1X starts shipping NEO humanoid robots to US homes


The Norway-founded company’s vertically integrated NEO factory in Hayward marks the first US-scale push to put a general-purpose humanoid robot into private homes, with shipments planned this year and a competitive field that is already crowded


1X Technologies has opened a 58,000-square-foot manufacturing facility in Hayward, California, to produce its NEO humanoid robot at consumer scale, with capacity for 10,000 units in year one and a target of more than 100,000 units annually by the end of 2027.

The Norway-founded, OpenAI-backed company described the plant as the first vertically integrated humanoid robot factory in the United States. First customer shipments are planned for 2026.

The factory currently employs more than 200 staff and is scaling. NEO is being manufactured with critical components built in-house, motors, batteries, structures, transmission systems, soft goods, and sensors, in a configuration the company describes as bottom-up American manufacturing.

The 💜 of EU tech

The latest rumblings from the EU tech scene, a story from our wise ol’ founder Boris, and some questionable AI art. It’s free, every week, in your inbox. Sign up now!

“This is more than just a factory opening, it’s proof that the future of humanoid robotics is being built right here in the U.S.,” 1X CEO and founder Bernt Børnich said in the announcement.

The Hayward facility is intended as a stepping stone to a larger plant under construction in San Carlos, California.

The product

NEO is positioned as a general-purpose home robot, designed to operate alongside humans in domestic environments rather than as an industrial bipedal for warehouses or factory floors.

The robot is available in three colours (Tan, Gray, and Dark Brown) and offered through two commercial models: an Early Access purchase at $20,000 with priority delivery in 2026, or a subscription at $499 per month.

NEO is powered by Nvidia’s Jetson Thor onboard computing platform and trained using Nvidia’s Isaac open robotics simulation framework.

Demand has reportedly outstripped initial expectations. The company says first-year production capacity sold out within five days of preorders opening in October 2025. 

1X raised $100 million in its push to bring NEO to market; the robot is designed with explicit safety constraints: it is light, soft to the touch, and configured without pinch-points or other hazards, a deliberate choice given the company’s ambition to deploy in private homes rather than industrial settings, where heavier and harder humanoids dominate.

NEO learns household tasks through embodied AI, the technique under which robots acquire skills by interacting with their environment. Customers can also manually demonstrate tasks using a VR headset and controllers, and the robot includes conversational functionality that Børnich has compared to ChatGPT.

Whether those capabilities translate to reliable performance across the variety of unstructured tasks a real home presents, the open question for every consumer humanoid, is something the customer shipments later this year will start to answer.

Two routes to market

Beyond the consumer product, 1X has structured its commercial strategy around a parallel enterprise track. In December 2025, the company struck a partnership with private equity firm EQT to deploy up to 10,000 NEOs to companies in EQT’s portfolio between 2026 and 2030 across facility operations, manufacturing, logistics, and healthcare.

The two-track structure gives 1X higher-margin enterprise revenue from its early units, plants and warehouses pay full price for performance, while the home model can scale down in cost over time. It is the same playbook electric vehicles followed, with luxury and commercial customers subsidising the consumer rollout.

The 10,000-unit annual production target is a meaningful number in a field where most humanoid robot makers are still measured in the hundreds. Tesla, however, is the comparison that matters most. Tesla’s China president Wang Hao described the Shanghai Gigafactory as a “golden key” to mass-producing the company’s Optimus humanoid robot. Tesla has discussed manufacturing a few hundred Optimus units in 2026, scaling to thousands and then tens of thousands annually by 2027 and 2028, with internal targets of one million units per year from Shanghai that have not been confirmed in any public filing.

Elon Musk’s long-stated goal is pricing Optimus below $20,000 per unit, the same price point at which 1X is selling Early Access NEOs today.

China’s humanoid robotics sector is moving rapidly in parallel. Unitree’s G1 and H1 robots are commercially available at price points well below Tesla’s indicated targets. Agibot, UBTECH, Fourier Intelligence, and a growing roster of Chinese startups are all targeting the same market.

