建立人工智能戰(zhàn)略 電信公司奠定了基礎(chǔ) Building an AI strategy - telcos put the foundations in place

tmforum

knowledge

AI

telcos

inplace

TMForum|March2024

BENCHMARK

buildinganstrategy:

putthe

foundations

Author:

MarkNewman,ChiefAnalyst

Editor:

IanKemp,ManagingEditor

Sponsoredby:

amd。cs人R

●

Tecnotree

servIcen◎

contents

03

aboutoursponsors

13

section2:

puttingtheright

skillsinplace,

byDawnBushaus

27

section5:

whereoperators

areapplyingAI

46

section8:

thefutureofAI,

byDeanRamsay

05

thebigpicture

17

section3:

gettingthebasicsright:

isCSPdataAI-ready?

Co-author:RichardWebb

34

section6:

theimpactof

generativeAI(GenAI)

49

section9:

keyfindingsand

recommendations

08

section1:

buildinganAI-ready

organization

23

section4:

architecting,building,

buyingandscalingAI

40

section7:

measuringthe

economicimpactofAI

52

additionalfeatures

&resources

2

3

BENCHMARK

aboutour

sponsors

amdocs

Amdocshelpsthosewhobuild

thefuturetomakeitamazing.

Withourmarket-leadingportfolioofsoftwareproductsandservices,weunlockourcustomers’

innovativepotential,empoweringthemtoprovidenext-generationcommunicationandmedia

experiencesforboththe

individualenduserandenterprisecustomers.Ourapproximately

29,000employeesaroundthe

globeareheretoaccelerate

serviceproviders’migrationtothecloud,enablethemtodifferentiateinthe5Gera,anddigitalizeand

automatetheiroperations.ListedontheNASDAQGlobalSelect

Market,Amdocshadrevenue

of$4.89billioninfiscal2023.

Formoreinformation,

visit

Ariaenablesenterprisesto

automatecomplexusageand

subscriptionbillinginanagile

marketenvironment.AriaBillingCloudistop-ratedbyleading

researchfirms,andinnovative

enterprisesincludingAdobe,

Comcast,Experian,Subaru,and

TelstradependonAriato

accelerateideation,become

customercentric,andgrow

recurringrevenues.

Ariaprovidestheonlycloud-

native,SaaSbillingand

monetizationplatformthatisbothrobustandscalableenoughto

helpCSPsseizenewmarket

opportunities.TheagilityofAria’splatformallowsCSPstomeetthedemandsoftheircustomersby

enablingthemtoquicklydeployawiderangeofmonetization

modelsacrossalltypesof

productsandbundles.

Formoreinformation,

visit:

.

aws

AmazonWebServices(AWS)istheworld’smostcomprehensiveandbroadlyadoptedcloud,

offeringover200fullyfeatured

servicesfromdatacenters

globally.Millionsofcustomersare

usingAWStolowercosts,

becomemoreagile,andinnovatefaster.AWSforTelecomis

harnessingtherobust,mature,

proven,andsmartplatform,to

deliverinnovativeusecasesthatarehelpingtelcostransform,

digitizeindustries,andreimaginetheconsumerexperience.With

thepoweroftheAWSGlobal

CloudInfrastructure,that’s

architectedtobeflexibleand

secure,AWSforTelecomhelp

CSPstaketheirbusinesstothe

nextlevel.Throughsimplifying

operations,cloudifyingnetworks,

monetizingassets,andinnovating

atscaleusingAIandgenerative

AIservices.

Cognizant:PoweringtheFuture

ofTelco

Cognizantisaleading

professionalservicescompany

dedicatedtohelpingCSPs

navigatethedigitalera.Weoffera

comprehensivesuiteofsolutions,

fromcorenetworkmodernisation

andcloudmigrationtocustomer

experiencetransformationand

AI-poweredautomation.Ourdeep

industryexpertiseandglobal

reachempowerCSPstounlock

newrevenuestreams,optimise

operations,anddeliver

exceptionalcustomerexperiences.

LetCognizanthelpyoubuilda

future-proofnetwork,empower

yourworkforce,andthriveinthe

ever-evolvinglandscape.Together,

we’reimprovingeverydaylife.See

howat

https://www.cognizant.

com/us/en/industries/cmt

-

solutions/communications

.

