中國半導(dǎo)體行業(yè)-微處理器：投資者反饋的五個(gè)領(lǐng)域-2021.7.28-22正式版

GreaterChinaMCU:OrderofPreference

Exhibit1:Orderofpreference–GreaterChinaMCUcompanies

Source:MorganStanleyResearchestimates.Note:Sharepricesasat28July,2021.

2

China'sMCULocalization:KeyCharts

Exhibit2:China'sMCUmarkettoexpandata9%CAGR,toUS$8bnby

Exhibit3:WeexpectChina'sMCUself-sufficiencyratiotoriseto23%

2025

by2025

25%

23%

20%

19%

15%

13%

11%

10%

8%

6%

5%

0%

2020

2021e

2022e

2023e

2024e

2025e

Source:WorldSemiconductorTradeStatistics(WSTS),MorganStanleyResearchestimates

Self-sufficiencyratio(RHS)

Source:WSTS,companydata,MorganStanleyResearchestimates

Exhibit4:GainsinChina-designedsemiconductormarketshare(in

Exhibit5:China'sMCUmarketbreakdown,byapplication,2020

percentagepoints)

NAND

14%

DRAM

8%

Imagesensors

8%

PowerDiscrete

7%

MCU

7%

RFSemis

6%

AnalogIC

4%

SemiEquipment

3%

SmartphoneChipsets

2%

PC/ServerCPU

1%

0%

2%

4%

6%

8%

10%

12%

14%

16%

2020vs.2025marketsharegain(%)

Source:Gartner,MorganStanleyResearchestimates

Source:Chyxx,MorganStanleyResearch

3

Exhibit6:GlobalMCUcompaniesincomparison:GreaterChinaplayersenjoyhighergrowthandvaluationpremium

70.0e

Espressif

60.0e

50.0e

Gigadevice

BES

Chipsea

SinoWealth

P/E

40.0e

ShanghaiFudan

2022

30.0e

TI

NXP

Infineon

20.0e

STM

Renesas

Nuvoton

10.0e

0.0e

0%

10%

20%

30%

40%

50%

60%

70%

80%

2020-23EarningsCAGR

Source:MorganStanleyResearchestimates,Refinitiv

Exhibit7:PlaysonChina'sMCUlocalizationarecheaperthanotherlocalizationplaysinChinawithhighgrowthpotential

170

150

StarPower

AMEC

130

P/E

110

2022

90

Maxscend

Espressif

70

CRMicro

Gigadevice

50

BES

Chipsea

ShanghaiFudan

30

SinoWealth

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

2022EarningsGrowth

Source:MorganStanleyResearchestimates,Refinitiv

4

China'sMCULocalization:KeyTables

Exhibit8:MCUcompanystockselectionmatrix

Product

Software&

Bundle

Supplychain

Industry

2022earnings

Policy

Management

growth

coverage

ecosystem

sales

management

position

risk

&valuation

Gigadevice

J

J

J

J

J

J

J

K

Bestechnic

K

K

J

K

J

J

J

J

SinoWealth

J

K

J

K

J

J

J

J

Chipsea

J

K

J

K

K

J

J

J

Nuvoton

J

K

J

J

J

K

J

J

ShanghaiFudan

J

K

J

J

K

K

L

K

Espressif

K

K

K

K

K

K

L

J

Source:MorganStanleyResearch.BasedonourassessmentsrelativetoourMCUcoverage.

Exhibit9:SummaryofkeycompaniesacrossMCUapplicationsinbothTaiwanandChina,highlightingthekeyplayerinourcoverageineachapplication

Marketsize

%oftotal

ChinaMktSize

%oftotal

Keygloballeaders

Chineseplayers

Taiwanplayers

Subsectors

(US$bn)

(US$bn)