China’s central and local governments have identified humanoid robots as a strategic technology, with subsidies and policy support that other regions have been slower to match. The competitive dynamic places 1X’s American-manufactured, vertically integrated approach against Chinese state-backed scale and Tesla’s automotive manufacturing infrastructure simultaneously.

Europe is also building. Neura Robotics, founded in Germany in 2019, has scaled to more than 600 employees and raised €120 million in January 2025. Founder David Reger has told TNW he sees Tesla as his only real competitor in the segment.

Europe’s humanoid robotics sector is positioning regulatory clarity, the AI Act, the updated Product Liability Directive, the General Product Safety Regulation, and the Machinery Regulation, as a competitive advantage, on the argument that investors and industrial partners commit resources where compliance risks are predictable.

The factory opening is the easy part. Manufacturing a humanoid robot at scale, while difficult, is fundamentally a known engineering problem with known suppliers and known cost curves. The harder question, the one no manufacturer has yet definitively answered, is whether a general-purpose home robot can perform the variety of unstructured tasks a private home demands at a level customers will pay $20,000 or $499 a month for. 

1X’s answer to that question is, in part, to ship and iterate. Robots produced at the Hayward facility are currently being routed to internal testing, validation, and the company’s own R&D Lab and Internal Home Testing programmes before customer deliveries begin.

The vertically integrated manufacturing approach was chosen specifically to enable rapid hardware iteration as feedback comes in. If that iteration speed is fast enough to close the gap between the demonstrations on the launch reel and the messy reality of the average American home is, ultimately, the bet behind the entire factory.



Source link

Leave a Reply

Subscribe to Our Newsletter

Get our latest articles delivered straight to your inbox. No spam, we promise.

Recent Reviews


The Samsung Keyboard supports glide typing, voice dictation, multiple languages, and deep customization through Good Lock. On paper, it’s a very capable and perfectly functional keyboard. However, it’s only when I started using it that I realized great features don’t necessarily translate to a great user experience. Here’s every problem I faced with the Samsung Keyboard, and why I’m permanently sticking with Gboard as my main Android keyboard.

I have been using Gboard and the Samsung Keyboard on a recently bought Galaxy S24, which I got at a massive discount.

Google’s voice typing doesn’t cut me off mid-sentence

Fewer corrections, fewer cutoffs, faster dictation

I might be a professional writer, but I hate typing—whether it’s on a physical keyboard or a virtual one. I type slower than I think, which I suspect is true for most people. That becomes a problem when I have multiple ideas in my head and need to get them down fast. It’s happened far too often: I start typing one idea and forget the other. Since jacking my brain into a computer isn’t an option (yet), I’ve been leaning more and more on voice typing as the fastest way to capture my thoughts.

Now, both Samsung Keyboard and Gboard support voice typing, but I’ve noticed that Gboard with Google’s voice engine is just better at transcription accuracy. It picks up on accents flawlessly and manages to output the right words. In my experience, it also seems to have a more up-to-date dictionary. When I mention a proper noun—something recently trending like a video game or a movie name—Samsung’s voice typing fails to catch it, but Google nails it.

That said, you can choose Google as your preferred voice typing engine inside Samsung Keyboard, but it’s a buggy experience. I’ve noticed that the transcription gets cut off while I’m in the middle of talking—even when I haven’t taken a long pause. This can be a real problem when I’m transcribing hands-free.

Gboard offers a more accurate glide typing experience

Google accurately maps my swipe gestures to the right words

Voice typing isn’t always possible, especially when you’re in a crowded place and want to be respectful (or secretive). At times like these, I settle for glide (or swipe) typing. It’s generally much faster than tapping on the keyboard—provided the prediction engine maps your gestures to the right word. If it doesn’t, you have to delete that word, draw that gesture again, or worse—type it out manually.