4

aboutoursponsors

CSGempowerscompaniesto

buildunforgettableexperiences,makingiteasierforpeopleandbusinessestoconnectwith,useandpayfortheservicesthey

valuemost.Ourcustomer

experience,billingandpaymentssolutionshelpcompaniesofanysizemakemoneyandmakea

difference.WithourSaaS

solutions,companyleaderscantakecontroloftheirfuture,andtapintoguidancealongthewayfromourmorethan5k-strong

experiencedglobalteam.Wanttolearnmoreabouthowtobea

changemakerandindustry

shaperlikeour1,000-plusclients?Visit

tolearnmore.

AtNokia,wecreatetechnologythathelpstheworldacttogether.

AsaB2Btechnologyinnovationleader,wearepioneering

networksthatsense,thinkandactbyleveragingourworkacross

mobile,fixedandcloudnetworks.

Inaddition,wecreatevaluewithintellectualpropertyandlong-termresearch,ledbytheaward-winningNokiaBellLabs.

Serviceproviders,enterprises

andpartnersworldwidetrust

Nokiatodeliversecure,reliableandsustainablenetworkstoday–andworkwithustocreatethedigitalservicesandapplicationsofthefuture.

●

servIcen◎

ServiceNowhelpsTelcosaccelerate

growth,maximizenetwork

investment,andreducecosts

acrosstheentiredigital

ecosystem—allwhiledelivering

seamlessexperiences.Beit

planningfor5Gorfiberbuildouts,managingdatacenters,launchinganddeliveringnewservices,or

automatingnetworkoperations,

ourpurpose-builtsolutionshelp

makeworkeasieracrosstheentireservicelifecyclewithautomationandAI.Builtforthefast-changing

worldofTelecom,theNowPlatformconnectspeopleanddatafor

greaterproductivityandinnovation.

/

Tecnotree

Tecnotreeisa5G-readydigital

BusinessSupportSystem(BSS)

player,withAI/MLcapabilitiesand

multi-cloudextensibility.Tecnotree

leadsthewayontheTMForum

OpenAPIConformancewith59

certifiedOpenAPIsincluding9

real-worldopenAPIs,atestament

tothecompany’scommitmentto

excellence,andcontinuously

strivingtodeliverdifferentiated

experiencesandservicestoboth

CSPsandDSPs.Ouragileand

open-sourcedigitalBSSStack

comprisesthefullrange(order-to-

cash)ofbusinessprocessesand

subscriptionmanagementfor

telecomandotherdigitalservices

industriescreatingopportunities

beyondconnectivity.Tecnotree

alsoprovidesFintechandB2B2X

multi-experiencedigital

marketplacetoitssubscriberbase

throughtheTecnotreeMoments

platformtoempowerdigitally

connectedcommunitiesacross

gaming,health,education,OTT,

andotherverticalecosystems.

TecnotreeislistedontheHelsinki

Nasdaq(TEM1V).

Tolearnmoreaboutoursponsors’viewsonAItechnologiesandstrategieswatchthesevideos:

Aria

Nokia

AWS

Servicenow

Cognizant

Tecnotree

CSG

BENCHMARK

thebigpicture

TheAIjourneystartedin2016or2017forsomeofthemoreprogressive,ambitiousoperators,andin2019or

2020formostothers.Butitisonlyinthepast12-15monthswiththeexplosionofinterestinGenerativeAI

(GenAI)thatithasbecomeakeyfocusformostboardroomsandthetechnologywhichcommunicationsserviceproviders(CSPs)arepinningtheirhopesontotransformtheirfinancialfortunes.

T

hepurposeofthisreportistounderstandwhereCSPsareontheirAIjourneys.Itis

basedonbriefingswithAI

decision-makersin20ofthe

world’slargestandleadingCSPs

globally.Webelievethey

representsomeofthemore

progressiveCSPsintermsoftheirinvestmentin,andcommitment

to,AItechnologies.Assuch,theypresentaslightlydistortedpictureoftheglobalstatusofCSPuseofAI.However,thechallengesand

opportunitiestheyspeakaboutinrelationtoAIarecommontothesectorasawhole.

