For a company trying to crack the economics of humanoid robotics, OS3 has settled on an unlikely proving ground: the hotel laundry room.
The San Francisco startup, backed by Y Combinator and run by just two founders, opened preorders in August for its H.A.L.E. 1.0 robot at a price that sits well below most rivals in the space. Businesses can reserve a unit starting under $10,000 with a $99 refundable deposit, which the company began accepting on September 3. Shipments won't arrive until fall 2026, but the early bet reflects a broader wager in the robotics industry: that the path to viable humanoids runs through radically cheaper hardware and video-trained intelligence that skips the expensive, time-intensive data collection that has stalled previous efforts.
Whether that thesis holds up remains to be seen. For now, OS3 is betting hospitality will provide the repetitive, high-volume tasks needed to prove the model works.
The robot itself is a 22-degree-of-freedom bimanual machine with a telescoping torso capable of lifting 180 pounds and reaching from floor to top shelf, according to technical specifications the company published. Each arm handles up to 15 pounds at peak load, and an omnidirectional base carries 440 pounds while fitting through standard doorways. The system runs a single whole-body control policy that OS3 describes as "tokens to torque," with planning handled in the cloud and execution onboard. No GPU rides along with the robot; battery life lasts a full workday on a standard outlet charge.
What sets OS3 apart, at least in its pitch, is the training approach. The company said it pretrained H.A.L.E. 1.0 on 150,000 hours of video spanning more than 2,000 real-world tasks. A reinforcement learning loop then refines performance using failure corrections gathered from actual deployments. New tasks typically require just two to four hours of human demonstrations via VR teleoperation or handheld gripper rigs, the company said. Field footage shows the robot folding linens, shuttling laundry loads, and packing parts.
The workflow OS3 envisions for hospitality is methodical: move linens, load washers, unload them, fold, and shelve. The company claims operating costs as low as $2.50 per hour, though detailed terms and conditions have not been published.
Rishabh Chanana and Chris Hailey make for an unconventional founding pair in robotics. Chanana studied robotics and AI at UC San Diego and later trained large language models at ServiceNow, following NLP research stints with Harvard and Georgia Tech. Hailey came from Coinbase, where he worked as a software engineer on the asset-addition team handling operations tied to more than $19 billion in trading volume, according to the company's Y Combinator profile. He holds a computer science degree from USC, earned on a Presidential Merit scholarship. As of mid-September, the startup listed just the two of them.
OS3 said it has placed units in hotels and is expanding into biotech labs, hospitals, and data centers, including a wet lab at Stanford. Deployment states include California, Michigan, and Texas, according to the company's site. These deployment claims represent company-level evidence and have not been independently verified.
For developers, OS3 offers H.A.L.E. 1.0 at $14,995 to purchase outright or $1,800 per month to lease. The pitch positions the startup as a platform provider supplying the robot, base model, and reinforcement learning pipeline, while partners build vertical applications and handle go-to-market strategy. The $99 preorder deposit routes through Stripe and can be refunded.
Pricing in the humanoid robotics space remains all over the map. Unitree's G1 launched in May 2024 at ¥99,000, roughly $13,600. Figure AI unveiled its Figure 03 model late last year but hasn't published a list price, focusing instead on enterprise deals like a manufacturing partnership with BMW. Tesla has floated long-term consumer price targets of $20,000 to $30,000 for its Optimus robot, though those figures remain aspirational and the company has yet to confirm external deployments. Apptronik suggested in 2023 that its Apollo robot might eventually cost under $50,000.
OS3's preorder agreement cautions that H.A.L.E. 1.0 is still in active development and that final specifications and delivery timing could shift. The company declined to disclose funding details beyond its Y Combinator backing. Forbes spotlighted OS3 in a September roundup of Y Combinator's summer batch, summarizing the startup's thesis: "The model is not the product, deployment is."
If that proves true, the test will come not in lab demos but in the basement laundry rooms of hotels across the country, where a $10,000 robot will either earn its keep or quietly fade into the crowded graveyard of robotics startups that couldn't scale.
