1X Neo Ships to Homes: Inside the Hayward Factory Opening and the Teleop-to-Autonomy Bet

The headline: On May 1, 2026, 1X Technologies opened its NEO factory in Hayward, California — billed as America’s first vertically integrated humanoid robot factory. NEO is now in production and shipping to consumer homes at $20,000 Early Access (priority delivery in 2026) or a $499/month subscription model. Every NEO rolling off the line runs the latest NVIDIA Jetson Thor as its onboard “Cortex” brain. But the most interesting part of the Neo story isn’t the factory or the hardware. It’s the controversial design choice at the heart of how Neo operates: when the robot doesn’t know how to do something, a 1X human teleoperator does it for you. That’s not a bug. That’s the business model.

🔗 See the robot: 1x.tech/discover/neo-home-robot — 1X Technologies’ official Neo product page with specs, pricing, and the order flow.

The Hayward factory opening

1X Technologies — the Norwegian-American AI robotics company backed by OpenAI — opened its NEO factory in Hayward, California on May 1, 2026. The announcement framing: “America’s first vertically integrated humanoid robot factory, with consumer shipments planned for 2026.”

The factory specifics:

• Located in Hayward, CA — Bay Area, US-based production
• Vertically integrated — critical components manufactured in-house
• Powered by NVIDIA Jetson Thor — every NEO ships with Jetson Thor as the core of its NEO Cortex (the robot’s brain), enabling real-time AI inference onboard for perception, reasoning, and decision-making
• Producing robots for R&D, internal home testing, and first customer shipments in 2026

The vertical-integration framing matters because it puts 1X in the same structural-advantage tier as Tesla (Fremont line conversion) and Figure (BotQ). Most humanoid programs rely on third-party actuator suppliers, third-party batteries, third-party chips. 1X manufacturing its own critical components in the US is a deliberate hedge against supply-chain disruption and a way to keep iteration cycles short — customer feedback flows into the next production batch without waiting on outside vendors.

The factory shipped its first units in early May. The first customer deliveries to consumer homes — the actual proof-point that 1X has been promising since opening preorders in late October 2025 — are targeted for mid-2026.

Specs and what NEO actually is

NEO’s hardware is unusually well-suited for the home, in a way that none of the factory-first humanoids (Figure 03, Tesla Optimus, Atlas, Digit) currently are.

• Height: 5’5” (1.65 m) — fits through standard doorways, navigates stairs
• Weight: 66 lbs (29.94 kg) — light enough to be safe around people and pets
• Lifting capacity: 150 lbs static, 55 lbs dynamic carry — strong enough to be useful
• Hands: 22 degrees of freedom — human-level dexterity, matching Tesla Optimus Gen 3 hands
• Body: Head-to-toe soft 3D-lattice polymer covering — compliant, safe for contact
• Actuation: Tendon-driven transmission with high-torque-density motors — produces smooth, compliant motion with less jerk and lower impact forces
• Noise level: 22 dB — quieter than a modern refrigerator. This is a spec no other humanoid company publishes because no one else is thinking about it.
• Connectivity: Wi-Fi, Bluetooth, optional 5G, mobile app control, three-stage speaker in pelvis and chest (yes, it doubles as a home audio system)
• Colors: Tan, Gray, Dark Brown — chosen to be furniture-friendly

The 22 dB noise level is the single most-discussed specification, and rightly so. A Boston Dynamics Atlas in operation is north of 70 dB. A Figure 03 in normal operation is high 50s. For factory deployments that doesn’t matter; nobody cares if a robot is loud on a Hyundai assembly line. For home deployments, the noise floor is everything. If your robot wakes the baby every time it moves, you don’t have a home robot. You have a different way to lose customers.

The soft 3D-lattice polymer body is the second design choice that signals “home-first.” Factory humanoids are armored to handle bumps from industrial equipment. Neo is upholstered to handle bumps from couches, doorframes, and pets. Different threat model, different design.

Pricing and availability

The pricing model is unlike anything else in the humanoid market:

• Early Access at $20,000 — one-time purchase, priority delivery in 2026, $200 refundable deposit at preorder
• Subscription model at $499/month — shipped at a later date than Early Access; the rent-rather-than-own option
• Available through 1X’s online store at 1x.tech/order

This is the first humanoid robot anyone can actually buy as a consumer in 2026. Tesla Optimus is internal-deployment only. Figure 03 is BMW-only. Atlas is Hyundai-and-DeepMind-only. Digit is RaaS-only at industrial customers. Neo is the only one with a “click here, enter your address” purchase flow.

The $20K target matches what Tesla and Figure are aiming at as a long-term consumer price, but 1X is the first to actually hit it for an actual product available to actual consumers. That’s the structural story of 1X right now: it’s not the most advanced, but it’s first to market in the consumer segment.

The teleoperation bet

Here’s where Neo gets controversial. NEO is not a fully autonomous robot in 2026. Industry estimates put its launch autonomy at 60-70% — meaning roughly a third of tasks it encounters require human intervention. For these tasks, Neo has “Expert Mode”: a human teleoperator at 1X takes over remotely, completes the task, and the robot learns from the demonstration.

