1X's 10,000 NEO preorders put home robotics data to the test

Why it matters

1X's 10,000 NEO preorder claim is less a sales victory than a test of whether consumer humanoid startups can turn early homes into proprietary training data without burning trust on privacy, safety or teleoperation costs.

Robert Scoble is making the sharpest case for 1X Technologies in consumer robotics: the important number is 10,000 NEO preorders, because a fleet in early adopters' homes could give Bernt Oivind Bornich's team the kind of real-world data loop that humanoid robotics has spent years trying to manufacture in labs.

Robert Scoble on X

The number is real enough to matter, but narrower than Scoble's wording. 1X launched NEO preorders on October 28, 2025, and Forbes reported it booked its first-year production capacity in five days, equal to 10,000 NEOs. That is a preorder and capacity milestone, not evidence that 10,000 robots have been delivered or that 1X has recognized hardware revenue from them.

Scoble's extra claim, that many of those buyers are "AI people in San Francisco," remains an assertion from Scoble. 1X has not published a buyer breakdown by geography, occupation, or customer type. The claim still lands because it matches the thing 1X has been building toward since Bornich shifted the company toward the home: a user base willing to live with a partially autonomous robot while 1X learns from the edge cases.

Bornich founded 1X in 2014 as Halodi Robotics with the goal of building general-purpose robots that can coexist with humans. 1X previously built EVE, a wheeled humanoid for enterprise environments, then pivoted to NEO with a home focus. The founder's bet has been consistent: if androids are going to work in human spaces, they need to spend time in human spaces.

That is why the preorder base matters more than another chore demo. A humanoid robot in a kitchen faces a different problem than a robot arm in a cell or a mobile robot in a warehouse. Homes are cluttered, private, unstructured and full of low-frequency tasks. The data is valuable because it is messy.

The preorder milestone is old, but the delivery test is current

Scoble posted his comment on July 10, 2026, but the underlying NEO preorder milestone dates to late October 2025. 1X later used that milestone to explain why it was building manufacturing capacity in California. On April 30, 2026, 1X announced that its NEO factory in Hayward had started full-scale production, with capacity for 10,000 NEOs per year and a plan to reach more than 100,000 units per year by the end of 2027.

The order page now frames the commercial offer clearly: NEO costs $20,000 to own or $499 per month, with a $200 refundable deposit and U.S. deliveries starting in 2026. 1X describes NEO as a 5-foot-6-inch, 66-pound home robot with a 154-pound lift rating, 55-pound carry rating, four-hour runtime, dual stereo fisheye cameras, and Nvidia Jetson Thor compute.

The hardware specs are only one side of the product. 1X markets NEO as an AI system that learns household chores over time. For tasks NEO does not know, 1X offers scheduled Expert Mode, where a 1X human expert can remotely supervise the robot and help complete the work.

That design makes the data strategy explicit. NEO arrives with basic autonomy, then gains capability through use, supervision and training. The first buyers are paying for a robot, but they are also joining the training process.

Bornich is selling the loop

1X's clearest public statement of the strategy came on June 4, 2026, when 1X announced its World Model Lab and hired Sam Sinha, previously a founding research scientist at Luma AI, as founding AI researcher and head of world models. 1X said the lab would train embodied world models using web-scale media, egocentric human video, simulation, remote-operated robot data and data from deployed NEO robots. The company argues that world models will help humanoids generalize, adapt, and self-learn, and that it must own the stack down to pretraining its own video foundation models.

Scoble's data-flywheel framing sits inside that strategy. The comparison to Tesla is useful only up to a point. Tesla's vehicle fleet gave Tesla a massive stream of real-world driving data, but homes are a harder deployment surface for consumer trust. A car looks outward onto public roads. A home robot looks inward, through cameras and microphones, at kitchens, bedrooms, pets, children and private routines.

TechCrunch reported in March 2025 that 1X planned to test NEO Gamma in a few hundred to a few thousand homes, with Bornich saying early adopters would help teach the system how to behave. The same report said 1X expected to rely on teleoperators for in-home testing and that customers would decide when a 1X employee could view a robot's surroundings for auditing or teleoperation. In December 2025, TechCrunch again pointed to the privacy issue around human operators being able to look through a NEO's eyes, while noting 1X had struck a deal to make up to 10,000 NEO robots available to EQT portfolio companies between 2026 and 2030.

Those details do not undercut the strategy. They define the product risk. If NEO works, 1X gets a proprietary stream of household interaction data that competitors cannot scrape from the web. If NEO needs too much human help, the operating cost and privacy burden could overwhelm the subscription economics.

The humanoid market is moving toward fleets

1X is not alone in making data collection the center of the humanoid race. Figure announced in September 2025 that it had secured more than $1 billion in Series C commitments at a $39 billion post-money valuation, with capital earmarked in part for scaling humanoid robots into homes and commercial operations, GPU infrastructure and advanced data collection. Apptronik raised $350 million in February 2025 to scale production of Apollo, with industrial customers in automotive, logistics and manufacturing.

1X's distinction is the home-first emphasis. Bornich is asking customers to accept a robot before the robot can do everything autonomously. That is a founder bet as much as a product roadmap: ship early into the hardest environment, collect the data, close the loop, and use manufacturing scale to make iteration faster.

The capital stack reflects that ambition. 1X raised a $23.5 million Series A2 led by the OpenAI Startup Fund in 2023, with participation from Tiger Global, Sandwater, Alliance Ventures and Skagerak Capital. In January 2024, 1X announced a $100 million Series B with participation from EQT Ventures, Samsung NEXT, Skagerak Capital and Nistad Group, saying it had raised more than $125 million in less than 12 months.

The next proof point is not another preorder count. It is whether U.S. customers actually receive NEOs in 2026, whether those robots can complete useful household work without constant hand-holding, and whether early buyers tolerate the data collection that makes the product improve.

Scoble is right about the important question, even if the buyer-mix claim is unverified. In consumer humanoids, the moat will not come from a keynote robot folding one shirt. It will come from deployed robots learning from thousands of real homes without losing the trust of the people who let them in.