自動(dòng)駕駛出租車將如何重塑城市世界

1、Executive summaryIn this report, we refine our picture of what the future of mobility could look like, despitealltheuncertaintiesaroundtechnologyandregulation.UBSEvidenceLab hasdevelopedareal-timesimulationtomodeltheeffectivenessofarobotaxifleet in New York City. Our complex dispatching algorithm pe

2、rforms dynamic optimal route generation and connects riders with vehicles in the most efficient way possible.This powerful simulation has enabled us to quantify a wide range of key metrics, including average wait and trip times, the number of kilometres driven by each robotaxi, the average fare, the

3、 fleet profitability margin, the excess capacity required, and the number of charging stations needed. The simulation also gives us unique insight to improve our forecasts for the potential size of the global robotaxi fleet, as well as implications across developed and emerging marketsOver the past

4、five years, we have tried to help investors form a view on how the urban world will be reshaped by the introduction of new forms of mobility and what that could mean for various industries. Our view remains:The launch of robotaxi fleets will contribute materially to growth in the ride- on-demand mar

5、ket. We see the robotaxi fleet management revenue pool reaching more than $2trn by 2030E.Not a winner-takes-it-all market. The market share split is uncertain. While the early entrants certainly have an edge, we note that the barriers to entry in an autonomous world, once the technology is available

6、, should drop materially.Sharp price deflation. We estimate the average trip cost will fall by at least 80% compared to the average trip today. Most of this will be passed through to the users/passengers andregulators.The mass adoption of robotaxis will have far-reaching implications across many ind

7、ustries, and will create new revenue opportunities. Sectors that shouldbenefitthemostincludeInternet,Utilities,TelcosandTires.UBS Evidence Lab has developed a real-time simulation to model the effectiveness of a robotaxi fleet in New York CityFigure 2: Ride-on-demand profitability mapUBS viewVolumeM

8、arket sizeMarket sharePriceRevenuesRobotaxifleetwillmateriallyboosttheride- on-demand revenueVolumeMarket sizeMarket sharePriceRevenuesProfitabilityNotawinnertakesitallsituationgiven lower barriers to entry andregulationProfitabilitySharppricedeflation(80%).Benefittobe passed through to users andreg

9、ulatorsVariable costsCostsFixed costsIncreasingtrendstodevelopmoreefficient optimiser and central costVariable costsCostsFixed costsDecreasing trends as vehicle becomes autonomous,Decreasing trends as vehicle becomes autonomous,fewercarsrequiredandpoolingSource: UBSWhats unique about this report?UBS

10、EvidenceLabhasdevelopedareal-timesimulationtomodeltheeffectiveness of a robotaxi fleet in NYC, leveraging a program-based mathematical model. At every15-secondframeinthesimulation,thealgorithmperformsdynamicoptimal route generation and passenger-vehicle assignment considering online vehicle capacity

11、 and rider demand. Vehicles that are not assigned in a given frame are subsequently rebalanced according to the outstanding riderdemand.UBS Evidence Lab has built two scenarios into the simulation:Scenario1non-commuter,withthegoalofminimisingtheNYCrobotaxi fleet size during weekdays andweekends;Scen

12、ario 2 commuter, with the goal of maximising the number of commuterswitharobotaxifleetapproximatelyequalinsizetothecurrentNYC taxi fleet, in addition to the existing demand fortaxis.UBSanalystscoveringatotalof11sectorsrangingfrombatterycellproductionto utilitieshaveleveragedthisuniquesimulationtomak

13、econclusionsontheimpact of the introduction of robotaxi fleets for their respectiveindustries.UBS Evidence Lab built acomplex dispatching algorithm to maximise efficiencies of the vehiclefleetScenario 1 - weekdayScenario 1 - weekendScenario 1 - weekdayScenario 1 - weekendScenario 2 - weekdayTrip sta

14、rt time18:0408:1417:57Trip end time18:1508:3318:22Fare paid (in $)10.0714.7022.97Profitability margin (%)93.3092.5092.32Trip length (in km)3.335.0111.00Trip time (in min)11.2319.2624.43Wait time (in min)1.252.060.86Energy consumption (in kWh)0.400.601.32Source: UBS Evidence LabAt this stage, the sim

15、ulation assumes one passenger per vehicle. However, we believe the dispatching algorithm could be further improved to reflect pooling the use of the vehicle by several passengers, which should lead to even further efficiencies.Formoredetailsonthemethodology,pleaserefertothefinalsection of this repor

16、t How does the robotaxi simulationwork?What are the key findings from the simulation?Our analysis of the data from the simulation has enabled us to gauge some key metrics, such as the average wait and trip time, the number of km driven byeach robotaxi, the average fare paid by the passengers, the pr