Now, both Samsung Keyboard and Gboard support glide typing, but I’ve noticed Gboard is far more accurate. That said, when I researched this online, I found a 50-50 divide—some people say Gboard is more accurate, others say Samsung is. I do have a theory on why this happens.

Before my Galaxy S24, I used a Pixel 6a, before that a Xiaomi, and before that a Nokia 6.1 Plus. All of my past smartphones came with Gboard by default. I believe Gboard learned my typing patterns over time—what word correlates to what gesture, which corrections I accept, and which ones I reject. After a decade of building up that prediction model, Gboard knows what I mean when my thumb traces a particular shape. Samsung Keyboard, on the other hand, is starting from zero on this Galaxy S24—leading to all the prediction errors. At least that’s my working theory.

There’s also the argument for muscle memory. While glide typing, you need to hit all the correct keycaps for the prediction engine to work. If you’re even off by a slight amount, the prediction model might think you meant to hit “S” instead of “W.” Now, because of my years of typing on Gboard, it’s likely that my muscle memory is optimized for its specific layout and has trouble adapting to Samsung’s.

Swiping vs typing.


Is Swiping Really Faster Than Typing on a Phone Keyboard?

Which typing method reigns supreme?

I mix three languages in one message, and Gboard just gets it

Predictive multilingual typing doesn’t get any better than this

I’m trilingual—I speak English, Hindi, and Bengali. When I’m messaging my friends and family, we’re basically code-mixing—jumping between languages in the same sentence using the Latin alphabet. Now, my friends and I have noticed that Gboard handles code-mixing much more seamlessly than Samsung Keyboard.

If you just have the English dictionary enabled, neither keyboard can guess that you’re trying to transliterate a different language into English. It’ll always try to autocorrect everything, which breaks the flow. The only way to fix this is by downloading a transliteration dictionary like Hinglish (Hindi + English) or Bangla (Latin). Both Samsung Keyboard and Gboard support these dictionaries, but the problem with Samsung Keyboard is that it can only use one dictionary at a time.

Let’s say I’m writing something in Latinized Bangla and suddenly drop a Hindi phrase. Samsung Keyboard will attempt to autocorrect those Hindi words. Gboard is more context-aware. Since my Hinglish keyboard is already installed, I don’t have to manually switch to it. Gboard can detect that I’m using a Hindi word even with the English or Bangla keyboard enabled, and it won’t try to autocorrect what I’m writing. This also works flawlessly with glide typing, which is a huge quality-of-life improvement over Samsung Keyboard.

This isn’t just an India-specific thing either. Code-mixing is how billions of people type every day—Spanglish in the US, Taglish in the Philippines, Franglais across parts of Europe and Africa.

Gboard looks good without me spending an hour on it

I don’t have time for manual customization

Samsung Keyboard is hands down the more customizable option, especially if you combine it with the Keys Cafe module inside Good Lock. You get granular control over almost every aspect of the keyboard—key colors, keycaps, gesture animations, and a whole lot more. While for some users, this is heaven, I just find it too overcomplicated and a massive time sink.

I don’t have the patience to sit and adjust every visual detail of my keyboard. Sure, it gets stale after a while, and you’d want to freshen it up, but I don’t want to spend the better part of an hour tweaking a virtual keyboard. This is where Gboard wins (at least for me) by doing less.

Android 16 brings Material 3 Expressive, which automatically themes your system apps using your wallpaper’s color scheme. With Gboard, all you have to do is change the wallpaper, and the keyboard updates to match—no Good Lock, no manual color picking. It’s a cleaner, more seamless way to keep your phone looking good without putting in the extra legwork.


The keyboard you don’t think about is the one that’s working

I didn’t switch to Gboard because Samsung Keyboard was broken. I switched because Gboard made typing feel effortless. If you’re a Samsung user who’s never tried it, it’s a free download and a five-second switch. You might not go back either.

Pixel 7 with the 8vim keyboard.


I Tried the Weirdest Android Keyboards So You Don’t Have To

Can strange layouts and gestures beat the good old-fashioned QWERTY?



Source link