The Physical AI Power List

The people shaping what machines can do.

2026 Q3 quarter-to-date · Provisional

Elon Musk, official portrait
Image: The Royal Society

Rank 1

Elon Musk

Co-founder and CEO · Tesla

96Strategic Power /100

Joint first: unusually integrated humanoid resources, agenda influence and options balance NVIDIA’s broader enabling reach.

All 18 leaders

What changed · August 2026

Who moved, and what moved them.

2 moved up · 7 moved down · 3 held · 1 new

July 2026August 2026#1 UBTech Walker S#1 UBTech Walker S#3 Agibot A2#2 Agibot A2#2 1X NEO#3 1X NEO#5 Figure 03#4 Figure 03#5 Kepler Forerunner K3#4 Kepler Forerunner K2#6 Kepler Forerunner K2#6 NEURA 4NE-1#7 NEURA 4NE-1#7 Apptronik Apollo#8 Apptronik Apollo#9 Boston Dynamics Atlas (…#9 Boston Dynamics Atlas (…#10 Tesla Optimus#10 Tesla Optimus#8 Unitree H2#11 Unitree H2#11 Fourier GR-2#12 Fourier GR-2#12 XPeng IRON#13 XPeng IRON
  • New entrant#5
    Kepler Forerunner K3
    Kepler Robotics

    New generation·This generation continues the published history of Kepler Forerunner K2, which supplied its opening read.

  • Moved down#11-3 places down
    Unitree H2
    Unitree · score unchanged

    Other robots moved·Unitree H2 moved from #8 to #11. Its Index score did not change; other robots moved.

  • Moved down#6-2 places down
    Kepler Forerunner K2
    Kepler Robotics · score unchanged

    Other robots moved·Kepler Forerunner K2 moved from #4 to #6. Its Index score did not change; other robots moved.

  • Moved up#2+1 places up
    Agibot A2
    Agibot (Zhiyuan Robotics) · score +0.15

    New evidence·New public evidence was added for working with people (2 accepted records).

Capability landscape · August 2026

The whole field on one plot.

0.300.640.981.321.661%7%13%19%25%Median of this month’s ranked cohort. Not a threshold.MEDIANMedian of this month’s ranked cohort. Not a threshold.MEDIANCAPABILITY (cohort range of 0–5)EVIDENCE (cohort range of 0–100%)Relative mode magnifies differences inside the current market. Absolute scores do not change.RELATIVE · CURRENT COHORTTesla Optimus: capability moved from 0.63 to 0.73 since the previous editionBoston Dynamics Atlas (electric): capability moved from 0.63 to 0.75 since the previous editionAgibot A2: capability moved from 1.25 to 1.40 since the previous editionFourier GR-2: capability moved from 0.58 to 0.57 since the previous editionFigure 03: capability moved from 1.05 to 1.20 since the previous editionApptronik Apollo: capability moved from 0.75 to 0.98 since the previous editionNEURA 4NE-1: capability moved from 0.98 to 1.00 since the previous editionUBTech Walker S: capability moved from 1.43 to 1.48 since the previous edition1X NEO: capability moved from 1.25 to 1.27 since the previous editionXPeng IRON: capability moved from 0.45 to 0.48 since the previous edition2Agibot A21.40▲0.154Figure 031.20▲0.151UBTech Walker S1.48▲0.0531X NEO1.27▲0.025Kepler Forerunner K31.18new

13 ranked

Select a robot

One evidence base · Three ways to use it

See the ranking

Every humanoid robot we track, ordered by what it can actually be shown to do.

Open the ranking Compare two robots

Check realness

Tell a demo from a deployment. Each score is linked to the public source behind it.

See the record How scores are built

Find fit

Start from the job you need done, then shortlist the machines that can realistically do it.

Find a robot
The current read · August 2026

A closer look at one robot.

UBTech

UBTech Walker S

Capability score
1.48
Proof score
0.11
Proof gap
1.36
12345Locomotion4/5 · 1.60Manipulation4/5 · 1.60Perception4/5 · 1.60Autonomy5/5 · 1.80Learning1/5 · 0.40HRI5/5 · 2.00Deployment5/5 · 2.00Safety2/5 · 0.80Capability score1.48Proof score 0.11
Spider view · 8 axes

One spoke per axis. Dashed graphite is the Proof score; filled magenta is the Capability score. Gaps in either polygon mark axes that are not evidenced — never zeros.

0 of 8 axes not evidenced
Hover a spoke or axis label to preview it here. Click to open the axis lens.

30 of 40 sub-axes carry scored evidence in this edition.

Open the UBTech Walker S report

Humanoids

What a humanoid actually does, step by step

  1. Step 01

    Sees and senses

    Cameras, microphones and touch sensors turn the room into one picture the robot can use.

  2. Step 02

    Understands the task

    It works out what it has been asked to do, and which objects in front of it matter.

  3. Step 03

    Plans ahead

    Before moving, it estimates what will happen next — where things fall, slip or get in the way.

  4. Step 04

    Moves and handles

    It chooses the actual movement: which hand, which grip, how much force.

  5. Step 05

    Stays stable

    Hundreds of times a second it adjusts the whole body so the work does not topple it.

  6. Step 06

    Learns from experience

    Between runs, recorded attempts are used to retrain the models so the next attempt is better.

Evidence

From vendor claim to real work

Six steps of proof
Each step is harder to fake than the one before it
NONEFULL01Vendor says02Vendor shows03Customertrial04Confirmed customeruse05Measured atwork06Repeated acrosssites
Step 4 · Confirmed customer use

A real customer or site is named on the public record.

To move up: published output, uptime or task metrics.

No tracked record sits at this step with a public source we can link to yet, so no example is shown.

Deployments tracked
Public records

No sourced deployment records are available here yet.

Careers

Who is hiring to build this.

Four career paths

  • Intelligence & Autonomy
  • Motion & Hardware
  • Systems, Validation & Deployment
  • Product & Delivery

One useful email a month.

No trackers. Unsubscribe anytime.