1X’s new tendon-driven hand for the NEO humanoid reframes robotic dexterity as a platform primitive — and the engineering choices are really business-model choices.
1X NEO HAND — KEY SPECS
25
Degrees of Freedom (22 fully actuated fingers/palm + 3 wrist)
On July 9, 2026, 1X Technologies unveiled the new hand for its NEO humanoid robot, framing it in explicitly platform terms: an API to the physical world. The hand ships with 25 degrees of freedom — 22 fully actuated across the fingers and palm, plus 3 at the wrist — distributed anatomically, with deliberate bias toward a thumb that genuinely opposes the fingers. None of the DoF are padded with passive spring joints. Every joint does real work.
The mechanical architecture is quasi-direct-drive: motors live in the forearm and pull tendons through the wrist at gear ratios of roughly 5:1 to 15:1. That is an order of magnitude lower than the 100:1–200:1 ratios typical in industrial actuators. The consequence is a hand that is light, low-inertia, and — critically — compliant by construction. When the hand hits an unexpected object, it yields. It does not fight back.
1X says hundreds of hands have already been manufactured, with capacity to produce 10,000 units this year. The finger assemblies are validated to millions of cycles. The hand is IP68-rated and food-safe. These are 1X’s own figures, not independently verified — but the production claim is the one that matters strategically, and we’ll come back to it.
FROM CAPABILITY TO PLATFORM — THE 1X ARC
2023–2024
1X ships NEO Beta; focuses on whole-body locomotion and data collection in real homes. Hand is a known bottleneck.
2025
Hundreds of NEO hands manufactured in-house; tendon-drive + low-gear-ratio architecture validated at scale. IP68 and food-safe certifications achieved.
July 9, 2026
1X unveils the NEO hand publicly. 25 DoF, fully force-controlled, backdrivable, tactile fingertips. Framed as “an API to the physical world.” Capacity: 10,000 hands/year stated.
The Bet
Dexterity + safe compliance + manufacturing scale = the interface layer between AI and the atom economy. Whoever nails all three owns that interface.
The key insight: 1X did not just build a better robotic hand — it made every engineering constraint a business-model constraint. Force control, backdrivability, and compliance are not safety features bolted on after the fact. They are the preconditions for deploying a humanoid in a home or a hospital, which is the only path to the real prize: a general-purpose labor platform at scale.
The Structural Read
The hand is the bottleneck of embodied AI. Everything upstream — the foundation models, the locomotion stack, the sim-to-real training pipelines — collapses in value if the end effector cannot reliably grip a peach, turn a doorknob, or hand a cup to a person without crushing it. 1X understood this and attacked the bottleneck on two simultaneous axes.
Axis one: sensing as the moat, not dexterity alone. The real breakthrough in the NEO hand is not the 25 DoF — competitors can count joints. It is that all 25 DoF are natively force-controlled and fully backdrivable, meaning every joint doubles as a force sensor. The hand knows its own pose and effort through closed-loop proprioception, without relying on vision. Tactile skin on the fingertips reads contact and shear in real time for adaptive gripping. This is the sensory infrastructure that makes a hand usable in unstructured environments — the kind of environments that constitute most of the world.
Axis two: vertical integration as the manufacturing moat. 1X builds the motors, custom electronics, and tendon systems in-house. This is the same own-the-whole-stack pattern that defines the current AI moment — from GPU-to-model at Nvidia, to silicon-to-inference at Apple, to custom ASIC-to-training at Google. The physical layer is now undergoing the same verticalization. Hundreds of hands made already, 10,000/year capacity stated. That is not a prototype cadence. That is a supply chain being built before demand fully arrives — a classic platform move.
In the Map of AI framework, control points shift as the stack matures. The physical layer — actuators, end effectors, sensors — is the emerging control point of embodied AI. Just as the API layer became a monetization surface for software AI, the hand-as-API becomes the monetization surface for physical AI. 1X’s framing is not marketing language. It is a claim about where the rent will be collected.
The choice to make compliance structural — baked into the gear ratio and the tendon drive, not layered on through software — is also a regulatory and liability choice. A hand that yields when it encounters a human is a hand that can pass safety certification in healthcare settings, eldercare, food handling. The compliance is what opens the addressable market. It is the unlock for every use case that requires physical proximity to people.
1X Technologies
“An API to the physical world.”
That phrase is doing a lot of work. An API is not a finished product — it is an interface layer that other systems build on top of. 1X is signaling that the NEO hand is infrastructure, not a point solution. The downstream applications — logistics, eldercare, manufacturing, home assistance — are the apps. The hand is the platform. The question is whether 1X can defend that platform position before larger players verticalize their own physical stacks.
Three Implications
IMPLICATION 1 — THE VERTICAL INTEGRATION RACE REACHES THE ATOM LAYER
1X’s decision to manufacture motors, electronics, and tendons in-house mirrors the stack-consolidation moves that defined the software AI era. The winners of embodied AI will not be systems integrators assembling commodity parts — they will be the companies that own the full physical stack from actuator to intelligence. Production capacity becomes the moat, not just IP. 10,000 hands per year (1X’s figure) is a number that starts to matter when deployment scales.
IMPLICATION 2 — FORCE CONTROL IS THE NEW CONTEXT WINDOW
In large language models, context window size became the proxy metric for capability. In physical AI, force-sensing fidelity and backdrivability will play the same role. A hand that knows what it is touching — without vision, through proprioception alone — can operate in the dark, in gloves, underwater, in dusty warehouses. The sensory richness of the hand is what determines which tasks are in or out of scope. 1X just raised the baseline.
IMPLICATION 3 — SAFE COMPLIANCE IS A MARKET-ACCESS STRATEGY, NOT A FEATURE
The humanoid market segments sharply by regulatory environment. Industrial settings can tolerate caged, high-force robots. Homes, hospitals, and schools cannot. Compliance-by-construction — built into the gear ratio, not patched in software — is what crosses that threshold. 1X is not just making a safer hand; it is qualifying for the largest and most defensible TAM in physical AI. Every competitor shipping high-gear-ratio hands is, by definition, locked out of those markets.
Business Engineer Framework
The Map of AI — Physical Layer as Control Point
The Map of AI tracks where control points form as the stack matures. The argument in 1X’s NEO hand is that the physical interface — the end effector, the sensor layer, the actuator architecture — is becoming the next control point, just as the model API was the control point of software AI. Whoever owns the interface between AI and atoms collects the rent. This piece maps exactly where 1X is trying to plant its flag.
1X just drew a line between two eras of robotics: the one where dexterity was the goal, and the one where safe compliance at manufacturing scale is the business. The NEO hand’s 25 force-controlled, backdrivable degrees of freedom are not a spec-sheet flex — they are the engineering expression of a market thesis: that the humanoid prize goes to whoever builds the interface layer between AI and the physical world, owns the stack from motor to fingertip, and ships enough units to learn faster than anyone else. Every engineering decision in this hand is also a strategic decision. That is what makes it worth paying attention to.
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.
