1X launches World Model Lab for humanoid autonomy

PLUS: BYD enters humanoid robotics with dealer sales, Figure 03 lands first major retail commercial deal, and US Army backs stealthy robotic logistics trucks


1X launches World Model Lab for humanoid autonomy

Welcome back to your Robot Briefing

1X Technologies just spun up a dedicated lab to build AI world models that let humanoid robots learn new tasks on the fly — no retraining required. The Norwegian startup hired a founding researcher from Luma AI after its NEO bot proved it could tackle completely unfamiliar work autonomously.

If they crack it, we're looking at robots that adapt like humans instead of needing months of fine-tuning for every new job. The real question: can world models scale fast enough to make general-purpose humanoids commercially viable, or will they hit the same wall that's kept AI robotics stuck in the lab?

In today's Robot update:

1X builds World Model Lab for adaptive humanoids
BYD jumps into humanoid robotics with dealer sales plan
Figure 03 lands first commercial deal with JCPenney parent
US Army funds stealthy autonomous logistics trucks
News

1X launches World Model Lab to scale humanoid autonomy

Comparison infographic contrasting traditional robotics, which requires years of task-specific programming, with 1X's new foundation model approach that achieves zero-shot execution, allowing robots to handle unfamiliar tasks without retraining.

Image Source: There's A Robot For That

Snapshot: 1X Technologies launched a dedicated lab to develop large-scale AI models for autonomous humanoids, hiring a founding researcher from Luma AI after internal breakthroughs enabled their NEO robot to execute completely new tasks without additional training.

Breakdown:

1X's existing world models recently enabled NEO to handle unseen tasks with zero-shot execution — meaning the robot performed new work without task-specific programming or demonstration.
The company argues robotics requires web-scale pretraining on diverse data sources including human video, simulations, and robot operating data, rather than fine-tuning narrow models for specific tasks.
The timing aligns with 1X's NEO factory going live, removing manufacturing constraints and shifting the bottleneck to AI development and real-world data collection.

Takeaway: A major robotics company is betting its roadmap on pretraining foundation models rather than programming specific tasks — the same architectural shift that unlocked ChatGPT's capabilities. If 1X is right about zero-shot execution, the timeline for general-purpose humanoids just compressed significantly, because each robot learns tasks without custom engineering.

News

BYD enters humanoid robotics, plans dealer network sales

Comparison chart illustrating BYD's distribution advantage, highlighting a massive network of over 30,000 dealerships ready to sell consumer robots compared to the zero retail footprint of typical tech competitors.

Image Source: There's A Robot For That

Snapshot: Chinese EV giant BYD confirmed it's developing humanoid robots for household use and plans to sell them through its automotive dealer network, while building an open platform that works with both BYD-developed and third-party robots.

Breakdown:

BYD will leverage its extensive auto dealer infrastructure for robot distribution — the same sales channel that moved it to global leadership in electric vehicles.
The company is building an open robot platform rather than a closed ecosystem, allowing both in-house and partner-developed products to operate within its framework.
BYD backs PaXini Tech, a robotics firm making dexterous hands and humanoid robots that's now exploring a Hong Kong IPO to fund expansion.

Takeaway: An automaker with proven manufacturing scale and a 30,000+ dealer network entering humanoids signals the market is moving beyond tech startups to industrial players with distribution advantages. BYD's open platform approach suggests they're positioning robots as consumer products sold alongside cars, not niche industrial equipment.

News

Figure AI lands first commercial humanoid deal with JCPenney parent Catalyst Brands

Snapshot: Figure AI signed a commercial agreement with Catalyst Brands, parent of JCPenney, Brooks Brothers, Aéropostale, Lucky Brand, and Nautica, to deploy Figure 03 humanoid robots at a 1.7 million square foot Reno distribution center, marking one of the first real commercial humanoid contracts with a major U.S. retailer.

Breakdown:

The Figure 03 fleet will handle sorting and packing at the Nevada facility, which received a $40 million upgrade in 2024 — the deployment targets repetitive logistics work rather than experimental tasks or curated demos.
Catalyst CEO Marc Rosen framed the deal around scalability and seasonality, positioning humanoids as flex labor that can ramp with retail demand cycles rather than fixed capacity additions, with the partnership designed to scale across the broader brand portfolio.
The agreement moves Figure from livestream demos and pilot tests to recurring commercial revenue tied to actual logistics throughput, with Brett Adcock's team targeting 100+ deployed units across the network as the relationship matures.

Takeaway: A first signed commercial agreement with a major retail conglomerate moves humanoid robots from "interesting demo" to "line item in a logistics budget." Operations leaders evaluating warehouse automation should note that the retail use case Figure picked, high-volume sorting and packing, is exactly where labor costs and turnover hurt the most, and where humanoid economics start working before they make sense in skilled assembly.

News

US Army backs stealthy robotic logistics trucks to solve battlefield supply gap

Snapshot: The US Department of Defense contracted American Rheinmetall and Harbinger to develop autonomous, hybrid-electric military cargo trucks designed to be affordable, expendable, and stealthy enough to operate in combat zones, reducing soldiers' manual supply hauling.

Breakdown:

The vehicles will have no cabs or manual controls and use hybrid-electric propulsion to run silently on battery power, reducing heat signatures and detection risk in combat zones.
The Pentagon wants platforms cheap enough to be "attritable" — meaning militaries can afford to lose them in dangerous situations rather than risk human drivers.
Initial focus is on autonomous tactical wheeled vehicles for combat logistics before expanding to next-generation platforms designed for coordinated human-robot operations.

Takeaway: Military logistics faces the same last-mile problem as commercial operations — eventually humans must manually carry supplies the final distance — but with higher stakes and willingness to pay for solutions. The Pentagon's emphasis on mass-producible and affordable autonomous vehicles suggests defense applications will drive down costs and prove reliability faster than waiting for consumer robotics to mature.

Other Top Robot Stories

Medtronic requested FDA clearance for its Hugo robotic surgery system in general and gynecological procedures, positioning the medical device giant to challenge Intuitive Surgical's dominance in hernia repair and other key surgical applications beyond the urologic procedures already approved in December.

JAKA unveiled Pi, a compact 1.22-meter humanoid weighing 42 kilograms with 27 degrees of freedom, marking the Shanghai-based collaborative robot maker's expansion into intelligent robotics with an open SDK platform targeting labs, schools, and office environments.

GlobalFoundries completed its acquisition of Synopsys' ARC Processor IP Solutions business, combining it with MIPS to deliver a software-to-silicon capability for Physical AI spanning RISC-V processor IP, custom design, and advanced manufacturing for automotive and industrial robotics applications.

Niqo announced its AI-powered agricultural weeding robotics business is on track to achieve profitability in its first full year of commercial operations, a rare milestone in agri-robotics driven by a one-time purchase model with no subscription fees and deployment across farms in India and the United States.

🤖 Your robotics thought for today:

1X just hired a Luma AI researcher and spun up a whole lab because their NEO bot handled brand-new tasks without any training. That's the zero-shot execution bet — same architecture shift that made ChatGPT work. If they're right, we just skipped years of task-specific programming. I'm watching whether other humanoid makers follow this playbook or double down on fine-tuning.

Enjoy your weekend,
Uli

1X launches World Model Lab for humanoid autonomy

Great! Check your inbox and click the link to confirm.
Please enter a valid email address.