17、ofitability margin of the fleet,theexcesscapacityrequired,theelectricityconsumptionofthefleet,andthe number of charging stations required. HYPERLINK l _bookmark0 Figure 4 shows only a fraction of the data that we have been able to extract from the simulation.Figure 4: Key numbers to remember-$2trnpo

18、tentialrevenuecouldbereachedfromrobotaxifleetsby2030 The fleet of taxis could be reduced by 67% today90%+profitability margin could be generated by managing robotaxi fleets Ittakesonly2fortherobotaxisfleetreachbreakeveneachday The cost of a (for the passengers) could reduce by80%+Only 9charging stat

19、ions are required cover the daily needsWith a daily running cost of $40daily, the market entry are very lowRobotaxi will price compete with public transport as cost per trip will rapidly fall 330k trip requests on a daily basis), (2) our new forecast split between developed markets (DM), EM, and Chi

20、na; and (3) comments from industry participants that developing the technology might take longer than initially planned. We reduce our forecast for the fleet of robotaxis from 26m to 11m in 2030. Furthermore, we now assume that robotaxis will represent c5% of new car sales (from 12% before) in 2030.

21、We have built an HYPERLINK /shared/d218tKFglY interactive model, which enables investors to gauge the size of the robotaxi fleet and the impact for various industries using their own assumptions.What are the key implications for autos?We estimate that the penetration of robotaxi sales will be 5% in

22、2030 (as a percentageofnewcarsales).Weexpecttheoverallfleettoreach11mvehiclesby then,equivalentto1%oftheglobalcarparcandc4%ofthemilesdrivenglobally. All those metrics should expand exponentially as the adoption rate of robotaxis increases further: we currently assume 8% of the urban population globa

23、lly in 2030 (including 20% in DM), rising to c37% in 2040.We estimate that robotaxis could represent on average 16m units sold per year between 2016 and 2050. In the long term, we see new car sales running c5-10% lower than our current estimates. However, in the medium term, new car sales should be

24、slightly supported until 2027 and then drop off as the adoption rate of robotaxisaccelerates.Then,newcarsalesshouldrecover,thanksto(1)thehigher utilisation rate of robotaxis (about 10 times higher than that of a private car); andthe faster replacement velocity we assume an average life of a robotaxi

25、 of about three years compared to about 10 years for a privatecar.We estimate that robotaxis will represent more than 10% of global EV sales in 2030.Thesimulationalsoshowedthatthedailycostofrunningaelectricrobotaxi fleet is about one-third cheaper than that for an ICE (internal combustionengine) fle

26、et.Robotaxiscouldalsohelpstabilisethegridduringpeakelectricitydemand(if there is lower demand for transport).Tire makers should benefit from the higher number of miles driven. Despite the shrinking car parc, the number of kilometres driven is set to increase for two key reasons: (1) the time spent i

27、n vehicles should increase due to the lower fares chargedtopassengers(weseeapricereductionofmorethan80%);and(2)there should be a shift from public transport to robotaxi fleets. The simulation shows that a robotaxi should drive about 6x more than a private car today. This would lead to the number of

28、kilometres driven per car increasing from 15,000/year toWe delay our adoption forecast by a year or two5-10% negative impact on new car sales in the long runSharp acceleration in EV penetrationTire makers best positioned to benefit in Autos sectorabout 15,600/year. Therefore, even if we estimate tha

29、t the fleet size could shrink by two-thirds with the introduction of autonomous vehicles, the kilometres driven would still increase on a like-for-like basis.We see the revenue opportunity for tire makers related to the launch of robotaxi fleets reaching $19bn in 2030E. To put things into perspectiv

30、e, this would be larger than what Michelin makes today selling passenger car tires. It would also support industry volume growth, with an annual growth rate of close to 1% between today and 2030E (and most of the impact would be felt after 2025). The netrevenuepool(i.e.adjustedforlowernewcarsales)fo

31、rtirescouldincreaseby up to 50% by 2040E.How do we connect the simulation with our global forecasts?As in the past, we use the forecast for the urban population provided by the UN. We then apply the adoption curves that we described earlier. For each region (developed markets, China and emerging mar

32、kets), we model different adoption curves based on income, infrastructure, cost of public transport, etc. We also assumethateachrobotaxiusermakestwotripsperdayinourbasecase.Finally, thesimulationshowsthatthefleetof4,500robotaxiscanservicearound330ktrip requestsdaily.Weapplythisratiotoquantifyhowmany

33、robotaxisarerequiredin each region.What are the largest revenue pools that could emerge from the mass adoption of robotaxis?The launch of robotaxi fleets should have far reaching implications across many industries. The mass adoption of robotaxis could materially boost the revenue pools of several s

