The

stateof

AI

inwarehousingGlobal

insights

from

supply

chain

leaderson

AI’seconomic

and

workforce

impact

–

2025MIT

center

forTransportation&

LogisticsMECALUXContentsIntroductionKey

findingsTheme

1:

Current

state

of

AI/MLadoptionAI

adoption

in

warehousing

reaches

a

new

level

of

maturity?

The

expanding

footprint

of

AI

in

warehouse

operations?

AI

becomes

part

of

the

daily

workflow?

A

tool

for

operational

complexity?

An

enabler

of

high-impact

valueTheme

2:

AI/ML

investment

and

ROIAI

investments

move

from

pilot

budgets

to

proven

returns?

Budget

allocation

and

ROI

from

AI?

AI

investment

drivers?

Payback

periods

reflect

the

scope

of

transformationTheme

3:

Current

AI/ML

implementation

challengesand

enablersOvercoming

challenges

to

accelerate

AI

adoption

in

warehouses?

Main

barriers?

Internal

capabilities

for

AI

implementation?

Resources

to

speed

up

AI

adoptionTheme

4:

AI/ML

impact

on

theworkforcePeople

atthe

centre

of

warehouse

transformation?

AI

raises

skills

across

the

board?

AI

creates

new

rolesTheme

5:

Future

AI/ML

implementation

outlook

and

prioritiesThe

next

wave:

From

prediction

to

decision?

Expanding

capabilities

and

investmentTheme

6:

Dominant

methodsand

technologiesFrom

prediction

to

generation:How

AI

methods

create

value?

The

technologies

powering

today’s

smart

warehouses?

Integration:

The

foundation

of

effective

AISynthesis

across

themes?

Key

insightsand

implications

for

AI

in

warehousing?

RecommendationsCross-sectional

analysisRegional

nuances

and

differences?

The

state

of

warehouse

automation

across

countries?

AI/ML

budget

allocation

by

country?

Barriers

to

AI

and

automation

adoption

vary

by

regionIndustry-specific

differencesClosing

remarksSurvey

methodology2I

MIT

IntelligentLogistics

SystemsLab

×

Mecalux

StudyMITcenter

forTransportation&

LogisticsAboutthe

surveyYear:

2025Participants:

2,000+

experienced

supply

chain

and

warehousing

professionals.Reach:

21

countries

across

Europe,

North

America,

Asia-Pacific

and

LatinAmerica.Industry

sectors:

Agriculture,

automotive,chemicals,

construction,

consumer

goods,e-commerce

&

retail,

energy,

food

&

beverage,

government,

logistics,

manufacturing,

pharma,

technology,

textiles

and

transport.Company

size:

Respondents

representedcompanies

ranging

from

100

to

over

5,000employees,

with

the

largest

share

coming

from

organisations

with

1,000–4,999

employees.Revenue:

Most

respondents

(32%)

representcompanies

with

annual

revenues

between$251–999

million,

followed

by

26%

in

the$51–250

million

range,

19%

above

$1

billion,

18%between

$10–50

million,

and

5%

under

$10

million.IntroductionA

data-driven

viewof

AI

adoption

inwarehouse

operationsTheMIT

Intelligent

Logistics

Systems

LabandMecalux

jointly

fielded

a

global

survey

ofmore

than

2,000

experienced

supply

chain

and

warehousing

professionals

to

assess

the

current

state,

investment

patterns,

challenges,

workforce

effects

and

outlook

for

artificial

intelligence

and

machine

learning

(AI/ML)

in

warehouse

operations.

To

ensure

the

findings

reflected

organisations

with

meaningful

levels

of

warehouse

activity

and

technologyadoption,

the

survey

focused

on

companies

with

at

least

100

employees.

Our

aim

was

to

establish

a

contemporary,

evidence-based

baseline

for

AI

adoption

in

warehousing

and

logistics.

The

survey

spanned

six

thematic

blocks:

current

state

of

adoption;

investment

priorities

and

return

on

investment

(ROI)

considerations;

implementation

challenges;workforce

impacts;

future

implementation

outlook

and

priorities;

and

dominant

methods

and

technologies,

as

well

as

questions

about

respondent

demographics.The

data

reveal

a

sector

in

transformation.

