《高盛：2025人工智能與數據中心全球電力需求激增研究報告：五大驅動因素的上下行風險及可靠性投資順風（英文版）（28頁）.pdf》由會員分享，可在線閱讀，更多相關《高盛：2025人工智能與數據中心全球電力需求激增研究報告：五大驅動因素的上下行風險及可靠性投資順風（英文版）（28頁）.pdf（28頁珍藏版）》請在三個皮匠報告上搜索。

1、Goldman Sachs does and seeks to do business with companies covered in its research reports.As a result,investors should be aware that the firm may have a conflict of interest that could affect the objectivity of this report.Investors should consider this report as only a single factor in making thei

2、r investment decision.For Reg AC certification and other important disclosures,see the Disclosure Appendix,or go to employed by non-US affiliates are not registered/qualified as research analysts with FINRA in the U.S.The Goldman Sachs Group,Inc.AI/data centers global power surge Five drivers of ups

3、ide/downside and the Reliabilityinvestment tailwindEQUITYRESEARCH|January 12,2025|6:17PM ESTWe continue to see data center power demand adding the equivalent of a top 10 power-consuming country by the end of the decade.As data center and broader electricity demand growth accelerates,we see five pote

4、ntial constraints that could drive upside or downside risk to our updated base case of more than 160%global data center power demand growth in 2030 vs.2023 levels:Data center capacity additionsPower capacity/infrastructureGreen data centers:Low-carbonlead times/incentives/costData center power inten

5、sity efficiency gainsAI results/innovationsIn this report,we provide our latest analysis of expected growth in data center capacity,power infrastructureand Green Reliability Premiums,frameupside/downside implications from customer responses to chip efficiency gains and highlight key metrics to consi

6、der the ultimate breadth of AIs potential innovations that could advance Sustainable Development Goals.Reiterate Reliability as a multi-year investment theme We reiterate our bullish view on the multi-year theme of Reliability and resulting Green Capex driven by the confluence of:(a)power demand gro

7、wth acceleratingto levels not seen in the US and Europe in more than 20 years;(b)Big Techs long-term low-carbon prioritization/ability to support Green Reliability Premiums;and(c)Adaptation needs to mitigate risk of aging infrastructure,wider summer/wintertemperatures and extreme weather events.We r

8、emain bullish on stocks of companies in the data center power supply chain that will see investment tailwinds to advance Reliabilityof power and water.Carly Davenport+1(212)357- Goldman Sachs&Co.LLCAlberto Gandolfi+39(02)8022-Goldman Sachs Bank Europe SE-Milan branchBrian Singer,CFA+1(212)902-Goldma

9、n Sachs&Co.LLCBrendan Corbett+1(415)249- Goldman Sachs&Co.LLCJames Schneider,Ph.D.+1(212)357-Goldman Sachs&Co.LLCNoteNote:The following is a redacted version of the original report published January 12,2025 30 pgs.Contributing AuthorsJames Schneider,Ph.D.+1(212)357-Goldman Sachs&Co.LLCMichael Smith+

10、1(212)357-Goldman Sachs&Co.LLCBrian Singer,CFA+1(212)902-Goldman Sachs&Co.LLCCarly Davenport+1(212)357-Goldman Sachs&Co.LLCEric Sheridan+1(917)343-Goldman Sachs&Co.LLCKash Rangan+1(415)249-Goldman Sachs&Co.LLCXavier Zhang+852 2978-Goldman Sachs(Asia)L.L.C.Joshua M.Frantz,CFA+1(917)343-Goldman Sachs&

11、Co.LLCDerek R.Bingham+1(415)249-Goldman Sachs&Co.LLCBrendan Corbett+1(415)249-Goldman Sachs&Co.LLCJohn Miller+1(646)446-0292 Goldman Sachs&Co.LLCJaskaran Jaiya+1 332 245-Goldman Sachs India SPLTyler Bisset,CFA+1 212 357-Goldman Sachs&Co.LLCGrace Chen+44 20 7774-Goldman Sachs InternationalVarsha Venu

