proportionally.
Platform Selection Decision
Buyers operating in warehouse logistics or bounded automotive assembly tasks can, on current evidence, initiate a structured procurement process for Digit with a defined pilot scope. Buyers evaluating Apollo should place it on a formal vendor watch list with a defined review trigger: re-evaluate when Mercedes-Benz or another named pilot customer publishes operational outcome data. Buyers evaluating Optimus should set a separate trigger: add Optimus to the active evaluation shortlist when external commercial availability is announced with confirmed pricing.
Neither Apollo nor Optimus provides sufficient operational evidence to support procurement commitment in 2026. Automation and Procurement Buyers who act on the evidence in this report can initiate a Digit pilot, commit Apollo and Optimus to a dated vendor review schedule, and avoid fleet-scale exposure to any platform until single-unit operational performance has been independently validated against their specific task requirements.
9. Executive FAQ
Buyer Decision Questions
The six questions below address the procurement decisions most frequently raised by enterprise buyers evaluating humanoid robot platforms in 2026. Answers draw exclusively from verified and company-claimed research data underpinning this report.
Q1. Is Digit commercially available and can we sign a contract today?
Agility Robotics offers Digit under a Robots-as-a-Service model with confirmed contracts in operation at GXO Logistics and Toyota Motor Manufacturing Canada. The indicative RaaS rate is approximately $8,500 per month per robot (company-claimed, SEC investor deck, June 2026). Contract terms, lead times, and site qualification criteria are not publicly disclosed; buyers should contact Agility Robotics directly before assuming availability.
Q2. What does it cost to deploy a humanoid robot for 12 months, all-in?
Platform cost for Digit is approximately $102,000 per year at the reported RaaS rate (company-claimed). Adding indicative integration, infrastructure, and training costs, a Year 1 total of $142,000–$282,000 per unit is a reasonable planning estimate. Apollo and Optimus have not published pricing. No vendor publicly discloses ancillary cost components; buyers must obtain itemised quotes before building a business case.
Q3. How long does a humanoid robot run before it needs recharging?
Industry analysis across the category identifies a sustained runtime of 1–4 hours, with typical continuous operation of 30–90 minutes before recharge in current commercial configurations. Digit v4 has a reported runtime of approximately 4 hours; Apollo uses swappable batteries with a comparable range (company-claimed). No platform currently meets the 95%+ uptime standard expected in enterprise logistics and manufacturing. Recharging logistics and shift scheduling should be primary operational planning variables.
Q4. When will Optimus be available for external customers to purchase?
Tesla has not announced a date for external commercial availability of Optimus as of mid-2026. CEO statements in April 2026 referenced production line aspirations for 2027 (company-claimed), but prior production targets for the platform have not been met on their stated timelines. Optimus should not be included in active procurement planning until Tesla announces confirmed external availability with pricing.
Q5. Do humanoid robots need safety cages or special facility modifications?
Digit v4 deployments have operated with physical fencing or safety separation in place; Agility Robotics states that v5 is designed to operate without this requirement (company-claimed, not yet validated in commercial deployment). Apptronik offers a wheeled-base Apollo variant positioned to align with existing industrial robot safety frameworks. No global harmonised standard currently exists for bipedal humanoid robots operating alongside humans. Site-specific safety assessment by qualified engineers is required for every deployment; Physical AI Journal does not evaluate or certify the safety characteristics of any platform.
Q6. How should we evaluate Apollo given that no outcome data has been published?
Apollo belongs on a vendor watch list, not an active procurement shortlist. The criteria for elevation to active evaluation are: publication of operational outcome data from the Mercedes-Benz programme or another named pilot; publication of pricing terms; and evidence of commercial revenue deployment at a named customer. Apptronik's $935M Series A and its Google DeepMind partnership on embodied AI are development-trajectory signals, not operational readiness signals. Set a formal review trigger at Q2 2027 and reassess against published evidence at that point.