Homeappliance

1.0

6%

0.5

10%

STMicro

Cypress

Sinowealth

Cmsemicon

Holtek,Nuvoton

Renesas

Weltrend

Consumer/IoT

1.6

10%

0.7

14%

Microchip

STMicro

Espressif

Beken

MindMotion

Nuvoton

BES

JieLi

CECHuada

Realtek

Smartcard

2.2

14%

0.8

15%

NXP

Infineon

FudanMicro

GuoxinMicro

CECHuada

Smartmeter

0.2

2%

0.1

2%

NXP

Microchip

FudanMicro

ShanghaiBelling

SinoWealth

Industrial

3.9

25%

1.0

19%

Microchip

STMicro

Gigadevice

Sinowealth

MindMotion

Nuvoton

Infineon

Healthcare

1.2

8%

0.6

11%

Infineon

TI

Chipsea

Geehy

Sonix

Hycon

Auto

5.3

34%

1.6

31%

Renesas

NXP

BYD

Autochips

ChipOn

Nuvoton

Infineon

STMicro

SineMicro

Ingenic

GigaDevice

TotalMCUMarket

15.5

5.2

Source:Gartner,MorganStanleyResearch.Dataasof2020.

5

China'sMCULocalization:Cycletrackers

1.WemonitorchannelpricesinChina

Exhibit10:STM32F103VET6SpotMarketPriceTrend

Exhibit11:GD32F103VET6SpotMarketPriceTrend

(Rmbincl.tax)

STM32F103VET6

(Rmbincl.tax)

GD32F103VET6

500

140

450

400

120

350

100

300

250

80

200

60

150

100

40

50

20

-

Jul-19Oct-19Jan-20Apr-20Jul-20Oct-20Jan-21Apr-21

Jul-21

-

STM32F103VET6

Jul-19Oct-19Jan-20

Apr-20Jul-20Oct-20Jan-21Apr-21Jul-21

Source:Bom.ai,MorganStanleyResearch

Source:Bom.ai,MorganStanleyResearch

WelookatthemonthlysalesfromTaiwaneseMCUplayers–thisisanindicatorforthecontractmarket:WeexcludeNuvotonbecauseitacquiredPanasonicSemiconductorinSeptember2020.

Exhibit12:Taiwan–MCUcompanies:monthlysalestrend

Exhibit13:Taiwan–MCUcompanies:monthlysalestrend(ex-Nuvoton)

6000

5000

4000

3000

2000

1000

0

Jan-15 Jan-16 Jan-17 Jan-18 Jan-19

TotalTWMCUCompaniesMonthlyRevenue(NTDmn)

Source:TEJ,MorganStanleyResearch

Note:CompaniesincludeNuvoton,Hycon,Weltrend,Sonix,Holtek

300%

250%

200%

150%

100%

50%

0%

-50%

Jan-20 Jan-21

YoYGrowth(RHS)

1800

140%

1600

120%

1400

100%

1200

80%

1000

60%

800

40%

600

20%

400

0%

200

-20%

0

-40%

Jan-15

Jan-16

Jan-17

Jan-18

Jan-19

Jan-20

Jan-21

TotalTWMCUCompaniesMonthlyRevenue(NTDmn)

YoYGrowth(RHS)

Source:TEJ,MorganStanleyResearch

Note:CompaniesincludeHycon,Weltrend,Sonix,Holtek,butexcludeNuvotonbecauseitstartedtoconsolidateNTCJfromSept2020

6

China'sMCULocalization–WeAreattheTippingPoint

KeyassumptionsdrivingourviewsonChina'sMCUlocalization

MCUlocalizationhasbeenslowinChina,butwebelieve2020/2021isthetippingpointtoanacceleratedpace.

WeexpectChina'sMCUmarkettocontinuetoshowstronggrowth,ata44%CAGR,2020-25,vs.China'sGDPgrowthof3%to8%.

WeexpecttheMCUlocalizationratioinChinatoriseto23%by2025,fromthecurrent6%.

WeexpecttheMCUTAMinChinatorisetoUS$8bnfromthecurrent

US$5bn,andtheChineseMCUvendors'TAMtoincreasefromUS$300mnto

US$2bnby2025.

WeintroduceasystematicmethodtoassessMCUcompaniesinChina,byevaluating:

MCUproductportfoliocoverageandhardwarespecs,

Firmwarecapabilityandecosystemcomprehensiveness,

Companyhistoryandmanagementexperience,

Salesbundlingsalescapability,

Supplychainmanagement,

Targetedmarketandindustryposition,

Financialperformanceandvaluationcomparison,and

Policyrisk.

WethinkChineseplayers'penetrationinautomotiveapplicationscouldstillbelimitedinthenext3-5years,butwebelievesomelocalMCUvendorscouldhavesomeexposureinnon-criticalautoMCUs.