12、gopal+1(415)393-Goldman Sachs&Co.LLCMadeline Meyer+44(20)7774-Goldman Sachs InternationalEmma Jones+61(2)9320-Goldman Sachs Australia Pty LtdEvan Tylenda,CFA+44(20)7774-1153|Goldman Sachs InternationalToshiya Hari+1(646)446-Goldman Sachs&Co.LLCAlberto Gandolfi+39(02)8022-Goldman Sachs Bank EuropeSE-

13、Milan branchBrian Lee,CFA+1(917)343-Goldman Sachs&Co.LLCNeil Mehta+1(212)357-Goldman Sachs&Co.LLCAdam Wijaya+1(212)357-Goldman Sachs&Co.LLCJohn Mackay+1(212)357-Goldman Sachs&Co.LLCFive key drivers of upside/downside risk to our data center power demand outlook Will AI server shipments be constraine

14、d by data center capacity?Our analysis1.led by our Telecom I nfrastructure team suggests a tightening market for data centerreal estate in the coming years but suf ficient capacity for our base caseexpectations for power demand.Will data center capacity be constrained by power infrastructure?Our ana

15、lysis2.led by our Utilities team suggests a combination of new generation additions andgreater utilization of existing capacity will be suf ficient to meet data center powerdemand with transmission/interconnection the greatest risk.Will power infrastructure be constrained by low-carbon optionality/c

16、ost?We3.believe Big Tech will continue to take an all-in approach to data center powersourcing,with continued willingness to pay Green Reliability Premiums while at thesame time prioritizing time-to-market.Will new-gen AI chips drive lower or higher aggregate power demand?We4.assume Big Tech cash fl

17、ow/budgets will be the key constraint,leaving upside risk ifthere are no constraints and downside risk if compute speed demand is finite.Will AI server demand be constrained by AI results/innovations?This will remain5.key to watch,particularly from a Sustainability perspective whether we seeaccelera

18、ted ef ficiency solutions in the health care,energy,agriculture and educationsectors.Exhibit 1:After being flat for 2015-19,we have seen data center power demand accelerate in 2020-23 and expect a 160%increase through the rest of the decade Global data center electricity consumption,TWh;includes AI

19、and excludes cryptocurrency 02004006008001,0001,200Data center power demand,TWhUS ex-AIRoW ex-AIRoW AIUS AISour ce:Cisco,IEA,Goldman Sachs Global Invest ment Resear ch,Masanet et al.(2020)12 January 2025 2Goldman SachsGS SUSTAINAI/data center power demand key driver of Reliability as multi-year inve

20、stment theme We believe the confluence of rising power demand,historical underinvestment in infrastructure,AI investment and rising temperatures/more extreme weather events will continue to drive rising tailwinds for investments in Reliability,i.e.,products that help maintain resiliency,particularly

21、 for water and power.We continue to see opportunity for investment in stocks levered to the theme globally,which we believe will be a priority for both policymakers and corporate/residential consumers.Infrastructure replacement and hardening both necessitate Reliability investment.Our meetings with

22、corporates,regulators and policymakers in 2024 and at our 2025 Global Energy,Utilities and Clean Tech Conf erence indicated increased recognition of the need for grid/water infrastructure hardening and modernization.This is due to both underinvestment in recent years as well as a wider range of expe

23、cted temperatures between summer and winter.We believe both policymakers and regulators will look to reduce risk of outages and as such prioritize measures that would improve Reliability and Resiliency.This multi-stakeholder focus on Reliability combined with Big Techs support for low-carbon technol

24、ogies should provide tailwinds for exposed stocks at the confluence of Adaptation,Decarbonization and AI/Data Center Power Demand that also show strong financial fundamentals,in our view.Exhibit 2:Generational Growth:Our Utilities Research teams expect USA and EU-27 electricity consumption accelerat

25、ing through the end of the decade to levels not seen in 20+years 5-yr CAGR for US and European electricity demand;forecasts from our US and Europe Utilities Research teams from our April 2024 reportsUSAEU-27-2%-1%0%1%2%3%4%5%1990199419982002200620102014201820222026E2030EUS electricity generation 5-y

26、ear CAGR-2%-1%0%1%2%3%4%5%20052007200920112013201520172019202120232025E 2027E 2029EEU-27 electricity generation 5-year CAGR Sour ce:EIA,EMBER,Goldman Sachs Global Invest ment Resear ch12 January 2025 3Goldman SachsGS SUSTAINExhibit 3:The confluence of power demand acceleration with Adaptation to agi