Wealsopublishtheresultsofa

surveyinthisresearchintheformofchartsthroughoutthereport.

Thesurvey–whichwealsoused

foranearlierreport,

GenerativeAI:

operatorstaketheirfirststeps

–

coversabroadercross-sectionofCSPs,withresponsesfrom104

executivesacross73CSPsglobally.

5

Twoofthe20operatorsthatwe

interviewed–ChinaMobileand

SKTelecom–arebuildingAI

businesseswithaclearobjective

tomonetizetheirinvestments.Theother18areinvestinginAIfortheirowninternalusage,principallytodriveefficienciesandimprove

customerexperience.Addressingthegrowingcomplexityoftheirbusinesses–andspecificallytheirnetworkoperationsfunctions–isalsoakeydriver.

AlltheoperatorsthatwespoketoarerampinguptheirinvestmentsinAI–beitbuildingteamsofAIexpertsorbuildingplatformsandtoolsinreadinesstostartscalingAIdeployments.

“I’dsayeverymeetingIgoto,

nearlywithoutexception,there’s

somecomponentthatdiscusses

whatwe’redoinginAI–nextyear’sinvestmentplanning,execution

skills,peopledevelopment,

I’dsayeverymeetingIgoto,nearly

withoutexception,there’ssome

componentthat

discusseswhatwe’redoinginAI–next

year’sinvestmentplanning,executionskills,people

development,

everything.”

MarkSanders,

ChiefArchitect,Telstra

6

thebigpicture

HowefectivelyisyourorganizationexploitingAIandmachinelearning?

WehavemadegoodprogressinusingAIandmachinelearningindiferentpartsofourbusiness(e.g.customerexperienceornetworkoperations)

Wehavesometrialsandproofsofconceptrunningbutwehaven'tdeployeditinourbusiness

WeareattheearlystagesofresearchingthepotentialofAIandmachinelearning

31%

14%

55%

TMForum,2023

everything,”saysMarkSanders,ChiefArchitect,Telstra.

However,itisimportanttodrawadistinctionbetween“traditional”typesofAI–whatwerefertointhisresearchaspredictiveAI–

andGenAI.

CSPinvestmentinpredictiveAI

has,untilnow,beenrelatively

modest.Mostoperatorshave

deployedahandfulofusecasesinareassuchasenergysaving,churnprediction,andnextbestaction

whichusesAIandreal-timedatatotargetcustomerexperiences.

Buttheyhavelackedtheskills,thetoolsandcompellingevidenceofgoodreturnsoninvestmentto

scalerollout.Furthermore,usecaseshavebeenmore“bolt-on”,ratheronesthatsitwithinan

overarchingarchitectureor

operatingmodel.

ThefocusonGenAI

Overthepastyearto15months

CSPs’AIstrategieshavepivotedtofocusonGenAI.Indeed,in

manyofourconversationsforthisreportGenAIwasthemaintopicofdiscussion.MostCSPshave

drawnupwellinexcessof100usecasesandareengagedina

processofaggressive

experimentation.Butveryfewhavebeenputintoproduction.

Weexpectthistochangeduring2024and,morelikely,in2025.

Somewhatironically,this

enthusiasmforGenAImayendupboostinginterestinpredictiveAI.WhilesomeusecasesinvolveonlyGenAI,manyothersrequirethe

twotobeusedintandem.

Furthermore,whenitcomesto

usingAIforautomationGenAIhasitslimitations,particularlyinthe

networkandthedrivetowards

autonomousnetworks.

InspeakingtoCSPsaboutthe

statusoftheirAIjourneysmost

conversationstendtoendup

focusingmoreonGenAIthan

predictiveAI.IsthisbecauseCSPsbelievethereisfarmorepotentialthaninmachinelearningand

predictiveAI?OrisitjustbecauseGenAIisthenewertechnology

andhascapturedtheimaginationofboardroomsandtheiradvisers?

AkeyreasonforGenAI’sappealisitsaccessibilitycomparedwith

machinelearningandpredictiveAI,andthefactthatitusesthe

sameinputsandprovidesthe

sameoutputsasthetoolsweuseinoureverydayworkinglives.

