1X TECHNOLOGIES
1X NEO
First consumer-ready home robot. 29 kg, quieter than a refrigerator. US deliveries from 2026. $20,000 outright or $499 / month.
Chapter III · A Briefing
The state of Physical AI, and the question it leaves open.
A short orientation, prepared for principals and engineering leaders who are evaluating humanoid platforms for residential service. What the technology now does, what it does not yet do, and the gap our work is designed to close.
§ 01
From rigid automation to adaptive autonomy.
Classical robots, the welding arm in a car plant, execute fixed instructions. A minor deviation breaks the routine. Physical AI is a new category: systems that perceive their environment, reason about it, and act in the unstructured physical world.
The shift, popularised by NVIDIA at CES 2025, is from programmed behaviour to learned competence, models that grasp gravity, friction, balance, and touch.
§ 02 · How a humanoid learns to fold a shirt
Three ingredients, none of them sufficient alone.
I
Simulated worlds
Platforms such as NVIDIA Omniverse model gravity, friction and collision. A robot can attempt millions of grasps without breaking hardware.
II
Real sensor data
Simulation cannot capture wet cloth or the give of a tomato. Cameras, LiDAR and tactile sensors record the physics that maths alone misses.
III
Imitation of humans
Companies film thousands of contractors across 50+ countries doing laundry and dishes, head-mounted iPhones become the most valuable training corpus in robotics.
$6B+ INVESTED IN HOUSEHOLD-TASK TRAINING DATA IN 2025 ALONE
§ 03 · The market in 2026
Three platforms now in or approaching residences.
TESLA
Optimus Gen 2
Demonstrates autonomous grocery handling, plant care, even board games. Trained on human video via neural networks.
FIGURE AI
Figure 03
Folds laundry and loads the dishwasher under the Helix VLA model. Named one of TIME's Best Inventions of 2025.
§ 04 · Capability matrix
What works today, what does not.
No system in 2026 reliably and unsupervised performs the full spectrum of household duties a person handles in a day. Honest assessment matters.
TASK
STATUS
CONSTRAINT
Folding laundry
PARTIAL
Possible, unreliable on unfamiliar garments
Loading the dishwasher
PARTIAL
Assisted, starting the cycle still requires a human
Watering plants
DEMONSTRATED
Demonstrated in structured environments
Unloading groceries
DEMONSTRATED
Demonstrated for known packaging
Climbing stairs
PARTIAL
Lab-stage; not reliable on unfamiliar staircases
Grasping unknown objects
PARTIAL
Error-prone with irregular shapes
Fully autonomous cleaning
NOT READY
Not market-ready; unstructured spaces overwhelm current systems
§ 05 · The horizon
"By 2060, forecasts suggest as many as three billion humanoid robots may coexist with humans, most of them in households, as personal assistants."
RECORDED FUTURE RESEARCH · 2025
§ 06 · The unresolved question
The next challenge is no longer primarily technical.
Hardware works. Models learn. The remaining questions concern safety, privacy, a humanoid sees and hears everything in the home, social acceptance, and who retains control of these systems inside the most private spaces of human life.
These are not engineering questions. They are questions of service, decorum and discretion, the subject matter of our academy for nearly two centuries. Maison Protocol exists to translate that discipline into the layer your platform is missing.
Sources
Compiled from publicly available reporting by the World Economic Forum, EY Global, NVIDIA, Boston Consulting Group, MIT Technology Review, TIME, and Recorded Future Research. Figures and product details current as of April 2026.