More

than

four

out

of

five

organisationsincreased

AI/ML

use

in

the

past

year,

and

most

expect

budgets

to

rise

further.

The

typical

payback

period—just

two

to

three

years

—

shows

that

AI

is

delivering

tangible

returns,

not

speculative

value.

Generative

AI

is

emerging

asthe

next

frontier

in

logistics,

speeding

up

process

design,

documentation

and

decision-making.

The

human

side

of

automation

is

also

evolving

in

positive

ways:

productivity

and

job

satisfaction

are

rising

together,

driven

by

new

roles,

training

and

upskilling.Warehousing

is

evolving

from

automation

to

intelligence,

where

data

and

algorithmscomplement

human

expertise.

The

next

competitive

edge

will

belong

to

those

who

treat

AI

not

asa

project,

but

asan

embedded,

measurable

capability

that

turns

insight

into

action

faster.Matthias

WinkenbachDirector,

MIT

Intelligent

Logistics

Systems

Lab4I

MIT

IntelligentLogistics

SystemsLab

×

Mecalux

StudyTheme

4:AI/ML

impacton

the

workforce?Positive

workforce

trends

dominate:

productivity

up

for

77.5%

oforganisations,

job

satisfaction

up

for

75.4%,

training

requirements

up

for76.4%

and

workforce

size

increasedfor

55.8%.?

AI

is

creating

—notreplacing

—jobs:

AI/ML

engineers

(60.1%),

automation

specialists

(58%),process-improvement

experts

(51.9%)

and

data

scientists

(40%).Theme

5:Future

AI/MLimplementation

outlook

and

priorities?

92.1%

of

firms

are

implementing

or

planning

AI

projects

in

the

near

term.

Only

1.7%

havenoplans;6.2%

already

have

extensiveimplementations.?

87%

expect

to

increase

AI

budgets

inthe

next

2–3

years

(50.3%

slight,

36.7%

significant).?

Main

investment

goals:

efficiency

(47.9%)

and

innovation

(31.1%),followed

by

cost

reduction

(10.5%)

and

competitive

differentiation(10%).Theme

6:Dominant

methods

and

technologies?

Most

valuable

methods

today:generative

AI

(70.3%),predictiveML

(58.4%),

computer

vision

(49.8%),

reinforcement

learning

(45.9%)

and

NLP(42.5%).?

Top

generative

AI

applications:automated

documentation/reporting

(55%),

layout

optimisation

(53.6%),process-flow

design

(53.3%)

and

code

generation

(52.5%).Theme

1:Current

state

of

AI/ML

adoption?

Nearly

9

out

of

10

warehousesnow

operate

at

automation

levels

beyond

basic

processes.?

57.5%

of

organisations

operateat

advanced

or

full

automationmaturity;

only

11.7%

remain

largely

manual.?

Full

automation

is

most

common

among

larger

firms

with

higherrevenue,

more

sites

and

more

employees.Theme

2:AI/ML

investmentand

ROI?

Most

companies

dedicate11–30%

of

warehouse-tech

budgets

to

AI/ML.?

Typical

payback

period:

2–3

years.?

Main

investment

drivers:

cost

savings,

customer

needs,

labour

availability,

quality,

safety,sustainability

and

competitivepressure.Theme

3:Current

AI/MLimplementationchallengesand

enablers?

Top

adoption

barriers:

technicalexpertise

(48.6%),

system

integration

(47.7%),

data

quality

(46.2%)

andimplementation

cost

(46.1%).?Key

enablers

for

faster

adoption:better

tools/platforms

(55.5%),

more

internal

expertise

(53.7%),largerbudgets

(51.9%),

clearroadmaps(50%)

and

external

consultants

(43%).Key

findingsMIT

IntelligentLogistics

SystemsLab

×

Mecalux

StudyI5

mEcaluxAlMITcenter

forTransportation&

LogisticsTheme

1:

Current

state

of

AI/ML

adoptionAI

adoption

in

warehousingreaches

a

new

level

of

maturityThe

expanding

footprint

of

AI

in

warehouse

operationsOur

results

show

that

AI/ML

adoption

in

warehousing

is

already

well

underway

and

expandingrapidly.