GenAIcantakeourconversations,theemailswewrite,the

documentsweuse,theonline

researchthatweconductandthepresentationswemakeandaddvaluetothem.

What’smore,atatimewhenevery

personintheCSPorganizationis

encouragedtobemorecustomer

centric,GenAIoffersthepromise

ofdeliveringinsightsandvalue

fromthemillionsofconversations

andmessagesthatpassbetween

customers,customerservice

agentsandonlineandmobile

platforms.Eventheleast

technology-savvypeopleinan

organizationcangettheirheads

aroundwhatGenAIcando.

Arealisticassessment

AIisneither,ontheonehand,the

answertothechallengesthatCSPs

facenor,ontheother,awildly

overhypedtechnologythatwillfail

tohaveameaningfulimpacton

operatororganizations.AIwillbe

adoptedbyoperatorsinthesame

waythatitwillbeadoptedby

almosteveryothersector.Butwill

itenableCSPstosignificantlycut

costs,tocreatenewproducts,

servicesandrevenuestreams,and

toredefinetherelationshipwith

theircustomers?

Thecostdiscussionis,perhaps,

themostinterestingone,and

lessonsshouldbelearnedfrom

theexperienceofpubliccloud

whichwasoriginallytoutedasa

cost-savingopportunityforCSPs.

Operatorsareunderpressureto

cuttheircostsbecauseofrevenue

7

thebigpicture

stagnation.InmanycasesAI-

enabledautomationcanhelp

employeesdotheircurrentjobsmoreefficientlyandmorequickly.

Butwilloperatorsbeableto

capturethesetimesavingsand

convertthemintotheeliminationofdifferentroles?Orwillitsimplyallowemployeestofocustheir

timeonothertasksandfunctionsthatcreatemorevalue?Ultimately,thiswillcomedownto

departmentalbudgeting,the

pressuretoreducecapexand

opex,andwhetherAIand

automationaretwoofthetoolsthatCSPsleanontocutcosts.

WeshouldalsonotforgetthatAIwillrequireoperatorstoinvestinnewroles,newtechnologiesandplatformsand,insomecases,newinfrastructure.Inaddition,aswe

seeinsection2,newskillstendtobemorecostlythanoldones.

Belowaresomeofthekey

findingsandrecommendationsfromourresearch.Forafullersummaryseesection9.

Keyfindings

Thereisnearlyuniversal

recognitionthatAIwilllead

toaparadigmshiftinCSPs’

operatingmodels.However,

veryfewoperatorshaveaclearpictureofwhatthatoperatingmodelwillorshouldbe.

AI-readyaccesstoclean,

quality,usabledataisthe

singlebiggestchallengeto

thedeploymentofAIatscale.

MostAIdeploymentstodayare“bolt-on”usecases

–specific,isolateddeploymentsratherthanonesthatrepresentpartofanoverallarchitectureandoperatingmodel.

TherearefewexamplestodayofAIusecases/deploymentsthathavedeliveredsignificantorclearlydemonstrablevaluetotheCSPbusiness.Those

areaswhereCSPshavehadearlysuccessincludeenergysaving,networkplanning/

smartcapex

andchurn

predictionmodels.

DespitethebuzzaroundGenAIveryfewusecaseshavebeenputintoproduction.ManyCSPsarehopingthat2024willbetheyearGenAI-enableduse

casesbecomeareality.

FuturepredictiveAIuses

casesareincreasinglyseen

asprincipallydelivering

valuetonetworkoperations

asCSPsseektobuild

autonomousnetworks.

GenAIismostlikelytosee

wide-scaledeploymentin

customer-facingfunctions

andasatoolfordriving

greaterproductivitygenerally.

0OperatorspendingonAI(asopposedtoanalytics)has

beenmodestuntilnow,but

manyCSPsarestartingtospendsignificantlymoreonhyperscalerservicesthat

embedGenAI.

AstheyembraceGenAICSPs

arebecomingincreasingly

reliantonhyperscaleservice

providersandthepubliccloud.

Veryfewoperatorsare

buildingAIinfrastructure.

