Figure 03, Apptronik Apollo, and Agility Digit are the three enterprise humanoid robots with actual paying customers behind them — not letters of intent, not internal pilots, not press releases — and after going through their real deployment records, funding levels, and what each company's actual paying customers report from the factory floor, the picture that emerges is less a three-way tie and more a story of one robot that's already proven itself, one that just raised nearly a billion dollars to catch up, and one that's ramping production faster than anyone expected.
For enterprise buyers, the question isn't which one has the best demo video. It's which one ships, works, and has the numbers to prove it.
This is the fourth installment in our humanoid robot series — previous pieces covered manufacturing and care deployment broadly, consumer home robots, and the 1X NEO vs Unitree G1 vs Tesla Optimus consumer-focused comparison. This one focuses entirely on enterprise deployment: who's actually working in factories and warehouses right now, and what the real ROI story looks like.
Quick Comparison Table
| Feature | Figure 03 | Apptronik Apollo | Agility Digit |
|---|---|---|---|
| Height / Weight | ~5'6" / ~70 kg | 5'8" / ~72 kg (160 lbs) | ~5'9" / ~65 kg |
| Payload Capacity | Not published (dexterous manipulation focus) | 25 kg (55 lbs) | ~16 kg tote handling |
| Battery / Runtime | Wireless charging for near-continuous operation | Hot-swappable battery packs, 4 hrs per pack | ~16 hrs endurance (longest in category) |
| Key Deployment | BMW Spartanburg (Jan 2026+, 40-unit fleet); BotQ factory at 1 unit/hour production rate | Mercedes-Benz Germany + Hungary; GXO Logistics; Jabil manufacturing | GXO Flowery Branch: 100,000+ totes; Toyota Canada RAV4 plant (7 units) |
| AI Partner | In-house Helix AI (multimodal, speech-to-speech) | Google DeepMind Gemini Robotics | In-house; Amazon robotics data integration |
| Funding Raised | $1.75B raised; $39B valuation | $935M+ Series A (closed Feb 2026) | $641M raised; $2.12B valuation |
| Pricing / Access | No public price; partner/pilot track only | Quote-only (~$200,000–$300,000 target) | RaaS: ~$10–12/hr or $20–25K/year; ~$250,000 purchase |
Agility Digit: The Only One Actually Generating Commercial Revenue
If there's one fact worth anchoring this entire comparison to, it's this: Agility Digit is the only humanoid robot in the world that was generating revenue from productive commercial work as of April 2026. Not internal deployment. Not a pilot offset by a partnership agreement. Paying customers, measurable output, recurring contracts. That distinction matters more than any spec sheet comparison.
The numbers are specific: Digit surpassed 100,000 totes moved at GXO's Flowery Branch facility, GXO signed an industry-first multi-year Robot-as-a-Service agreement, Toyota Canada signed a commercial agreement for seven units at its RAV4 plant, and Amazon — which invested in Agility — is openly using Digit in its fulfillment operations as part of its broader robotics strategy. These aren't LOIs (letters of intent). They're contracts with delivered hardware and tracked KPIs.
Agility's battery endurance lead is also significant and rarely discussed in comparison coverage. At up to 16 hours per charge, Digit runs longer on a single charge than either Figure 03 or Apollo by a wide margin, and its cloud fleet management platform lets operators monitor multiple units, track KPIs, and deploy updates without per-unit intervention. RoboFab in Salem, Oregon has capacity for 10,000 units per year — the clearest near-term production commitment of the three.
The honest limitation is task specialization. Digit's current proven workflow is overwhelmingly tote handling — moving, placing, and sorting standardized bins in warehouse environments designed for human-scale movement. "General-purpose humanoid" is the brand positioning. "Very good tote mover" is closer to the current commercial reality. Fine-manipulation tasks, variable surfaces, and unscripted environments are where Digit's proven deployment record gets thinner.
Apptronik Apollo: The Best-Funded Newcomer With the Deepest Industrial Partnerships
Apollo's February 2026 close of over $935 million in Series A funding — with a strategic partnership with Google DeepMind to power the robot with Gemini Robotics — was the largest single financing round in enterprise humanoid history. That capital is earmarked for production scaling through Jabil, one of the world's largest contract manufacturers, and for deepening the Mercedes-Benz pilot beyond Germany and Hungary if KPIs hold.
Apollo's engineering profile is genuinely differentiated from Digit in one key dimension: modularity. Apollo's torso can mount on bipedal legs, a wheeled base, or a stationary platform using the same core robot, hedging against the still-open question of which mobility approach wins for different factory tasks. Its 25 kg payload, hot-swappable battery system, and five-digit hands with a more robust dexterity profile than Digit's tote-optimized grippers make it a credible candidate for a wider range of intralogistics and light manufacturing tasks.