27、ng infrastructure,extreme weather events and wider/higher temperature ranges is helping to drive signfiicant investment towards energy/water Reliability Where we see the confluence of various macro trends into Sustainability-related mega-themes we believe will drive investment tailwinds Sour ce:Gold

28、man Sachs Global Invest ment Resear ch12 January 2025 4Goldman SachsGS SUSTAINLatest outlook on data center power demand and emissions We believe data demand driven in part by AI and in part from deceleration in non-AI efficiency gains will catalyze generational growth in global power demand.Our ana

29、lysis suggests a 160%-165%increase in data center power demand by 2030 vs.2023 levels.I n the US,this implies that data centers will contribute a 1%CAGR to overall US power demand;our Utilities team in its April 2024 report expects overall US power demand CAGR of 2.4%through 2030.We see data centers

30、 adding a 0.3%CAGR to overall global power demand.Our base case implies data center power demand moves from 1%-2%of overall global power demand to 3%-4%by 2030.I n the US,the pace of mix increase is even greater,more than doubling by 2030 from 4%in 2023.I f global data center growth in 2030 vs.2023

31、levels were its own country,it would be a top 10 global power consumer.In our updated estimates,we reflect:Greater contribution from AI driven by a greater weighting on our server shipmentnsupply-based analysis.Narrower contribution ex-AI based on some of the additional demand we assumenfrom AI to c

32、annibalize a portion of what would otherwise be non-AI demand.Weassume about 12%annual growth in ex-AI workload by decade-end,deceleratingfrom a historical 20%-25%.Greater global weighting of power demand in the US based on recent reporting ofnUS historicals for data center power use by a study comm

33、issioned by the US DOE.I ntegration of our Telecom I nfrastructure teams outlook for data center capacitynbuildout.We look at the expected utilization of data center capacity af ter adjustingfor power usage ef f ectiveness(PUE)and our Telecom I nfrastructure teams outlookfor data center vacancy rate

34、s.Exhibit 4:As efficiency gains have decelerated,data demand growth is driving a surge in datacenter power use,with AI acceleration expected to continue Datacenter electricity consumption,TWh(LHS)and power efficiency gains ex-AI,%(RHS)0%2%4%6%8%10%12%14%16%18%20%22%24%01002003004005006007008009001,0

35、001,1001,200Power efficiency gains,%Data center power demand,TWhData center power demand,ex-AIAIEfficiencyGains DeceleratingNon-AIpower efficiency gainsPower Demand IncreasingSour ce:Masanet et al.(2020),IEA,Cisco,Goldman Sachs Global Invest ment Resear ch12 January 2025 5Goldman SachsGS SUSTAINEmis

36、sions outlook:We continue to assume about 40%of data center power growth is sourced by renewables,much of which we expect to come from power purchase agreements.We assume modest additional nuclear capacity online globally by the end of the decade to support data center demand,but we expect to see ac

37、celeration of added capacity in the 2030s.We continue to see rising carbon dioxide emissions from data centers by decade-end about 100%vs.2023 levels.Exhibit 5:We expect AI servers in operation will grow sharply through 2030 even as revenues from AI server shipments decelerate in 2027-30 AI servers

38、in operation implied by our global TMT team forecasts and implied annual AI server revenues Exhibit 6:We see AI power demand growing rapidly even as power use per AI server falls later in the decade due to mix shift and expected efficiencies AI power use,TWh(LHS);max power use per AI server,KW(RHS)$

39、0$50$100$150$200$250$300$350$40002,0004,0006,0008,00010,00012,00014,00016,000202220232024E2025E2026E2027E2028E2029E2030EAI server shipment revenues,$bnAI servers in operationLower power AI servers in operationHigh power AI servers in operationRevenues from AI server shipments0.01.02.03.04.05.0050100

40、150200250300350202220232024E2025E2026E2027E2028E2029E2030EAI power demandPower use per AI serverAI power use,TWhMax power use per AI server,KWSour ce:Goldman Sachs Global Invest ment Resear chSour ce:Goldman Sachs Global Invest ment Resear chExhibit 7:We see data center emissions doubling in 2030 vs