The customer-facing framing from 1X’s own announcement: “For any chore that NEO doesn’t yet know, owners could schedule a 1X teleoperator to guide it — helping the robot learn while getting the job done. From Day 1, NEO will perform functions like opening doors for guests, fetching items, and turning off the lights at night. It will grow in abilities with every software update.”

This is controversial for an obvious reason — a human stranger looking through your robot’s cameras into your house is a privacy concern. 1X has thought about this and built in:

• “No-go” zones — owners can designate rooms or areas the robot is not allowed to enter, ever
• Face blurring — Neo can blur faces of people it interacts with, on by default
• Explicit user consent — teleoperator access requires the owner to actively schedule and approve sessions
• Encrypted streams — all teleoperator video feeds are encrypted

But the more interesting framing comes from outside 1X. The Robot Report’s analysis in late 2025 made the case that teleoperation is not a step backward but actually the best route to reliable autonomy. The argument: no humanoid robot in 2026 can complete a full household chore from start to finish — not laundry, not dishwashing, not cleanup. Pretending otherwise is what’s been wrong with humanoid demos for the last three years. By embracing teleoperation as a business model, 1X gets two things at once: (a) recurring revenue from a service that actually works today, and (b) a continuous stream of high-quality demonstration data that trains Neo toward real autonomy.

It’s exactly the model that’s worked for autonomous vehicles. Waymo and Cruise both started with remote-assist teleoperators and used that human-supervised driving data to train the autonomy. The “fully autonomous” framing came later. 1X is doing the same thing for home robotics, just earlier in the public messaging.

What Neo can do today (and what it can’t)

Demoed capabilities at launch:

• Opening doors for guests
• Fetching items
• Turning off lights
• Vacuuming
• Carrying groceries
• Loading/unloading the dishwasher
• Object manipulation in the kitchen (including some cooking demos with chef Nick DiGiovanni)
• Navigating rooms in response to voice commands
• Conversational AI assistance (OpenAI integration provides language understanding)

What Neo can’t reliably do without teleoperation in 2026:

• Full laundry workflow (empty pockets → treat stains → sort → wash → dry → fold → put away)
• Cleaning up after a meal start-to-finish
• Handling unexpected situations (a pet getting in the way, a kid moving things)
• Anything involving high-precision manipulation outside its trained scope

The autonomy gap is real. 1X is not hiding it. The pitch to Early Access buyers is essentially: “Buy in now, get priority delivery, support the teleop bridge while we train the autonomy, and watch Neo get better every month via over-the-air updates.”

The OpenAI angle

The other key piece of the 1X story is its strategic relationship with OpenAI. 1X is one of the few humanoid companies with deep OpenAI backing, which provides both capital and access to OpenAI’s frontier language and reasoning models. Neo’s conversational AI is OpenAI-powered, and 1X’s broader autonomy roadmap depends on OpenAI’s research output.

This creates an interesting bet: if OpenAI’s models continue to improve at the rate they have been, Neo’s autonomy gap should close faster than Tesla’s or Figure’s, which rely on in-house models. The downside is dependency — 1X’s pace is partially set by OpenAI’s roadmap, not entirely by 1X’s own engineering.

For context on where Neo sits relative to the rest of the humanoid market, see our Top 5 Humanoid Robots of 2026 roundup. Neo ranks fifth there — last among the top 5 — specifically because of the teleoperation reliance. But it’s also the only one in the top 5 you can actually buy and have shipped to your home in 2026, which is its own kind of milestone.

The honest take

Neo is not the most advanced humanoid in the world. It’s the most advanced home humanoid that’s actually shipping. Those are very different categories.

For early adopters who understand what they’re buying — a developing platform with growing autonomy, teleoperator support for what the AI can’t yet do, monthly software updates expanding the skill set — Neo is a remarkable product. For anyone expecting a fully autonomous household helper out of the box, it isn’t there yet and 1X has been honest that it isn’t there yet.

The bet is that by 2027-2028, Neo’s autonomy gap closes substantially as the deployment-data flywheel does its job. Early customer households generate the high-quality demonstration data needed to train more general autonomy. The teleoperators move from “doing tasks for the robot” to “supervising the robot doing tasks itself” to “no longer needed for most tasks.” It’s a multi-year arc, not a one-time launch.

Bottom line

The Hayward factory opening is the moment 1X transitioned from “company with preorders” to “company shipping product.” That matters disproportionately, because consumer-facing humanoid robots have been promised for years and delivered by no one. 1X is the first.

The teleoperation model is a controversial design choice that doubles as a pragmatic business model. It will be the thing that either makes Neo successful (by ensuring tasks actually get done while autonomy matures) or holds it back (if privacy concerns dominate consumer adoption). The next 12 months of customer feedback will tell.

If you’re considering Neo: understand that you’re buying a developing product, not a finished one. If you want a humanoid robot shipping to your home address in 2026, this is the only option that exists. If you want full autonomy, wait until 2027 or 2028.