Wemaintainourviewthatthelogicsemiup-cyclecouldpeakin4Q21or1Q22.MCUcompaniesalsobenefitfromtheup-cycle,albeitwithmuchlessASPexpansionvs.thedistributors.

WebelieveGigaDeviceisthekeyproxyandcouldbethestrongestplayerforgeneralpurposeMCUsinChina.However,theendmarketisfragmentedandbigenoughformultiplelocalplayerstogrow,includingTaiwanesevendors.

GlobalMCUplayers(STM,NXP,Renesas,Infineon,TI)arelikelytolosesomemarketsharebutcouldcontinuetogrowinabsoluteterms,givenmarketgrowthandincreasingexposuretotheautosegment.WeexpectGMtoshowsteadyimprovementthankstobetterproductmix.Overall,wethusmaintainourviewontheglobalpeers.

7

Webelieve2021willbethebaseyearforChina'sMCUlocalization,andprogresswillnowaccelerate

Weexpectlocalizationtoacceleratefrom2021,basedon:

Technologicalimprovement:Manylocalcompanieshavemanagedtoclosethehardwaregapwithforeignvendors.Thereisnotmuchperformancedifferencebetweenleadinglocalvendorsandforeignplayers.SomelocalplayerscouldalsoleverageTSMC'slowpowerprocesstoreducestandbypowerconsumption.

Morecomprehensivesoftwaresupportandecosystem:Overtheyears,localplayershavestartedtoworkmorecloselywithendcustomersandsystemintegrators.MoreandmoreMCUseminarsarebeingheldtohelptoexpandtheecosystemandsoftware.

Enhancementoftheproductportfolio:Perourassessment,localvendorsareintroducingmoreMCUproductstocovermoreMCUapplications.Forexample,GigaDevicehasbroadeneditsproductroadmaptoinclude400-500MCUs.Wealsofindthattalentfromoverseasisjoininglocalfirms,helpingtostrengthenMCUproductportfolios.

Foundrycapacityconstraint:China'sfoundryconstraintopenedawindowforlocalvendorstoenterthesupplychain.OncetheChinesesuppliersenterthesupplychain,itwould,ingeneral,bedifficulttobereplacedbyforeignplayers,especiallyifthere'snosignificantdifferenceinquality.

Increasingurgencytolocalize:Thefoundryconstrainthasalsospurredlocalizationawareness.NotonlyUSvendors,butalsoEuropeanandJapanesevendorshaveprioritizedlocalcustomers,especiallyintheautosegment,amidoverallfoundryconstraint.Chinesecustomersarelikelytobemorewillingtogiveopportunitiestolocalvendorsinthefuture,evenintheabsenceofsupplytightness,givenChina'slocalizationinitiative.

WeexpectChina'sMCUlocalizationprogresstoacceleratefrom2021.

Weexpectthelocalsufficiencyratetogrowfrom5-10%in2020to20-25%by2025,andthenrisefurtherto40-45%by2030.

Currently,thenon-autoandnon-industrialmarketsaccountfor50%ofChina'sMCUmarket.

8

Exhibit14:MCUlocalsufficiencyrate

25%

23%

20% 19%

15%

13%

11%

10%

8%

6%

5%

0%

2020 2021e 2022e 2023e 2024e 2025e

Self-sufficiencyratio(RHS)

Source:MorganStanleyResearch

TheMCUmarketinChinacouldalsooutgrowtheglobalmarket

Ontopofhigherpenetration,wealsoexpecta9%CAGRfortheMCUTAMinChinain,2020-25,fasterthanglobalMCUTAMgrowthat7%.China'sMCUmarketgrowthistiedtoconsumerIoT,industrialIoT,andincreasingEV/ADASpenetration,whichweexpectwillenjoyhighergrowthratesthanthetraditionalautoandconsumerelectronicsmarketsin2020-25.