Scope & Disclaimer
This report provides market intelligence, deployment evidence, and commercial analysis for informational purposes only. It does not constitute legal, financial, investment, engineering, safety-certification, or professional advice of any kind. Physical AI Journal and Sekason Research Limited do not evaluate, certify, or endorse the safety, technical conformity, or fitness-for-purpose of any humanoid robot platform or deployment described herein. Company-claimed figures — labelled as such throughout this report — originate from manufacturers and vendors and have not been independently verified by Sekason Research Limited; they should not be relied upon as confirmed facts without independent validation. Any procurement, deployment, or capital-allocation decision should be taken only after independent professional, technical, and legal assessment.
For full terms, see physicalaijournal.org/disclaimer.
References & Strategic Sources:
This report is backed by authoritative research, independent verification, and structured analytical methodology.
Agility Robotics | "Digit Moves Over 100,000 Totes in Commercial Deployment" | Undated (cited Jun 2025) | agilityrobotics.com | Company-Claimed (Tier 3)
GXO Logistics / GlobeNewswire | "GXO Signs Industry-First Multi-Year Agreement with Agility Robotics" | June 27, 2024 | globenewswire.com/news-release/.../2905556 | Company-Claimed (Tier 3)
Agility Robotics | "Mercado Libre and Agility Robotics Announce Commercial Agreement" | Mid-2025 | agilityrobotics.com | Company-Claimed (Tier 3)
Agility Robotics | "Commercial Agreement with Toyota Motor Manufacturing Canada" | February 19, 2026 | agilityrobotics.com | Company-Claimed (Tier 3)
Automotive Manufacturing Solutions | "Toyota Canada signs humanoid robot deal with Agility" | February 19, 2026 | automotivemanufacturingsolutions.com/.../2611551 | Verified (Tier 2)
Agility Robotics | "Agreement with Schaeffler Group" | November 13, 2024 | agilityrobotics.com | Company-Claimed (Tier 3)
Apptronik | "Apptronik closes over $935M Series A" | February 11, 2026 | apptronik.com/news-collection/ | Company-Claimed (Tier 3)
Apptronik | "Welcome to Robot Park..." | June 30, 2026 | apptronik.com/news-collection/ | Company-Claimed (Tier 3)
Apptronik | "Apptronik and Mercedes-Benz Enter Commercial Agreement" | March 15, 2024 | apptronik.com/news-collection/ | Company-Claimed (Tier 3)
Agility Robotics | "Agility Opens New Fremont Facility..." | July 16, 2026 | agilityrobotics.com | Company-Claimed (Tier 3)
Agility Robotics | "Agility Robotics Strengthens Executive Leadership Team..." (CFO hire) | July 23, 2026 | agilityrobotics.com | Company-Claimed (Tier 3)
SEC / Churchill Capital XI | Churchill Cap XI investor deck (Ex. 99.2, SPAC merger filing) | June 2026 | sec.gov/Archives/.../0331-39577...exhibit992.htm | Verified (Tier 1)
RoboZaps | "Digit Review 2026: Specs, Price & Deployments" | May 19, 2026 | blog.robozaps.com/.../agility-robotics-digit-review | Verified (Tier 2)
RoboZaps | "Challenges in Humanoid Robotics" | January 30, 2026 | blog.robozaps.com/.../challenges-in-humanoid-robotics | Verified (Tier 2)
Amazon Robotics | "Amazon announces new fulfillment center robots, Sequoia and Digit" | 2023 | aboutamazon.com/news/...amazon-introduces-robotics | Verified (Tier 1)
Silicon Valley Robotics Center (SVRC) | "State of Robotics 2026" | March 2026 | roboticscenter.ai/en/state-of-robotics-2026 | Verified (Tier 2)
Technology Org / Humanoid Robotics Technology | "Tesla's Optimus Roadmap" (Musk statements) | January 2025 | humanoidroboticstechnology.com/industry-news/ | Verified (Tier 2)
Reuters | "China's AI-powered humanoid robots..." | May 13, 2025 | reuters.com/world/china/chinas-ai-powered-humanoid-robots-2025-05-13 | Verified (Tier 1)
Reuters | China robots conference coverage | June 9, 2024 | reuters.com/world/china/ | Verified (Tier 1)
Comments