Nearly

9

out

of

10

warehouses

now

operate

at

automation

levels

beyond

basic

processes,while

about

6

out

of

10

report

having

implemented

some

form

of

AI

or

ML.

A

majority

of

respondents

—

57.5%

in

total—describe

their

organisations

as

operating

at

advanced

or

full

automation

maturity

(44.6%

and

12.9%,

respectively),

with

a

further

30.8%

reporting

a

moderate

level

of

automation.

Only

11.7%

say

their

facilities

remain

at

basic

or

fully

manual

operations

(Figure

1).Figure

1Current

level

of

warehouse

automation

maturityNo

automation

(fullymanual

operations)Basic

automation

(some

automated

equipment/processes)Moderate

automation

(multiple

automated

systems)Advanced

automation

(highly

automated

with

some

AI/ML)Full

automation

(comprehensive

AI/MLintegration)

How

would

you

rate

your

organisation’s

current

level

of

warehouse

automation

maturity?6I

MIT

IntelligentLogistics

SystemsLab

×

Mecalux

Study

2.1%

9.6%

12.9%30.8%44.6%Nearly

9

outof

10

warehousesnow

operateat

automation

levelsbeyond

basicprocessesAI

becomes

partof

the

daily

workflowThe

operational

footprint

of

artificial

intelligence

is

similarly

broad:

43.1%

report

that

AI/ML

supports

between

a

quarter

and

half

(26–50%)

of

theirwarehouse

processes,

and

38.9%

say

it

coversmore

than

half

(51–75%)

(Figure

2).

This

means

that

for

most

companies,

AI

is

not

confined

toa

single

function

—it

is

embedded

across

daily

workflows

such

as

order

picking,

inventory

management,equipment

maintenance

and

workforce

scheduling.Momentum

continues

to

build:

83%

of

firmsincreased

AI/ML

use

in

the

past

year

(Figure

3).

The

pace

of

adoption

suggests

that

AI

is

becoming

a

standard

tool

for

staying

competitive

rather

than

an

optional

innovation

experiment.Significantly

decreasedSomewhat

decreasedNo

changeSomewhat

increasedSignificantly

increased

Compared

to

oneyear

ago,

how

hasyour

organisation’s

use

of

AI/ML

in

warehouse

operations

changed?0%

of

warehouseoperations1–25%

of

warehouseoperations26–50%

of

warehouseoperations51–75%

of

warehouseoperations76–100%

of

warehouseoperations

What

percentage

of

your

warehouseoperations

currently

utilise

AI/ML

technologies?Figure

2Warehouse

operations

utilising

AI/ML

technologiesFigure

3Change

in

AI/ML

adoption

over

the

last

year

(12

months)MIT

IntelligentLogistics

SystemsLab

×

Mecalux

Study

I7

0.6%

9.8%

mEcaluxMITcenter

forTransportation&

Logistics28.4%38.9%0.2%43.1%54.6%13.4%3.4%7.6% 100–249employees

250–499employees

500–999employees

1,000–4,999employees●5,000

or

moreemployees.