Exhibit15:China'sMCUTAM Exhibit16:China'sMCUshare,byapplication,2020

9.0

8.1

8.0

7.6

7.0

7.0

6.5

6.2

6.0

4.6

4.9

4.9

5.2

5.0

4.0

3.8

4.0

3.0

2.0

1.0

0.0

2015

2016

2017

2018

2019

2020

2021e

2022e

2023e

2024e

2025e

ChinaMCUMarket(US$bn)

YoYGrowth(RHS)

25%

20%

15%

10%

5%

0%

-5%

Source:Gartner,MorganStanleyResearchestimates

Source:Gartner,MorganStanleyResearch

9

WebelievetherewillbehigherpenetrationofsmartappliancesinChinathaninmanyothercountries.Weforecast12%annualgrowthforsmartappliancesthrough2030,basedonourforecastthat100%penetrationwillbeachievedbythen,asdrivenbyChina'surbanizationobjective,whichdrivesoveralldemandforhomeappliances.Thiscompareswitha2019-30CAGRof5%fortheoverallhomeappliancesegment.Inturn,weprojectthenumberofIoTappliancesperhouseholdwillriseto7unitsby2030fromjust1today,alsodrivenbyChina'surbanizationinitiatives.

Exhibit17:MarketsizeofChina'shomeapplianceindustry

Rmb'bn

1,800

1,415

1,600

1,400

1,200

810

818

818

880

940

1,000

797

616

660

800

600

400

200

-

2015

2016

2017

2018

2019

2020E2021E2022E

2030E

Whitegoods

Smallappliances

Others

Source:iResearch,AllViewCloud(AVC),MorganStanleyResearchestimates.

Exhibit18:MarketsizeofChina'sIoT-enabledsmarthomeappliances

Rmb'bn

1,600

1,415

1,400

1,200

1,000

680

800

615

600

421

529

315

400

201

71

117

200

-

2015

2016

2017

2018

2019

2020E2021E2022E

2030E

Whitegoods

Smallappliances

Others

Source:iResearch,AVC,MorganStanleyResearchestimates.

Ontheindustrialside,IoTtechnologyisbeingintegratedwithmanufacturingprocessesacrosssupplychains,productsandservices.IIoTallowsanetworktocollectsignificantamountsofdatafromconnectedmanufacturingtools,equipment,and/orvarioussetsofhardwaredevices.

GivenChina'ssizablenumberoffactoriesandworkersinmanufacturing,alongwithsubstantialdata,wethinkthereissignificantpotentialindevelopingIIoTtechnologies.AccordingtoBostonConsultingGroupdata,asof2015,USmanufacturingcostswerelessthan5%higherthaninChina.Chinaisthereforefacingpressuretoupgradeitsindustries,particularlyinviewofgreatercompetitionfromcheapermarketsintherestofAsia.WebelieveitspushtodevelopanindustrialIoTandcreateitsownecosystemwillhelpitmaintainacompetitiveedgeinmanufacturing.

10

Exhibit19:IndustrialIoT–positivedataflowforvaluecreation Exhibit20:China'sIIoTecosystemadoptionphasesandrespectivebeneficiaries

Source:MorganStanleyResearch.

Source:MorganStanleyResearch.Note:ReferstoSanyHeavyIndustries.

TheautomotivesegmentisanotherimportantmarketforMCUs.AutoMCUsarecalledelectroniccontrollerunits(ECU).Weexpectnewenergyvehicles(NEV)andautonomousdriving(fromlevel1tolevel4)todrivestrongerMCUdemandinChina.ForNEV,weforecastEVpenetrationinChinatorisefrom<10%in2020to24%by2025.Semicontentforadvanceddriverassistancesystems(ADAS)willalsoincreaseincomingyears;wenotethatlevel3semidollarcontentisUS$400percar,onaverage,vs.level2beingbelowUS$100.

Exhibit21:Bull,base,andbearcases–NEVannualsalesvolume

Source:ChinaAssociationofAutomobileManufacturers(CAAM),MorganStanleyResearch.e=MorganStanleyResearchestimates.

11

Exhibit22:Averagesemiconductorcontentpercar,bylevelofautomation

Source:InfineonTechnologiesAG

FasterMCUmarketgrowth+acceleratedMCUpenetration=6xmarketsizeby2025

Weprojectthatarisingself-sufficiencyrate,alongwithongoingMCUmarketgrowth,willyieldadditionalTAMofaroundUS$1.5bn(fromUS$250-300mnin2020toUS$2bnby2025),basedon:

A9%CAGRforMCUinChinain2020-25e,and

Localsufficiencyrateto20-25%in2025eand40-45%in2030e,from<5%in2020(basedonforecastsderivedfromChina'surbanizationinitiatives).