The honest gap is that Apollo is still primarily in pilot and validation mode rather than recurring revenue mode. Named customers — Mercedes-Benz, GXO Logistics, Jabil — are real, and the early reporting from those pilots is positive. But the deployment scale is a fraction of Digit's proven tote count, and no multi-year recurring RaaS contract equivalent to the GXO deal has been publicly confirmed for Apollo. The Google DeepMind partnership is also worth watching carefully: it gives Apollo a powerful AI engine, but it means Apollo's core intelligence roadmap is in the hands of a partner rather than in-house, which introduces a dependency that some enterprise procurement teams have flagged as a risk factor.
Figure 03: The Fastest Production Ramp, With a Deliberately Limited Buying Path
Figure 03's commercial story reads differently from the other two, because Figure has made a deliberate choice to restrict who can buy its robots rather than maximizing unit volume. The BotQ factory in San Jose is building at one robot per hour as of April 2026, targeting 12,000 units per year — impressive production velocity — but those units are flowing to internal R&D, data collection, and a small number of select commercial partners, not to open procurement.
BMW's Spartanburg plant is the flagship deployment, and it's legitimately significant: Figure 03 began deploying there in January 2026 following an 11-month Figure 02 run that loaded over 90,000 parts across 1,250 operational hours, and it's the first deployment where Figure 03's fourth-generation hands with tactile pads, palm cameras, and in-hand memory functions are being tested at production scale. Figure 03's wireless charging for near-continuous operation is also a meaningful differentiator — both Apollo's hot-swap battery and Digit's long-endurance approach involve interrupting the robot's work cycle; Figure's model doesn't.
The clearest honest challenge with Figure 03 from an enterprise buyer's perspective is access. There is no public price, no open purchasing path, and no confirmed pricing even for enterprise customers. Figure's $39 billion valuation implies a bet on scarcity-driven premium positioning — select partners, controlled rollout — rather than the volume-first approach Agility is taking with RaaS. For a company trying to evaluate whether humanoids belong in their operations in 2026, "submit a partner inquiry and wait" is a harder starting point than GXO's published hourly rate.
What This Means for Enterprise Buyers in 2026
- Need proven commercial deployment with measurable ROI and a clear pricing model? Agility Digit is the only realistic choice right now. The GXO and Toyota contracts are real, the RaaS pricing is published, and 100,000 totes is a number you can actually benchmark against.
- Have the runway to absorb a multi-month integration cycle, need 25 kg payload with modular mobility options, and want Google DeepMind's AI stack behind the robot? Apollo is the right conversation to have, especially if your operations overlap with the Mercedes or Jabil pilot environments. Just go in expecting quote-only pricing and a partnership relationship rather than a SKU.
- Want access to the most advanced manipulation hardware and the fastest production ramp, and you're a large manufacturer with the volume to get a partner allocation? Figure 03's BMW track record and BotQ production rate make it the most technically ambitious of the three — but only accessible through a relationship that takes time to establish.
One thing all three companies agree on, whether or not they say it out loud: the gap between "commercial deployment" and "production-ready at scale" is still very real. Every confirmed deployment is still task-specific. No humanoid is running general-purpose, unscripted factory work at scale in 2026. The companies that will be in the best position for 2027 and 2028 are the ones accumulating real-world operational data now — and on that specific metric, Digit's 100,000-tote lead is harder to close than any spec sheet advantage.
Frequently Asked Questions
Which enterprise humanoid robot has the most real-world deployments in 2026?
Agility Digit leads by a significant margin — it's the only humanoid generating revenue from commercial work, having moved over 100,000 totes at GXO's warehouse, signed a multi-year RaaS contract, and secured paying agreements with Toyota Canada, all of which are revenue-generating deployments rather than pilots.
What is the price of Apptronik Apollo?
Apollo has no published list price — it's quote-only, with a target range of roughly $200,000 to $300,000, accessible through direct enterprise sales engagement rather than a standard purchasing channel.
How does Figure 03 differ from Figure 02?
Figure 03 features fourth-generation hands with tactile sensing pads, palm cameras, and in-hand memory functions that significantly expand manipulation capability, along with wireless charging for near-continuous operation — improvements specifically designed for the wider range of tasks beyond the sheet-metal loading BMW pilot Figure 02 focused on.
What does RaaS mean for humanoid robots?
Robot-as-a-Service means the enterprise customer pays an operating rate — Agility's Digit is priced at roughly $10 to $12 per operating hour, or $20,000 to $25,000 per year — rather than purchasing hardware outright, which lowers the capital barrier and shifts maintenance responsibility to the robot maker.
Which company has the most funding in enterprise humanoid robotics?
Figure AI leads with $1.75 billion raised at a $39 billion valuation, followed by Apptronik at $935 million closed in its Series A in February 2026, and Agility Robotics at $641 million raised with a $2.12 billion valuation.