Benchmarks

Less Fragile Each Time, Not a Better Demo

Progress in physical AI shows up as each deployment being less fragile than the last, not as a better demo.

Georgian + NewtonX Physical AI Survey, Wave 1 · Reporting base: Buyers, n=85 unless a benchmark states otherwise · July 24 to Aug. 21, 2026

Benchmark H1.1

What Has Changed Over 12 Months

Survey question

Compared with 12 months ago, how has each of the following changed for the systems you are closest to?

Audience
All qualified buyers, n=85
In depth
Hypothesis 1 — Less Fragile Each Time, Not a Better Demo

Six items, each answered on the same five-point scale. “Improving” below combines “Improving clearly” and “Improving slightly.”

What Has Changed Over 12 Months

79%

of robotics buyers say the share of the task the system handles on its own has improved over the past 12 months

All qualified buyers, n=85

Data for What Has Changed Over 12 Months
CategoryImproving (net)
The share of the task / workflow the system handles on its own79%
How often a human has to step in76%
How long it takes to stand up a new deployment66%
The amount of custom engineering each new deployment needs64%
How well the system recovers on its own after it gets stuck64%
The support / firefighting burden as more units are deployed58%
0%20%40%60%80%The share of the task / workflow the systemhandles on its ownThe share of the task / workflow the system handles on its own — Improving (net): 79%79%How often a human has to step inHow often a human has to step in — Improving (net): 76%76%How long it takes to stand up a newdeploymentHow long it takes to stand up a new deployment — Improving (net): 66%66%The amount of custom engineering each newdeployment needsThe amount of custom engineering each new deployment needs — Improving (net): 64%64%How well the system recovers on its ownafter it gets stuckHow well the system recovers on its own after it gets stuck — Improving (net): 64%64%The support / firefighting burden as moreunits are deployedThe support / firefighting burden as more units are deployed — Improving (net): 58%58%

All qualified buyers, n=85

Benchmark H1.2

Is Each Deployment Getting Easier?

Survey question

Overall, compared with earlier deployments, are the deployments you are closest to becoming easier to implement consistently over time, or does each new deployment require a similar level of effort?

Audience
All qualified buyers, n=85
In depth
Hypothesis 1 — Less Fragile Each Time, Not a Better Demo

Single-select, so the shares add to 100%.

Is Each Deployment Getting Easier?

81%

of robotics buyers say each new deployment is easier to implement than the ones before it

All qualified buyers, n=85

Data for Is Each Deployment Getting Easier?
Category% of buyers
Clearly easier to implement consistently than earlier deployments19%
Somewhat easier over time62%
About the same each time11%
Still requires substantial customization each time5%
Too early to tell4%
0%20%40%60%Clearly easier to implement consistentlythan earlier deploymentsClearly easier to implement consistently than earlier deployments — % of buyers: 19%19%Somewhat easier over timeSomewhat easier over time — % of buyers: 62%62%About the same each timeAbout the same each time — % of buyers: 11%11%Still requires substantial customizationeach timeStill requires substantial customization each time — % of buyers: 5%5%Too early to tellToo early to tell — % of buyers: 4%4%

All qualified buyers, n=85

Benchmark H1.3

What Operational Impact Have You Seen?

Survey question

Where you have deployed robots / physical AI, what impact have you seen over the past 12 months?

Audience
The 61 buyers already in production — scaled or in limited production. Buyers still piloting were not asked
In depth
Hypothesis 1 — Less Fragile Each Time, Not a Better Demo

“Improved” below combines the three improvement bands and “worsened” combines the three worsening bands.

What Operational Impact Have You Seen?

85%

of robotics buyers already in production report throughput improving over the past 12 months, the strongest of the five operational measures. Safety incidents move least, at 64%.

The 61 buyers already in production

Data for What Operational Impact Have You Seen?
CategoryImproved (%)
Throughput / output85%
Labor cost74%
Uptime / equipment availability74%
Error / defect rate (a decrease is positive)72%
Workplace safety incidents (a decrease is positive)64%
0%20%40%60%80%100%Throughput / outputThroughput / output — Improved (%): 85%85%Labor costLabor cost — Improved (%): 74%74%Uptime / equipment availabilityUptime / equipment availability — Improved (%): 74%74%Error / defect rate (a decrease ispositive)Error / defect rate (a decrease is positive) — Improved (%): 72%72%Workplace safety incidents (a decrease ispositive)Workplace safety incidents (a decrease is positive) — Improved (%): 64%64%

The 61 buyers already in production