1warehouse

2–5warehouses●

6–10warehouses

11–20warehouses

More

than20

warehousesLess

than

$10

million(Less

than

€9.5

million)$10–50million(€9.5–47.5

million)$51–250

million(€48–238million)$251–999

million(€238.5–950.5million)$1

billion

or

more(€951

million

or

more)Figure

6Share

of

AI/ML-enabled

operations

by

number

of

warehousesFigure

5Share

of

AI/ML-enabled

operations

by

company

sizeFigure

4Share

of

AI/ML-enabled

operations

by

company

revenue0%

1–25%26–50%51–75%76–100%0%

1–25%26–50%51–75%76–100%0%

1–25%26–50%51–75%76–100%50%40%30%20%10%0%50%40%30%20%10%0%50%40%30%20%10%0%8I

MIT

IntelligentLogistics

SystemsLab

×

Mecalux

StudyInventory

optimisation

algorithms

61.7%

Automatedpicking

systems

56.4%Route

optimisation

55.5%Demand

forecasting

51.7%Predictivemaintenance

48.4%Computer

vision

systems

46.5%Naturallanguageprocessing

25.2%None

currentlyimplemented

1.7%Other

0.1%

What

types

of

AI/ML

technologies

are

currently

implemented

in

your

warehouse

operations?*An

enabler

of

high-impact

valueCurrent

AI/ML

deployments

concentrate

on

proven,

high-

impact

capabilities

with

clear

operational

value:

inventory

optimisation

algorithms

(61.7%),

automated

picking

systems

(56.4%),

route

optimisation

(55.5%),

demand

forecasting(51.7%),

predictive

maintenance

(48.4%)

and

computer

vision

(46.5%).

The

adoption

of

natural

language

processing

(NLP)

technologies

(25.2%)

is

still

lagging

(Figure

7).Technology

effectiveness

improves

with

an

organisation’s

automation

maturity.

Among

fully

automated

sites,the

share

of

respondents

rating

a

technology

as

“veryeffective”

rises

to

56.5%

for

inventory

optimisation,

57.1%

for

demand

forecasting,

59.7%

for

route

optimisation,

50.6%

for

predictive

maintenance,

58.5%

for

automated

picking

and

68.4%

for

NLP.

Mature

users

are

not

just

using

more

AI—

they

are

using

it

better.A

tool

for

operational

complexityA

company’s

level

of

AI/ML

maturity

—

that

is,

the

extent

to

which

its

operations

haveadopted

automated

systems

and

technologies

—

rises

sharply

with

its

scale

andoperational

complexity.

Our

results

indicate

that

full

automation

is

more

prevalent

among

firms

with

higher

annual

revenues,

larger

warehouse

networksand

larger

individualfacilities

(Figures

4,

5

and

6).

Smaller

organisations,

while

less

automated

overall,

represent

the

next

frontier

for

adoption

asAI

becomes

more

accessible

and

scalable.Figure

7AI/ML

technologies

implemented*Multiple

responses

allowed;

results

do

not

add

up

to

100%MIT

IntelligentLogistics

SystemsLab

×

Mecalux

Study

I9

mEcaluxMITcenter

forTransportation&

LogisticsTheme

2:

AI/ML

investment

andROIAI

investments

move

from

pilotbudgets

to

proven

returnsBudget

allocation

andROI

from

AIThe

overall

approach

to

AI/ML

investment

in

warehousing

ispragmatic.

Most

organisations

dedicate

between

11%

and

30%

of

theirwarehouse-technology

budgets

to

AI

and

machine-learning

initiatives(42.4%

at

11–20%

and

32.8%

at

21–30%)

(Figure

8).Figure

8Budget

allocated

to

AI/ML

technologiesLess

than5%5–10%11–20%21–30%More

than

30%

Whatis

your

organisation’s

annual

budget

allocation

for

AI/ML

warehouse

technologies?

(As

%

of

overall

warehouse

tech

budget)10I

MIT

IntelligentLogistics

SystemsLab

×

Mecalux

Study

2.5%

12.9%本報(bào)告來源于三個(gè)皮匠報(bào)告站（）,由用戶Id:349461下載,文檔Id:980188,下載日期:2025-12-04

9.4%42.4%32.8%Typical

payback

periods

cluster

around

two

to

three

years

(34.7%),followed

by

three

to

five

years

(27.3%)

andoneto

two

years

(22%)

(Figure

9).

