24x Throughput: Figure Scales Manufacturing to One Robot Per Hour

Figure has increased its manufacturing throughput by 24x in under 120 days, moving from producing one robot per day to one robot per hour.
The company has manufactured over 350 Figure 03 units to date, with a target of 55 humanoids to be built this week alone.
A new AI milestone, System 0 (S0), allows the Figure 03 to traverse stairs and uneven terrain "zero-shot" by integrating vision into its whole-body control policy.
Manufacturing maturity at the BotQ facility has reached an 80% first-pass yield, supported by the production of over 9,000 custom actuators.

In the race to achieve industrial-scale humanoid deployment, Figure has shifted from "hockey stick" projections to tangible assembly line results. The Sunnyvale-based robotics firm announced today that its BotQ manufacturing facility has achieved a 24x increase in throughput over the last four months.

The company is now producing one Figure 03 humanoid every hour, a significant jump from its January rate of one unit per day. CEO Brett Adcock confirmed that the company is on track to manufacture 55 robots this week, signaling that the transition from research-level prototyping to mass manufacturability is effectively complete.

A ground-level perspective from the center of a production floor, showing symmetrical rows of white and charcoal-grey Figure 03 humanoid robots stretching into the background of the BotQ industrial facility. — Figure has achieved a 24x increase in manufacturing throughput in under 120 days, scaling from one robot per day to one per hour. This ramp-up at the BotQ facility has already resulted in the delivery of over 350 Figure 03 units as the company targets an eventual production capacity of 50,000 robots per year.

Inside the BotQ Machine

The acceleration at BotQ—the facility Figure claims is central to its supply chain decoupling strategy—is driven by a custom manufacturing execution system (MES) running across 150 networked workstations. Unlike the retired Figure 02 fleet, which relied on lower-volume processes, the Figure 03 was designed for this specific ramp-up.

Key manufacturing metrics released by the company include:

Production Volume: Over 350 Figure 03 units have been delivered to date.
Component Scale: The facility has manufactured 9,000 actuators across 10 distinct SKUs.
Yield Rates: End-of-line first-pass yield has surpassed 80%, while the dedicated battery line has achieved a 99.3% yield over 500 packs.

To ensure reliability at this volume, each robot undergoes 80 functional verification tests, including "burn-in" sessions that involve thousands of cycles of squats, presses, and jogging to identify hardware infants deaths before they leave the factory.

Figure

@Figure_robot

·Follow

Today we’re giving an update on ramping F.03 production at BotQ In the last 120 days, Figure scaled manufacturing 24x - from 1 robot/day to 1 robot/hour We will manufacture 55 humanoid robots this week

Watch on X

3:39 PM · Apr 29, 2026

1.5K

Read 87 replies

System 0: Closing the Perception Gap

Alongside the hardware surge, Figure unveiled a major software update to its System 0 (S0) AI model. Previously, the robot's whole-body controller was "blind," relying on proprioception (sensing its own limb positions) to walk on flat ground. While stable, this made stairs and ramps impossible without hand-tuned code.

Humanoids daily

@humanoidsdaily

·Follow

Figure has cleared the gap to industrial scale, ramping manufacturing 24x in under 120 days at its BotQ facility. Now producing one robot per hour, the expanding fleet has unlocked "System 0"—a new AI model for perception-conditioned, zero-shot stair traversal.

Watch on X

4:17 PM · Apr 29, 2026

Read 4 replies

The new "perception-conditioned" S0 integrates RGB images from the head cameras directly into the control policy. By passing vision through a stereo model that creates a 3D representation of the world, the robot can now "see" the obstacles it is stepping on.

A close-up view of the head and neck assembly of a Figure 03 robot as a technician in black gloves carefully aligns and attaches the sleek black head to the chassis. — Every robot manufactured at BotQ undergoes 80 functional verification tests; this rigorous quality focus has helped Figure achieve an 80% first-pass yield for its third-generation humanoids.

Remarkably, Figure claims this capability was achieved via zero-shot transfer. The AI was trained entirely in simulation across thousands of randomized terrains and deployed to physical hardware without any real-world fine-tuning. This effectively bridges the "sim-to-real" gap that has long hindered autonomous locomotion in unstructured environments.

Solving the "Long Tail" of Failures

The purpose of scaling the fleet to hundreds of units is not just for commercial delivery, but for data collection. Adcock notes that a larger fleet allows Figure to encounter "invisible" failures that only appear after thousands of cumulative operating hours.

A front-facing, ground-level perspective showing long rows of white Figure 03 robots standing on black bases, emphasizing the scale of the manufacturing output at the California facility. — The 24x increase in manufacturing throughput allows Figure to deploy larger fleets into real-world environments, generating the data streams required to solve the 'long tail' of edge-case failures.

To manage this growing population of machines, Figure has deployed an internal Field Service Management system and a Fleet Management System. These tools allow the company to track real-time health and deploy Over-the-Air (OTA) updates to the entire fleet simultaneously. This infrastructure is critical for the 24/7 autonomous operations Figure is currently running in Sunnyvale.

An overhead, high-angle view of dozens of Figure 03 humanoid robots in both white and charcoal-grey finishes, standing in staggered, uniform rows on a polished industrial floor. — By scaling production at the BotQ facility to one robot per hour, Figure has expanded its fleet to over 350 Figure 03 units, which now serve as data-collection engines to accelerate the development of autonomous capabilities.

As production approaches the target of 50,000 units per year, Figure is shifting its focus from high-frequency bugs to the "long tail" of edge cases. With a production rate of one robot per hour, the company is now generating the data streams required to harden its Helix 02 architecture for the messy reality of both industrial floors and residential homes.