Thistranslatesintoa40-50%revenueCAGR,2020-25,or30%revenueCAGRin2020-

Wealsohighlightthatmostlocalvendorsarestillverysmall,andoperateinafragmentedmarket,suggestingthatthehigher-qualitylocalplayerscouldmorereadilyenjoystrongrevenuegrowthoverthenextfewyears.

12

Exhibit23:TotaladdressablemarketforChina'sMCUvendors,2020-25e

2000

25%

23%

1800

1600

19%

20%

1400

1200

13%

15%

1000

11%

800

8%

10%

6%

400

5%

200

0

0%

2020

2021e

2022e

2023e

2024e

2025e

ChinaMCUVendor'sTAM(USDmn)

Self-sufficiencyratio(RHS)

Source:MorganStanleyResearchestimates

Exhibit24:TotalrevenueofChina'sMCUvendorsin2020vs.TAMofChina'sMCUvendorsin2025

(USDmn)

2,000

1818

1,800

MCURevenue

Otherrevenue

1,600

1,400

1,200

1,000

800

600

400

200

-

BES

VendorsMCUChinaforTAM

Nuvoton

GigaDevice

FudanShanghai

WealthSino

Espressif

Chipsea

Source:Companydata,WSTS,MorganStanleyResearchestimates

ToomanyMCUcompaniesinChina–whicharelikelytoprovidesoundinvestmentopportunities?Introducingourstockselectionmethodology

TheMCUmarketremainsveryfragmentedinChina,withatleast30companiesinthesegment.Weintroduceourcompanyselectionmatrixtoidentifycompetitiveplayersingrowthareas.Weevaluatecompaniesbasedonthefollowingeightcriteria:

MCUproductportfoliocoverageandhardwarespecs

Firmwarecapabilityandecosystemcomprehensiveness

Companyhistoryandmanagementexperience

Salesbundlingcapability

13

Supplychainmanagement

Targetedmarketandindustryposition

Financialperformanceandvaluationcomparison

Policyrisk

Exhibit25:MCUcompanystockselectionmatrix

Product

Software&

Bundle

Supplychain

Industry

2022earnings

Policy

Management

growth

coverage

ecosystem

sales

management

position

risk

&valuation

Gigadevice

J

J

J

J

J

J

J

K

Bestechnic

K

K

J

K

J

J

J

J

SinoWealth

J

K

J

K

J

J

J

J

Chipsea

J

K

J

K

K

J

J

J

Nuvoton

J

K

J

J

J

K

J

J

ShanghaiFudan

J

K

J

J

K

K

L

K

Espressif

K

K

K

K

K

K

L

J

Source:MorganStanleyResearch.BasedonourassessmentsrelativetoourMCUcoverage.

Valuationcomparisonwithglobalandlocalpeers

China'sMCUcompaniestendtocarrymoreexpensivevaluationsvs.globalpeers.Wethinkthisismainlybecauseoftheirhigherperceivedgrowthrate,giventheampleroomtogrowinthefuture.Still,whencomparingwithotherhigh-growthlocalizationplaysinChina,suchassemiequipment,RFsemis,andpowersemis,thevaluationsappearreasonable,giventhelowerPEGpremium.

Exhibit26:GlobalMCUcompaniesincomparison:GreaterChinaplayersenjoyhighergrowthandvaluationpremium

70.0e

Espressif

60.0e

50.0e

Gigadevice

BES

Chipsea

SinoWealth

P/E

40.0e

ShanghaiFudan

2022

30.0e

TI

NXP

Infineon

20.0e

STM

Renesas

Nuvoton

10.0e

0.0e

0%

10%

20%

30%

40%

50%

60%

70%

80%

2020-23EarningsCAGR

Source:MorganStanleyResearchestimates,Refinitiv

Exhibit27:China'sMCUlocalizationplaysarecheaperthanotherlocalizationplaysinChinawithhighgrowthpotential

170

150

StarPower

AMEC

130

P/E

110

2022

90

Maxscend

Espressif

70

CRMicro

Gigadevice

50

BES

Chipsea

ShanghaiFudan

30

SinoWealth

0%

10%

20%

30%

40%

50%

60%

70%

80%

90%

2022EarningsGrowth

Source:MorganStanleyResearchestimates,Refinitiv;note:Chipsea's2022earningsgrowthhighergivennon-operatinglossesin2021.