This

means

that

many

companies

are

already

seeing

measurable

returns

within

a

single

business

cycle

a

sign

that

AI

is

delivering

real

operational

and

financial

value,

not

just

long-term

potential.Figure

9Payback

period

for

AI/ML

investmentsReturn

on

investment

is

measured

primarily

throughtangible,

performance-driven

outcomes

such

asinventory

optimisation

(63.1%),

improved

throughput

(57.6%),

labour-cost

reduction

(52.1%),

error

reduction

(48.9%),energy

savings

(47.6%)

and

customer

satisfaction

(38.3%),

which

closely

mirror

the

technologies

deployed

(Figure

10).Figure

10ROI

metrics

employedLess

than1

year12

years2

3

years3

5

yearsMore

than

5

yearsToo

early

to

tell

Not

applicable

/

We

don

t

measure

AI/ML-related

paybackInventory

optimisationImproved

throughput

Labour-cost

reduction

ErrorreductionEnergy

savingsCustomer

satisfaction

Not

applicable

/

We

don

t

measure

AI/ML-relatedROI

4.8% 22.0% 34.7%

27.3%

7.7%

2.5% 1.0%

63.1%

57.6%

52.1% 48.9%

47.6%

38.3% 0.8%

What

hasbeen

yourtypical

payback

period

for

AI/ML

implementations?

How

doyou

measureROI

on

AI/MLimplementations?**Multiple

responsesallowed;

results

do

notadd

up

to100%MIT

IntelligentLogistics

SystemsLab

×

Mecalux

StudyI11

mEcaluxMITcenter

forTransportation&

LogisticsAI

investment

driversCommon

justifications

for

AI/ML

investments

appear

to

be

multifactorial.Most

respondents

rate

cost

savings,

customer

requirements,

labour

availability,

quality

improvement,

safety

enhancement,

sustainability

goals

and

competitive

pressure

as

very

important

or

critical

driversin

their

decision-making.

This

mix

of

motivations

underscores

how

AI

adoption

is

shaped

as

much

by

productivity

or

cost

efficiency

as

by

workforce

and

sustainability

goals.Moreover,

our

results

show

statistically

significant

relationships

between

allocated

budget

levels

and

the

perceived

importance

of

several

ofthese

factors

(Figures

11

and

12).Figure

11Budget

allocation

by

perceived

importance

ofcompetitive

pressure

as

a

driver

of

AI/ML

investmentsHow

important

are

sustainability

goals

in

your

decision-making

process

forinvesting

in

AI/ML?●

Not

important●

Somewhatimportant●

Very

important●

CriticalHow

importantis

competitivepressurein

yourdecision-makingprocess

forinvestingin

AI/ML?●

Not

important●

Somewhatimportant●

Very

important●

CriticalFigure

12Budget

allocation

by

perceived

importance

of

sustainabilitygoals

as

a

driver

of

AI/ML

investmentsLess

than5%5–10%11–20%21–30%More

than30%Less

than5%5–10%11–20%21–30%More

than30%50%40%30%20%10%0%50%40%30%20%10%0%12I

MIT

IntelligentLogistics

SystemsLab

×

Mecalux

Study

mEcaluxMITcenter

forTransportation&

LogisticsIntelligentLogistics

Systems40%35%30%25%20%15%10%5%0%Too

early

Not

applicable

/to

tell

We

don’t

measureAI/ML-related

paybackBasic

automation(some

automatedequipment/processes)40%35%30%25%20%15%10%5%0%Too

early

Not

applicable

/to

tell

We

don’t

measureAI/ML-related

payback●Less

than

$10

million(Less

than

€9.5

million)Payback

periods

reflect

the

scope

of

transformationExpected

investment

payback

periods

vary

by

a

company’sAI/ML

maturity

level

andother

firm

characteristics.

Larger

enterprises,

higher

budget

commitmentsand

multi-site

warehouse

networks

tend

to

experience

longer

payback

horizons,

consistent

with

thegreater

complexity

and

integration

scope

of

their

transformations

(Figures

13,

14,

15

and

16).

For

these

organisations,

AI

is

nota

one-off

investment.