Rather,CSPinvestmentsarefocusedondevelopingusecasesandbuildingplatforms

forteamsacrossthebusinesstousedifferentAItools.

OuranalysisofCSPcapexandopexpresentstwoscenariosforthepotentialsavingsthat

couldresultfromthesuccessful

useanddeploymentofAI.

Ina“bullish”scenariototal

capexandopexfortelecoms

operatorsgloballyfallsby

9.1%from$1.8trillionto

$1.64trillion.A“bearish”

scenarioseesa2%decline

to$1.77trillion.

Recommendations

Operatorsneedtodevelop

easyaccesstodatathatsits

acrosstheentireorganization

iftheyaregoingtodeployAI

atscale.

Ifoperatorsaretoshapehow

AIwillchangetheirbusinesses

theymustfundamentally

embracethesenew

technologiesandmakethem

integraltotheiroperations.If

theydonot,theyriskbeing

lockedintotechnologychoices

andarchitectures.

CSPsmustnotrepeatthe

mistakestheymadeinpublic

cloudwheremanytriedto

competewithhyperscaler

serviceproviders.Theirfocus

inAIshouldbeonthe

deploymentofusecasesand

thecreationofplatformsfor

deliveringAIcapabilitiesto

theirinternalorganization.

ForoperatorstodeployAIat

scaleandtoindustrialize

deploymentstheyneedto

embracemachinelearning

operations(MLOps)and

FinOpsinthesamewaythat

thesoftwaresectorwas

transformedwiththeuseof

DevOps.

Inthefirstsectionwelookatthe

differentapproachesCSPsare

takingtobuildorganizationsable

toharnessAItechnologies.

BENCHMARK

section1:

buildinganAI-ready

organization

Findingtherightbalancebetweencreatingcentralfunctionsandcapabilitiesontheonehand,and

democratizingaccesstoAItoolsontheother,isamajorchallengeforCSPs.Asaresultavarietyofroles,organizationalalignmentsandgovernancefunctionsarebeingdevelopedtogetthisbalanceright.

O

peratorshadbeen

buildingandevolving

theirdataanalytics

organizationsforseveralyears

beforeAIemergedasastrategic

priority.AIfunctionsand

organizationstendtositontop

ofthesedataanalyticsteams–

althoughinsomecasesthey

arepartofthesameteam–and

theirsizeandremittendsto

beafunctionofhowmuch

AItechnologyanoperatoris

buildingandwhat,andhow

much,itisbuying.

Anyfutureorganizationalfunctionandstructureneedstofactorin

thefollowingconsiderations:

AIisatoolforinnovation,andanyoneintheorganizationcanpotentiallybeaninnovator.

Assuch,AIneedstobemadeavailabletoeveryteamin

thecompany.

8

Thelevelofcomfortwith

embracingnewtechnologies

variesenormously.Every

effortneedstobemadeto

makeAIeasytouse,even

forteamswithlowtechnologycapabilities.

Reusingsuccessfulusecasescanbeamajorchallenge.ThecentralAIteam/platform

needstobeabletoseewheresuccessfulusecasesarebeingrolledoutandfacilitatetheir

reuseacrosstheorganization.

Costmanagementrisks

becomeaseriousissueifthedemocratizationofaccesstoAIisnotaccompaniedbythedeploymentofnew

FinOps

practices

forAI.

Thereisariskofincurring

technicaldebtthroughthe

uncontrolledpurchaseand

deploymentofdifferentAI

modelsandservicesacross

theorganization.AnAI

governanceprogramneeds

tobeputintoplacetoreduce

theriskofincurringdebt

while,atthesametime,

encouragingusageand

experimentation.

Inthissectionwelookatsome

examples,fromourresearch,of

thedifferentorganizational

approachesthatCSPsaretaking

toharnessAI.

CreatingaCXOroleforAI.