14

WhereAreWeintheMCUCycle?

TheMCUcyclemoveswiththelogicsemicycle,butcouldlastslightlylonger

Keyupdatesonoursemicyclesignposts:Overall,therehavebeennosignificantturningpointsinthepasttwomonths,asidefrom:

Inventory–PCchannelandOEMinventoriesincreasedfurther,soperhapssupplyiscatchingup.

Demand–smartphonedemandinChinastabilizedinJunewiththe618salespromotion.Ontheotherhand,clouddemandin2Hisstrongerthanexpected.

Foundrysupply–thereispotentialrisktoSMIC's28nmwafercapacityexpansiongivenanequipmentlicenseissue.

Latestageinlogicsemiinflation:Weexpectfurtherpriceincreasesin3Q21incertainsub-segments,suchasLCDdriverICs,powersemi,memory,rawwafers,andtrailing-edgewaferfoundry,rangingfrom5%to10%.Ontheotherhand,smartphonesemis(imagesensors,5GSoC,RFcomponents)andcryptomininghardwareareunlikelytogetfurtherpricehikes.Companies’3Q21guidanceonpricingandgrossmarginvs.Streetexpectationsshouldbekeytodeterminewhetherstockscancontinuetomakegains.

TheMCUcycleispartiallytiedtothelogicsemicycle...Wethinktheoveralllogiccyclepeakcouldhappenasearlyas4Q21,whenfoundrytightnessstartstoeaseandpricehikesstop.TheglobalMCUup-cycleisbasicallydrivenbyoverallfoundrytightnessinthelogicsemimarket.Thus,webelievetheMCUcyclemoveswiththeoveralllogicsemicycle.Intermsoftiming,MCUsupplyincreasesfromthefoundrysidecouldbelaterthanforlogicsemiproducts.That'sbecauseMCUsrequireeFlashfabricationexpertise,whichcouldstillbeabarriertoentry,especiallyfortheChinesefoundriesvs.otherfoundryproducts.

…butGMhasnotshownsignificantexpansion:WealsohighlightthatMCUcompanies'GMshaven'tshownsignificantimprovementfromthesemiup-cycle.ThisismainlybecausetheMCUmarketremainsquitefragmented,andMCUproductsbytheirnaturearenotcommodities.Also,alotofcustomershaveverycloseworkingrelationshipswiththeMCUvendors,sothere'sbeennosignificantpricingadjustmentfromtheMCUsuppliersinthepastquarters.Infact,mostMCUcompaniesjustpassedthroughthecostsfromfoundrytoclients,andsomeeventriedtosharesomeofthecostpressurewithcustomers.

Webelievethissuggeststhatmarginscouldshowsomeresiliencyifthere'sadown-cycleinto2022.However,incontrast,spotmarketMCUpricinghasrisen20-60xYTD,inresponsetoshortage.Thespotmarketaccountsforlessthan10%ofthetotalMCUmarket.Weviewitasagoodsentimentindicator,butcontractsofferedbyMCUvendorsaremoremeaningfulwhenjudgingthecycle,inourview.

15

AutoisabigapplicationforMCU,at30-40%ofthemix,whichweviewassustainabl:

AutoMCUproductionhasahigherratioofin-houseproductionbytheglobalMCUpeers.Thissuggeststhatthesupply/capacityincreasewillbemoremoderatethanthefoundryproductionfornon-autoMCU.TheMCUcycleforautoscouldlastlonger,into4Q21oreven1H22,whichcouldhelpcushionthepotentialdownside.

Exhibit28:Grossmargintrendscomparisoninthetechvaluechain–weareinthelatestageoflogicsemiinflation

Source:Companydata,MorganStanleyResearch(e)estimates

WhyownMCUdesignhousesifthere'sadownturn?