It

is

a

long-term

capabilityprogramme

that

requires

more

coordination

but

yields

broader

value

once

fully

realised.Smaller

firms,

by

contrast,

often

benefit

from

faster

returns

because

they

can

deploy

AI

solutions

with

less

disruption.Figure

13Expected

AI/ML

investment

payback

periods

by

AI/ML

maturity

levelFigure

14Expected

AI/ML

investment

payback

periods

by

company

revenue1–2

years2–3

years3–5

yearsMore

than5

years1–2

years2–3

years3–5

yearsMore

than5

yearsModerate

automation

(multiple

automated

systems)Advanced

automation

(highly

automated

with

some

AI/ML)Full

automation(comprehensive

AI/

ML

integration)14I

MIT

IntelligentLogistics

SystemsLab

×

Mecalux

Study●

$1

billion

or

more(€951

million

or

more)No

automation

(fully

manual

operations)●$10–50million(€9.5–47.5

million)●

$51–250

million(€48–238million)(€238.5–950.5million)●

$251–999

millionLess

than

1

yearLess

than

1

year40%35%30%25%20%15%10%5%0%Too

early

Not

applicable

/to

tell

We

don’t

measureAI/ML-related

paybackLess

than

5%5–10%11–20%21–30%More

than

30%40%35%30%25%20%15%10%5%0%Too

earlyNot

applicable

/to

tell

We

don’t

measureAI/ML-related

payback●More

than

20

warehousesSmaller

firms

see

returns

sooner

because

AI

deployment

is

simpler

and

disruption

is

lowerFigure

16Expected

AI/ML

investment

payback

periods

by

number

of

warehouses

in

operationFigure

15Expected

AI/ML

investment

payback

periods

by

AI/ML

budget

allocation1–2

years

2–3

years3–5

yearsMore

than5

years1–2

years

2–3

years3–5

yearsMore

than5

yearsMIT

IntelligentLogistics

SystemsLab

×

Mecalux

StudyI15

mEcaluxMITcenter

forTransportation&

Logistics●11–20

warehouses●

6–10

warehouses●

2–5

warehousesLess

than

1

yearLess

than

1

year●

1

warehouseTheme

3:

Current

AI/ML

implementationchallenges

and

enablersOvercoming

challenges

to

accelerate

AI

adoption

in

warehousesMain

barriersThe

main

challenges

to

AI/ML

adoption

in

warehouses

are

skills,

system

integration,data

quality

and

cost

—

the

core

ingredients

for

any

successful

digital

transformation.

Nearly

half

of

respondents

selected

technical

expertise

requirements

(48.6%),integration

with

existing

systems

(47.7%),

data

quality

and

availability

(46.2%)

and

cost

of

implementation

(46.1%)

among

their

top

four

challenges.

These

are

followed

byemployeeresistance

and

training

(39.7%),ROI

uncertainty

(30.5%)

and

vendor

selection

ormanagement

issues

(25.3%)

(Figure17).In

simple

terms,most

organisations

face

the

same

problem

set:

connecting

new

tools

to

old

systems,

ensuring

reliable

data

anddeveloping

the

right

in-house

skills

to

makeAI

work.Figure

17Main

barriers

to

AI/ML

adoptionTechnical

expertiserequirementsIntegration

with

existing

systemsData

quality/availabilityCost

of

implementationEmployeeresistance/trainingROI

uncertaintyVendor

selection/managementChange

management

What

are

the

main

challenges

you’ve

faced

in

implementing

AI/ML

solutions?* 30.5% 25.3%

15.7%

48.6%

47.7%16I

MIT

IntelligentLogistics

SystemsLab

×

Mecalux

Study*Multiple

responses

allowed;

results

do

not

add

up

to

100%46.2%46.1%39.7%50%40%30%20%10%0%VerylimitedSomewhat

limitedModerateSomewhat

strongVery

strongHow

would

you

rate

your

organisation’s

internal

capabilities

for

AI/ML

implementation?●Change

managementInternal

capabilities

for

AI

implementationAtthe

sametime,

many

companies

showencouraging

signs

of

readiness.

Across

five

internal

capability

areas

—

technical,

data,

project,

vendor

and

change

management

—

organisations

report

“somewhat

strong”

or

“very

strong”,

with

“veryst