34、ectors ranging from utilities to semis. We leverage this unique simulationinordertoconnectthedots.Inthissection,analystscovering11sectors haveleveragedtherobotaxisimulation,identifyingthekeymetricsthatmatterthe mostandansweringthekeypivotalquestionsthataremostrelevant. HYPERLINK l _bookmark1 Figure

35、5 shows the potential revenue boost the launch of robotaxi fleets could haveonvariousindustries.Our HYPERLINK /shared/d218tKFglY interactivemodeldiscussedlaterenablesinvestors to change the key assumptions. with a potential net revenue uplift of close to 50%Figure 5: The launch of robotaxi fleets co

36、uld have a material impact on the revenue potential of various industries HYPERLINK /shared/d218tKFglY (/shared/d218tKFglY)Boost from the launch of robotaxi fleetsToday2030E2040E$2,293bn$9,831bnx4$237bn$1,151bnx4$237bn$1,151bn$2,000bnx7x5AV productionx7x5TireElectricityCharging stations$168bn$19bn$1

37、38bn$3,000bn$319bn$42bnx8$3,000bn$319bn$42bnx8x7Battery cellsSemiconductors$16bn$35bn$98bnx3x3x4$42bn$18bn$72bnx4Source: UBS estimatesNote:ForUtilities,wecomparepowerpricesatthepumpwithawholesalelevelutilityrevenuepool,whichiswhytheactualrevenueupsidewilllikelybemoderately below 10%Figure 6: What do

38、es the robotaxi simulation mean for various industries?SectorTherobotaxisimulationshows.Our key conclusions areOEMsDailyrevenueoftherobotaxifleetis$3.3m(4,500vehicles)Givenprofitabilityandlowercyclicalityofrevenues,OEMsshouldconsiderenteringthefield RunningcostofaEVrobotaxiisathirdcheaperthananICEta

39、xiFurther acceleration of EV penetrationLong-term new car sales could be 5-10% lower due to robotaxi adoptionSuppliersRobotaxi fleet profitability marginis90%Oncethevehiclebecomesautonomous,barrierstoentrybecomemateriallylower Runningcostofrobotaxiis$40dailyRobotaxi sales will represent 10% of BEV s

40、ales in2030TiresRobotaxidrivesonaveragec250kmdaily(or90kkmp.a.)Tireconsumptionofrobotaxisis6xhigherthanthatofprivatecars Sizeoftaxifleetcouldbereducedbytwo-thirdsRevenueboostcouldreach$19bnin2030and$138bnin2040Numberofkmdrivenwillincrease(upto20 xperhour)Netrevenueupliftcouldbe50%oftodaysindustrysal

41、es RunningcostofEVrobotaxiisonethirdcheaperthanforanICEtaxiMix impact (larger tires) should bepositiveInternetDailyrevenueofrobotaxifleetis$3.3m(4,500vehicles)Technology&platformcompanieshaveampleflexibility/optionalityinpotentialbusinessmodels Robotaxi fleet profitability marginis90%Despitenear-ter

42、minvestments,AVdeploymenthasthepotentialtoimproveprofitabilityofTakesupto2hoursforfleettoreachprofitabilitybreakeveneachdayoperating the network Trip fare could be reduced by 80%Daily running cost of a robotaxi is $40UtilitiesDailyelectricityconsumptionis65,000kwh(for4,500robotaxis)Revenuecouldreach

43、$42bnin30and$300bnin40,or6%ofglobalelectricityconsumption RunningcostofanEVrobotaxiisathirdcheaperthanforanICEtaxiChargingandgridupgradesonlyrepresent10%oftotalcapexovernext10yearsOnly9chargingstationsrequiredtocoverdailyelectricityneedsoffleet Electricity represents c14% of robotaxis dailycostTelco

44、Theaveragedailyutilizationrateofthefleetisc50%TherevenuepoolcouldreachUS$100-120bnby2030translatingintorevenuepercarofUS$10kper annumTheaveragefareis$10andtheaveragetriplengthis10minIt is not clear how much incremental capex will be requiredEachrobotaxidrivesc250kmdailyThelatencyrequirementsforrobot

45、axiscouldbehigher,demandingahigherdegreeofreliabilityBatteriesRunningcostofanEVrobotaxiisathirdcheaperthanforanICEtaxiRevenueboostcouldreach$35bnin2030and$98bnin2040,6xhighervstodaysmarketsize Robotaxiwilldrivec90kkmperyearRobotaxi sales will represent 10% of BEV sales in2030We assume life-cycle of

46、robotaxi will be 3 yearsEach robotaxi will have to (fully) recharge every two daysSemisRunningcostofaEVrobotaxiisathirdcheaperthananICEtaxiRevenueboostcouldreach$18bnin30and$72bnin40,almost2xtodaysmarketsize Long-termnewcarsalescouldbe5-10%lowerduetorobotaxiadoptionRobotaxicouldbehelpfulaccelerantof