41、.2023 levels,net of impact of power purchase agreements(PPAs)from technology companies Carbon dioxide emissions in millions of tons(LHS);percent of 2022 energy emissions(RHS)0.0%0.3%0.6%0.9%1.2%1.5%1.8%2.1%0100200300400500600700Data center emissions covered by PPAsData center cumulative incremental

42、emissions%of 2022 global energy emissions(Million tons of CO2)(%of 2022 energyemissions)Sour ce:IEA,Goldman Sachs Global Invest ment Resear ch12 January 2025 6Goldman SachsGS SUSTAINKey learnings from the January 6-8 Global Energy,Utilities and Clean Tech Conference Bullish on appetite for data cent

43、er growth.Takeaways from our Global Energy,Utilities and Clean Tech Conf erence suggest pent-up interest from data center customers and conservatism by utilities in base-casing data center power demand growth into longer-term growth guidance.Utilities highlighted much greater magnitude of generation

44、 demand being discussed with data center customers than managements are willing to base case.There appears to be mutual interest among utilities,regulators and hyperscalers in ensuring rising data center demand does not drive reliability or af fordability issues to broader power customers.This is pa

45、rtly a driver of the conservative approach by utilities until contracts are signed that underpins investment towards further acceleration in power demand.Energy companies at the conference highlighted diversity of drivers of the power surge.Utilities continue to highlight that AI is not the sole dri

46、ver of power demand growth.Non-AI cloud-related demand continues to be strong,complemented by reshoring in the US and electrification globally.Our Utilities teams continue to expect Generational Growth in overall power demand 5-year CAGRs of US and European power demand not seen in 20+years.Sourcing

47、 power in near,medium and longer term:Existing gas capacity,renewables,natural gas,nuclear.Commentary at the Global Energy,Utilities and Clean Tech Conf erence suggested that sourcing new power capacity not already in queue is easiest for renewables(particularly solar),given long lead times for comb

48、ined natural gas turbines(2028+)and the very long lead time for nuclear generation.Utilities continued to highlight the relative cost competitiveness of renewables even as there is uncertainty on the sustainability of f ederal incentives.Overall,we expect greater weighting on flexing existing natura

49、l gas capacity and adding renewables(augmented with battery storage and natural gas peaking plants)in the near term,a pickup in natural gas combined cycle additions in the medium term and a rising focus on nuclear capacity additions for the 2030s.For additional details please also see our US Utiliti

50、es teams takeaways report.New survey of European utilities suggests European power demand set to inflect Following 15 years of declining power demand(consumption in Germany is equivalent to 1990,the year of reunification),the strong increase in connection requests(Data Centers,manufacturing,charging

51、 points,etc.)should lead to an inflection in power demand on a 2-3 year basis,bringing major tailwinds to power utilities.Our European Utilities team assumes 1%-2%annual increases in 2025-27,gradually rising to+3%by later in the decade.Our European Utilities team surveyed power utilities in our cove

52、rage,concluding that the number of connection requests received by power distribution operators(a leading indicator of future demand)has risen exponentially,mostly driven by Data Centers(DC).We estimate a DC pipeline of c.170 GW,equivalent to about 1/3 of European power consumption.Although only a p

53、ortion may be 12 January 2025 7Goldman SachsGS SUSTAINconverted,we estimate a potential 10%-15%boost to power demand over the coming 10-15 years.Exhibit 8:Our European Utilities teams recent survey suggests covered European Utilities have received more than 150 GW of connection requests from data ce

54、nters European data center connection requests split by geographic region(GW,%)Germany,26%Italy,13%Spain,13%UK,6%Others,42%DC Connection Requests Europe156 GWSour ce:Company dat a,Goldman Sachs Global Invest ment Resear chFor additional details please also see our European Utilities teams report,Dat

55、a Centers and AI:the time is now for Europe.12 January 2025 8Goldman SachsGS SUSTAINWill new-gen AI chips drive lower or higher aggregate power demand?Data center power demand was flattish in 2015-19 even as workload demand nearly tripled.This largely results from ef ficiency gains as:(a)data center

56、 workload has shif ted from higher energy-intensity traditional data centers to more ef ficient cloud and hyperscale data centers;and(b)cloud/hyperscale data centers have separately become more energy ef ficient,which we attribute in part to innovation and hyperscale/cloud consolidation.But starting