AstheprofileofAI,andthe

expectationsforwhatitcan

deliver,haverisenithasbecome

increasinglyimportantthatthere

isexpertiseandrepresentationat

–orjustbelow–boardlevel.Two

oftheCSPsthatwespoketofor

thisreporthaveappointedChief

AIOfficers(CAIOs).SteveJarrett,

section1:buildinganAI-readyorganization

previouslyGroupSeniorVice

President,DataandAI,atOrangeInnovation,wasappointedChief

AIOfficeroftheFrenchoperatorinDecember2023.InhisLinkedInprofileJarrettsayshisteam“l(fā)eadsour‘DataDemocracy’cultural

andtechnicaltransformationofthecompany”.InhisroleJarrettalsoleadstheselectionofkeypartnersandtheidentificationofpotentialacquisitions.

BThasaChiefDataandAIOfficer

–DeepikaAdusumilli–whojoinedthecompanyinOctober2023.

AdusumillireportstoHarmeenMehta,ChiefDigitaland

InformationOfficer.

WhileaCAIOisresponsibleforAI,aChiefDigitalOfficer(CDO)looksafterall“digital”technologies.In

CSPsthistendsnottoincludethe

networkitselfwhichisthe

responsibilityofaChiefTechnologyOfficeroraChiefNetworkOfficer.

SKTelecomisanotherCSPwithahigh-profileAIrole.ChungSuk-

guenisChiefGlobalAIOfficerat

SKTelecom,buthisroleis

principallyfocusedonGenAIandbuildingproductsthatcanbe

leveragedinternallyandmonetizedthroughpartnershipswithother

CSPsand,potentially,softwareandservicesvendors.

RatherthanhavingasingleCXO

responsibleforAI,someCSPs

divideresponsibilitiesbetween

executives.AfricantelecomsgroupMTNhasanAIsteeringcommitteeco-chairedbytheGroupCIOandChiefStrategyandTransformationOfficer,ChikaEkeji.

ChiefDataOfficerisalsoarolewhichexistswithinmanyCSP

organizations.Perhapsthe

highest-profileChiefDataOfficerintelecomswasChemaAlonsoatTelefónica,althoughhehasnow

transitionedtoChiefDigital

OfficerwithresponsibilityforAIincustomersolutions.TelefónicastillhasaChiefDataOfficer,butthisisnowaninternalrole,with

9

ChinaMobile

andSKTelecomareinvesting

aggressivelyinAItodelivernewproductsand

services.

responsibility,forexample,forthecompany’sdataarchitecture.

Settingupasubsidiary

organizationtodevelopAI.ManyCSPshavesetupsubsidiariesinrecentyearstobuildcapabilities,productsandservicesinnewtechanddigitalbusinessareas.But

thesehavefocusedonbuilding

newproducts,servicesand

revenuestreamsratherthan

deliveringcapabilitiesbackinto

theorganization.

ProximusAdaisanaffiliateof

Belgiantelecomsoperator

Proximusthatisspecifically

focusedondeveloping

cybersecurityandAIcapabilities.ItsworkissplitbetweenmeetinginternalProximusrequirements

anddeliveringservicestoB2B

customers.ProximusChiefDigitalInformationOfficer(CDIO),

AntoniettaMastroianni,saysAdawascreatedtohelpfindthe

requisiteskillsincybersecurity

andAIandtobuildaflexible

organizationalculturethatwasdifferentfromthecorebusiness.

Splittingtheinternaland

customer-facingAIfunctions.

RatherthanbuildingacentralAI

organization,Telefónicahas

decidedtosplitresponsibilitiesinto

two:customerinsightsand

innovation;andnetwork,ITandinternalAI-driventransformation.

ChiefDigitalOfficer,ChemaAlonso,

whocreatedthevoice-controlled

AIplatformAurain2017,

is

responsibleforcustomerinsights,

whileJuanManuelCaroBernatis

DirectorofAutonomousNetworks,

responsibleformanagingthe

wholelifecycleofnetworksandAI.

Thissplittingofresponsibilitiesis

particularlyinterestinggiventhat

thefocusofGenAIismoretowards

customer-facingratherthan

networkfunctions,whilepredictive

AIisincreasinglyatechnology

usedfornetworkautomation.

AIasanewbusinessfunction.

ChinaMobileandSKTelecomare

investingaggressivelyinAIto

delivernewproductsandservices.

