1X Technologies just demonstrated finger-level dexterity in Neo — and the deeper story is not about hardware. It’s about who captures value when robots stop being prototypes and start being platforms.
Wired’s July 2026 coverage of 1X Technologies’ Neo robot surfaced something the company’s own press releases undersell: Neo’s hands move fast — uncommonly, unsettlingly fast — with individual finger articulation that puts it ahead of most competitors on the single capability that has historically made humanoid robots useless in unstructured environments. The demo showed Neo manipulating small objects, adjusting grip mid-motion, and recovering from slip — all in real-time, without pre-scripted choreography.
1X, the Norwegian robotics company backed by OpenAI’s corporate fund, has taken a deliberately different path from Figure AI and Agility Robotics. Where competitors have chased bipedal locomotion headlines, 1X has quietly built around manipulation dexterity and data flywheel economics. Its EVE platform — a wheeled, torso-forward design — has been generating real-world training data from warehouse deployments since late 2023. Neo is the next-generation body that inherits that data advantage.
The timing matters. Tesla’s Optimus team, Physical Intelligence (pi), and Apptronik are all converging on the same 2026–2027 window for general-purpose manipulation. 1X is not just shipping a better robot — it is trying to establish a data moat before the hardware commodity wave arrives.
1X TECHNOLOGIES — CAPABILITY TIMELINE
2022 — Company Founded
1X (then Halodi Robotics) pivots to humanoid-adjacent form factors; OpenAI fund interest begins
2023 — EVE Deployment
First commercial wheeled units enter warehouse environments; real-world dexterous manipulation data collection begins
2024 — $125M Series B
OpenAI-led round; Neo humanoid design revealed; 1X positions as the “home robot” contender
July 2026 — Neo Dexterity Demo
Wired documents finger-speed breakthrough; home deployment pilots confirmed for H2 2026
The key insight: Neo’s fast fingers are not the product — they are the proof that 1X’s data flywheel is working. Every hour EVE spends in a warehouse is training data that teaches Neo how to handle the physical world. The hardware demo is a data-infrastructure announcement in disguise.
The Structural Read
The humanoid robot race is being narrated as a hardware competition. That framing will age badly. What 1X is actually building is a Product Overhang — capability that has been accumulating invisibly in EVE’s warehouse deployments, and that surfaces all at once in Neo as a seemingly sudden leap in dexterity.
This is the same dynamic that made GPT-4 look like a discontinuity when the underlying scaling had been linear for years. 1X’s EVE units have been generating proprietary manipulation telemetry in real commercial environments — data that Figure AI, Apptronik, and even Tesla’s Optimus cannot replicate in a lab. When Neo ships into homes, it will not be starting from scratch. It will be inheriting thousands of hours of contact-rich, real-world dexterous behavior.
The competitive moat 1X is building is not the hand design. It is the dataset that trained the policy model running inside that hand. Hardware commoditizes. Proprietary behavioral data does not.
Product Overhang Doctrine
“Capability builds invisibly until it surfaces all at once. The demo that looks like a breakthrough is almost always the moment a hidden accumulation becomes visible. 1X’s Neo is not a step forward — it is the revelation of a staircase that was already climbed.”
There is a second structural layer here: OpenAI’s investment in 1X is not passive. OpenAIneeds physical-world grounding data to extend its models beyond text and image. 1X needs foundation model intelligence to make its manipulation policies generalizable. This is a bilateral data dependency dressed up as a venture bet. As Neo deploys into homes, the behavioral data it generates feeds back into OpenAI’s physical AI research — a loop that neither party has fully disclosed but that makes 1X structurally different from every other robotics startup raising on vibes.
Three Implications
IMPLICATION 1 — DATA MOATS ARRIVE IN ROBOTICS BEFORE HARDWARE MOATS DO
The company that deploys the most robots in real environments earliest will accumulate the most diverse manipulation training data. 1X’s EVE-to-Neo pipeline is a proof of concept for this flywheel. Competitors who skip the “boring” commercial deployment phase to chase consumer headlines will find themselves training on synthetic data against a rival training on reality.
IMPLICATION 2 — OPENAI’S PHYSICAL AI STRATEGY IS MORE COHERENT THAN IT APPEARS
OpenAI’s 1X investment, combined with its robotics API ambitions and embodied AI research, forms a physical-world data acquisition strategy. The $125M is not just a financial bet on humanoids — it is an option on proprietary real-world grounding data that no language model training corpus can provide. If physical AI becomes a distinct capability layer, OpenAI has a seat at the table that Google DeepMind and Anthropic currently do not.
IMPLICATION 3 — THE HOME ROBOT MARKET WILL BE WON ON TRUST, NOT SPECS
Dexterous fingers will get Neo into product reviews. They will not get it into living rooms. The critical variable for consumer humanoid adoption is not capability — it is the permission architecture: what data does the robot collect, who owns it, and how is it governed. 1X has not yet published a serious answer to these questions. The company that builds a credible privacy and data-sovereignty model for in-home robots will have a durable consumer moat that no hardware spec can replicate.
Business Engineer Framework
The Map of AI — Where Does 1X Sit in the Stack?
1X Technologies operates at the intersection of three layers in the Map of AI: the model layer (policy networks for manipulation), the data layer (proprietary real-world telemetry), and the application layer (commercial and consumer deployment). Understanding which layer a company controls — and which it rents — determines where margin will concentrate as the robotics stack matures. The Map of AI framework maps 200+ companies across 9 layers to show exactly who wins when commoditization hits.
Neo’s freaky-fast fingers are real, and they matter — but the story 1X is actually telling is about a data flywheel that has been running quietly in warehouses for three years, and that is now surfacing as a capability lead that looks, to the outside world, like a sudden hardware breakthrough. The companies that win the physical AI era will not be the ones with the most elegant robot designs. They will be the ones that understood, early, that the body is just the sensor for collecting the data that trains the mind.
Gennaro is the creator of FourWeekMBA, which reached about four million business people, comprising C-level executives, investors, analysts, product managers, and aspiring digital entrepreneurs in 2022 alone | He is also Director of Sales for a high-tech scaleup in the AI Industry | In 2012, Gennaro earned an International MBA with emphasis on Corporate Finance and Business Strategy.
Scroll to Top
Discover more from FourWeekMBA
Subscribe now to keep reading and get access to the full archive.
