Figure AI showcases humanoid robots in 24/7 livestream

๐กSee how Figure AI's humanoid robots handle real-world logistics tasks in a 24/7 live environment.
โก 30-Second TL;DR
What Changed
Figure AI initiated a 24/7 livestream to demonstrate robot capabilities
Why It Matters
This public demonstration signals a shift toward deploying humanoid robots in logistics and warehouse environments. It sets a new standard for transparency in the robotics industry.
What To Do Next
Monitor Figure AI's livestream to observe how their motion planning handles edge cases in real-time package manipulation.
Key Points
- โขFigure AI initiated a 24/7 livestream to demonstrate robot capabilities
- โขThe robots are shown performing complex physical tasks like package handling
- โขThe initiative aims to showcase real-world reliability and progress in embodied AI
๐ง Deep Insight
Web-grounded analysis with 24 cited sources.
๐ Enhanced Key Takeaways
- โขThe 24/7 livestream showcased Figure AI's Helix 02-powered robots autonomously sorting over 88,000 packages for approximately 72 hours, significantly surpassing an initial 8-hour target with "zero failures".
- โขFigure AI terminated its partnership with OpenAI in early 2026, opting to develop its proprietary Helix AI models in-house for high-rate robot control, citing "very little" technical value from the collaboration.
- โขThe company has garnered substantial investment, including a $675 million Series B round in February 2024 and over $1 billion in a Series C round in September 2025, elevating its post-money valuation to $39 billion.
- โขFigure AI plans to scale production through its BotQ manufacturing facility, aiming to produce 12,000 humanoids annually, and employs a "Robot-as-a-Service" (RaaS) model with monthly subscriptions around $1,000 per robot.
- โขBeyond industrial applications, Figure AI is developing its Figure 03 robot for domestic tasks, with alpha testing in real homes commencing in late 2025, signaling an expansion into the consumer robotics market.
๐ Competitor Analysisโธ Show
| Company / Robot | Focus / Key Features | 2025 Shipments (approx.) | Pricing (if available) |
|---|---|---|---|
| Figure AI (Figure 01/03) | General-purpose, logistics, warehousing, automotive, household tasks; AI-powered | 150 units | ~$1,000/month (RaaS) |
| Agility Robotics (Digit) | Logistics, warehouses, material handling; bipedal | 150 units | Null |
| Apptronik (Apollo) | General-purpose, factories (stacking, machine tending); modular, human-robot collaboration | Null | Null |
| Boston Dynamics (Atlas, Spot, Stretch) | Atlas (research humanoid), Spot (industrial inspection), Stretch (warehouse unloading) | Null (Atlas research, Spot/Stretch commercial but not humanoid) | Null |
| Tesla (Optimus) | General-purpose, long-term for homes/factories; public sales forecasted 2027 | 150 units | Null (public sales forecasted 2027) |
| 1X (NEO) | Home/consumer focus; pre-orders, initial tele-operation | Null | ~$20,000 or $499/month |
| Unitree (R1, G1) | General-purpose, affordable; R1 ($5,900), G1 ($16,000) | 5,500 units (highest) | R1: ~$5,900, G1: ~$16,000 |
| UBTECH | Consumer-friendly, STEM education, retail demos; smooth motion, expressive faces | 1,000 units | Null |
๐ ๏ธ Technical Deep Dive
- Robot Models: Figure 01 (prototype), Figure 02 (industrial-grade), Figure 03 (next-generation, consumer-focused).
- Dimensions & Weight (Figure 01/03): Approximately 168 cm (5'6") tall, 60 kg (132 lbs).
- Payload Capacity: Up to 20 kg (44 lbs).
- Speed: Maximum walking speed of 1.2 m/s (2.7 mph).
- Power Source & Runtime: Battery pack with approximately 5 hours of runtime per charge.
- Actuators: Custom electromechanical actuators at every joint, fully electric (no hydraulics).
- Degrees of Freedom (DoF): 40+ (Figure 01), with 7 DoF per arm, 6 DoF per leg, and additional DoF in hands, head, and torso. Dexterous multi-fingered hands with individual finger control and force sensing.
- Perception: Equipped with RGB cameras, depth sensors, LiDAR, and tactile sensors for 3D environmental mapping, object recognition, and physical interaction feedback. Figure 03 features palm cameras and tactile sensors detecting forces as small as 3 grams.
- AI System:
- Initially co-developed with OpenAI, utilizing vision-language AI models for common sense reasoning from images, speech-to-text/text-to-speech, and autonomous task planning.
- Transitioned to proprietary Helix AI (Helix 01, Helix 02) in 2025/2026.
- Helix is a Vision-Language-Action (VLA) model that can control up to two robots at once.
- Helix 02 runs on dual onboard GPUs: System 2 for high-level planning (7โ9 Hz) and System 1 for low-level motor control (200 Hz).
- Enables robots to interact with the world without extensive manual training, pick up nearly any small household object, understand spoken instructions, describe what it sees, reason about tasks, and execute multi-step actions autonomously.
- All inference runs onboard without cloud dependency for core task execution.
๐ฎ Future ImplicationsAI analysis grounded in cited sources
โณ Timeline
๐ Sources (24)
Factual claims are grounded in the sources below. Forward-looking analysis is AI-generated interpretation.
- techrepublic.com
- oodaloop.com
- wikipedia.org
- robozaps.com
- letsdatascience.com
- thomasnet.com
- sacra.com
- forbes.com
- robozaps.com
- visualcapitalist.com
- eesel.ai
- standardbots.com
- businessinsider.com
- robozaps.com
- ieee7017roboethics.design
- qviro.com
- aparobot.com
- startupintros.com
- apnews.com
- prnewswire.com
- figure.ai
- humanoid.guide
- businessinsider.com
- youtube.com
Weekly AI Recap
Read this week's curated digest of top AI events โ
๐Related Updates
AI-curated news aggregator. All content rights belong to original publishers.
Original source: Ars Technica AI โ