Thefocusforbothoperatorsis

buildingtheirownLargeLanguage

Models(LLMs)–andsolutionsand

capabilitiesthatsitontopofthem

–andsellingaccesstothemto

enterprisesandotherCSPs.

ChinaMobile’s

Jiutianbusiness

hasbeensetupbothtoprovide

servicesinternallytothe

organizationandtodeliverB2B

services.InOctobertheoperator

announcedaJiutianGenAImodel

whichitsayshas“wonsupport”

fromlargeenterprisesincluding

ChinaOceanShippingandChina

RailwayConstructionCo.The

modelhastrainedmorethan2

milliontokens–abasicunitof

textorcodethatanLLMusesto

processandgeneratelanguage–

andhas“knowledge”ineight

industriesincludingtelecoms,

energy,andsteelandtransport,

saysChinaMobile.

“Webelievethatexternalsuccess

istherealmeasureofsuccess

[ofJiutian],”aChinaMobile

spokespersontoldTMForum.

EstablishinganAIcenterof

excellence.Inoursurvey53%of

CSPssaidtheyhadestablished

acenterofexcellenceforAI

(seechartonthenextpage).But

theprecisesize,shapeandrole

ofthesevariesconsiderably.

section1:buildinganAI-readyorganization

HowareyouapproachingtheGenAIopportunity(chooseallthatapply)?

Wehavesetupacenterofexcellence

Wehaveidenti?edfamiliesofusecasesandspeci?cusecaseswithineachfamily

Weareworkingwithpartners/technology?rmstoworkondiferentusecases

WehavediferentfunctionsanddepartmentsexperimentingwithGenAIacrosstheorganization

53%

45%

53%

59%

TMForum,2023

DutchtelcoVodafoneZiggo,for

example,hasanAIcenterof

excellencewhichincludesthe

company’sdatascienceteam.

Telefónica,meanwhile,hasaglobalcentreofexcellencewithinits

networksandITgroupfunctionthatisresponsiblefordataand

AIarchitectures,forthemigrationtowardsacommondatamodel,andforresearchingAI

technologiesandsolutions.And

MiddleEasttelecomsgroupe&hasafederatedcenterofexcellence

witheachofthekeydepartments/functionshavingarepresentative.AnAIgovernancefunctionsitsontopofthiswithamissiontoensurethatusecasesarereusedacross

differentdivisions.

AIasaplatformfunction.Some

operatorshavebuilt–orare

building–platformsdesignedto

makeAIaccessibletodifferent

partsoftheorganization.Inthe

caseofGenAI,forexample,a

platformmaysitabovetheone,ormany,LLMsthattheCSPisusingtoensurethatthebestone–in

termsofcost,accuracyandsoon–isusedforaspecificrequirement.Forexample,VodafonehasanAIplatformwhichalsoprovides

self-servicetoolsandeducationalmaterialsfordifferentteamstobuildtheirownusecases.

SKTelecomhaswhatitcallsanIntelligencePlatformthrough

whichtheentireorganizationgainsaccesstotheLLMitisdeveloping.TheplatformcapturesallthedataabouthowtheLLMisbeingused,theapplicationsandtheinputs

andoutputs,andthisisthenfed

backintothemodeltohelptrainittobecomemoreaccurate.

Theprocessofbringing

individualsandfunctionstogethertohelpdrivebetteruseofAIis

havingtheaddedbenefitof

helpingtobreakdowndivisionalsiloes.IthasthepotentialtohelpempowertheITfunctionand

demonstrateitsvalueasadriverforbusinessimprovementby

interpretingandshowcasingdatagarneredfromacrossthe

organization,particularlywhen

datafromonedepartmentor

functionisblendedwithdatafromanother.Inoursurvey,53%of

respondentssaidtheyhave

differentfunctionsand

departmentsexperimentingwithGenAIacrosstheorganization

(seechart).

AIgovernanceimperatives

Manyofthegovernance

requirementsforAIsitwithinan

existingdatagovernanceprogram.However,additionalAI-specific

guardrailsareneededtoensure

thatAItoolsandsystemsare,andremain,safeandethical.With

GenAItechnologydevelopingso

quicklythechallengeisputtingin

placegovernancethatremains

rel