47、thesemiconductorcontentpercarBy2030EL5penetrationcouldbe5%ofnewcarsalesShift to EV adds $550 per car to the average of c$450 per cartoday)L4/L5 could carry as much as $2,000 per carInsurance Insurance costs represent 18% of robotaxi operatingcostsInsurance could be reduced, but liability is likely t

48、o increase, leading to stable insurance costsoverallSize of taxi fleet could be reduced by two-thirdsCar rental Daily revenue of robotaxi fleet is $3.3m (4,500vehicles)Rental companies appear like natural partners for platforms Robotaxi fleet profitability marginis90%Rentalcompaniesalreadyhaveyearsw

49、orthofdataoncustomerTakes only two hours for fleet to reach profitability breakevenWe estimate that cleaning represents 5% of daily operating cost of fleetReal EstateNetworkinthesimulationhas770kilometresofroadLaunch of fleets could reduce on-street parking, freeing up land foralternativesLocationde

50、sirabilitymaychangeifcommutetimecanbespentmoreefficiently In long run, we see limited threat to central businessdistrictCapGoods Only9chargingstations(6chargerseach)requiredtocoverdailyelectricityReduction in new car sales could have material impact since auto capex forms c10% of sectorsalesRunningc

51、ostofanEVrobotaxiisathirdcheaperthanforanICEtaxiShifttowardsEVcouldcreateincrementaldemandforautomationplayersESGRunningcostofanEVrobotaxiisathirdcheaperthanforanICEtaxiTransition technologies for low-carbon mobility are developingrapidlyTripfarecouldbereducedby80%Ifjobswereputatrisk,orifaccesstoaff

52、ordable(public)transportbecamea“postcodelottery”, regulators would have reason to impose conditions ofoperationSizeoftaxifleetcouldshrinkbytwo-thirdsEither structure (concentrated ownership or distributed ownership of robotaxis) could bring benefitSource: UBS estimatesNote:ForUtilities,wecomparepowe

53、rpricesatthepumpwithawholesalelevelutilityrevenuepool,whichiswhytheactualrevenueupsidewilllikelybemoderately below 10%Figure 7: Impact of the mass introduction ofrobotaxisRobotaxi impactSales growthEBITmarginROICValuationOverallCommentAuto OEMset es to st frm seg a cr to eg a Auto suppliersStrong se

54、nsor volume growth ahead but steep price deflationTire makersBenefit through higher number of kilometers driven and better mixBatteriesEV penetration boosted due to cost advantage vs ICECar rentalRental companies appear natural partners for the platformsInternetPlatforms have ample flexibility/optio

55、nality in potential business modelsSemisRobotaxis could sharply boost the semi content (EV + AV)UtilitiesElectricity consumption boosted; investments requirement manageableTelecomSharp increase in data consumptionChemicalsFurther challenges for chemical companies with ICE exposureCapital GoodsEV wil

56、l have fewer moving parts and less bearings vs ICEInsuranceManageable impact, as traditional motor fleet premium size is limitedReal-EstateShift in location desirability and change-of-use could bring opportunitiesReal-EstateBrokersMarket-wide impact on real estate values would be de minimisSource: U

57、BS estimatesFigure 8: Stocks positively and negatively impacted by the theme Impacted Sector Positively Stock Negatively StockGMPSAAuto OEMsVWFCAHellaVeoneerAuto suppliers Valeo AptivLeoni Tire makersMichelin Pirelli BatteriesLG Chem SamsungSDI Car rentalSixtpref.BaiduInternetAlphabet Lyft Yandex In

58、fineonMelexisSemiconductorsNVIDIA TSMC EnelUtilitiesE.ON Iberdrola AT&TTelecomVerizon China Mobile AMT UmicoreJohnson MattheyLGChemicalsEMS-ChemieChemicalsAsahiKaseiBASFToray Sika Albemarle Capital GoodsSiemensSKF SandvikInsurance Zurich AXAHastings Real Estate BrokersCWKReal EstateRetail*Residentia

59、l*Industrial*Source: UBS estimates; Note: *Real Estate subsectors provided, given the breadth and fragmentation of the market (over 2000 listed real estate companies globally)What are key players saying about the launch of robotaxi commercial fleets?Lyft HYPERLINK /shared/d2I5Mf1x4w (LYFTInitiation)

60、istakingadual-prongedapproachtoAVdevelopment:both firstparty(1P)throughitsLevel5EngineeringCenterandthirdparty(3P)through itsopensourcedplatformandpartnershipswithAptiv,Waymoandothers.Lyfthas deployedafleetofrobotaxis(withasafetydriver)inLasVegassinceJanuary2018, and has performed over 35,000 trips