Thoughpastperformanceisnoguaranteeoffutureresults,historydoestendtorepeatfortechcyclesandcanofferguidanceastohowsegmentsmightperform.Inpreviousdownturns,Asiantechstocks'EV/salesmultiplescompressed~25%.Earningsdecline13-100%inatypicaldowncycle,andrevenuefalls2-47%inaspanoffourquarters.Dispersionwithinthesectorissignificantonthewaydown.Stockscorrect46%,onaverage(2001-2002;2008-2009;2020downturns),withsemimaterialsandMLCCtendingtobetheweakestperformersvs.cloudhardware,memory,andICdesignoutperforming.Thus,wethinkMCUdesignhousescouldbefavorableduringacyclicaldownturn.

16

Exhibit29:Shareperformancebysegmentfollowingthe3Q18downturn

Source:Refinitiv,MorganStanleyResearch

Exhibit30:ICDesignandDataCenterHardwarestockshavebeen Exhibit31:…whilesalesinSemiEquipmentandSmartphone

moredefensiveindownturns... Hardwarehavebeenaffectedmostfrompeaktotrough,withan

averagedecreaseof22%

Source:Companydata,Refinitiv,MorganStanleyResearch

Source:Companydata,Refinitiv,MorganStanleyResearch

17

DisclosureSection

TheinformationandopinionsinMorganStanleyResearchwerepreparedoraredisseminatedbyMorganStanleyAsiaLimited(whichacceptstheresponsibilityforitscontents)and/orMorganStanleyAsia(Singapore)Pte.(Registrationnumber199206298Z)and/orMorganStanleyAsia(Singapore)SecuritiesPteLtd(Registrationnumber200008434H),regulatedbytheMonetaryAuthorityofSingapore(whichacceptslegalresponsibilityforitscontentsandshouldbecontactedwithrespecttoanymattersarisingfrom,orinconnectionwith,MorganStanleyResearch),and/orMorganStanleyTaiwanLimitedand/orMorganStanley&CoInternationalplc,SeoulBranch,and/orMorganStanleyAustraliaLimited(A.B.N.67003734576,holderofAustralianfinancialserviceslicenseNo.233742,whichacceptsresponsibilityforitscontents),and/orMorganStanleyWealthManagementAustraliaPtyLtd(A.B.N.19009145555,holderofAustralianfinancialserviceslicenseNo.240813,whichacceptsresponsibilityforitscontents),and/orMorganStanleyIndiaCompanyPrivateLimited,regulatedbytheSecuritiesandExchangeBoardofIndia(“SEBI”)andholderoflicensesasaResearchAnalyst(SEBIRegistrationNo.INH000001105);StockBroker(BSERegistrationNo.INB011054237andNSERegistrationNo.INB/INF231054231),MerchantBanker(SEBIRegistrationNo.INM000011203),anddepositoryparticipantwithNationalSecuritiesDepositoryLimited(SEBIRegistrationNo.IN-DP-NSDL-372-2014)whichacceptstheresponsibilityforitscontentsandshouldbecontactedwithrespecttoanymattersarisingfrom,orinconnectionwith,MorganStanleyResearch,andtheiraffiliates(collectively,"MorganStanley").

Forimportantdisclosures,stockpricechartsandequityratinghistoriesregardingcompaniesthatarethesubjectofthisreport,pleaseseetheMorganStanleyResearchDisclosureWebsiteat/researchdisclosures,orcontactyourinvestmentrepresentativeorMorganStanleyResearchat1585Broadway,(Attention:ResearchManagement),NewYork,NY,10036USA.

Forvaluationmethodologyandrisksassociatedwithanyrecommendation,ratingorpricetargetreferencedinthisresearchreport,pleasecontacttheClientSupportTeamasfollows:US/Canada+1800303-2495;HongKong+8522848-5999;LatinAmerica+1718754-5444(U.S.);London+44(0)20-7425-8169;Singapore+656834-6860;Sydney+61(0)2-9770-1505;Tokyo+81(0)3-6836-9000.AlternativelyyoumaycontactyourinvestmentrepresentativeorMorganStanleyResearchat1585Broadway,(Attention:ResearchManagement),NewYork,NY10036USA.

AnalystCertification

Thefollowinganalystsherebycertifythattheirviewsaboutthecompaniesandtheirsecuritiesdiscussedinthisreportareaccuratelyexpressedandthattheyhavenotreceivedandwillnotreceivedirectorindirectcompensationinexchangefor