57、 in 2020,ef ficiency gains have significantly decelerated,which we attribute to more limited than prior period opportunities for mix shif t away from traditional centers and cloud/hyperscale consolidation.We continue to expect innovations that can drive efficiency gains that lower power intensity.We

58、 have already see new generations of AI servers that increase computing speed at a much greater pace than max power consumption,thus representing meaningful reductions in power intensity.However,as demand for AI training grows in the medium term and for inf erence longer term,we see demand growth we

59、ll exceeding the ef ficiency improvements that are leading to meaningful reductions in high power AI server power intensity.We assume non-AI server power efficiency gains which averaged around 15%annually in 2015-19 but decelerated to around low single digits in 2020-22 will remain relatively low.We

60、 do model a slight re-acceleration to an average of 3%in our base case in 2024-30 as industry discussions suggest continued ef forts towards innovation ef ficiencies,especially around the power intensity given the prospects for significant power needs ahead.We see greater potential for efficiency ga

61、ins from AI servers and assume 8%annualized gains for high-power servers through the end of the decade.We assume 5%-8%annual power ef ficiency gains per year to reflect expected future server Exhibit 9:Data center workload demand nearly tripled between 2015-2019 but electricity consumption from data

62、 centers was flat Data center workload demand(RHS)in million compute instances;data center power demand(LHS)in TWh Exhibit 10:Data center efficiency gains and the shift to cloud/hyperscale have been critical drivers of the moderate increase in data center power demand,but decelerating efficiency gai

63、ns have helped to drive a pickup in power demand from data centers in recent years Data center power intensity(LHS)in KWh per compute instance;share of cloud/hyperscale data centers(RHS)as%of workload 02505007501,0001,2501,500050100150200250300350400450201520162017201820192020202120222023Data center

64、 power sourced from others(LHS)Data center power sourced from renewables(LHS)Total data center workload demand(RHS)(TWh)(mn computeinstances)60%65%70%75%80%85%90%95%100%05001,0001,5002,0002,500201520162017201820192020202120222023Traditional data center power intensity(LHS)Cloud/hyperscale data cente

65、r power intensity(LHS)Overall data center power intensity(LHS)Cloud/hyperscale share of compute instances(RHS)(KWh/compute instance)(%share)Sour ce:Masanet et al.(2020),Cisco,IEA,Goldman Sachs Global Invest ment Resear chSour ce:Masanet et al.(2020),Cisco,IEA,Goldman Sachs Global Invest ment Resear

66、ch12 January 2025 9Goldman SachsGS SUSTAINinnovation relative to demand-based energy intensities where saw higher annual ef ficiencies early in lif e cycle,we assume slower pace in a supply-based approach to reflect timing of adoption.Will AI chip power intensity reduction lead to higher power consu

67、mption or lower power consumption?We have seen new AI innovations increase both computing speed and max power consumption per server but increase computing speed at a much greater pace,representing meaningful reduction in power intensity.As demand for GPUs grows,there is still notable company-projec

68、ted intensity reductions.As an example of how innovations have reduced power intensity per server but increased overall power per server:The NVI DI A DGX A100 system is listed to net 5 petaFLOPS and consuming 6.5 kWnmax,or 1.30 kW per pFLOPS.The more recent NVI DI A DGX H100 system is listed at 32 p

69、etaFLOPS andnconsuming 10.2 kW max,or 0.32 kW per pFLOPS.The new generation NVI DI A DGX B200 system using the new Blackwell chips isnlisted to net 72 petaFLOPS(training)and consuming 14.3 kW max,or 0.20 kW perpFLOPS.Our global TMT team expects AI server units shipped to grow at a 76%CAGR in 2024-26

70、;whether demand,budgets or neither is the constraint to growth is critical for energy use forecasting.A key question impacting compute demand is whether that demand is pent-up(i.e.,available new servers will be bought regardless of budget),not pent-up and constrained by demand itself,or constrained

71、by customer budgets.I n other words,will customers buy equal amounts of the more powerful servers as they would the less powerful ones?I f in a scenario in which a customer initially desires to buy 10 AI servers,and a new generation is announced with 10 x the compute speed for 5x the price,will the

72、customer:Buy the same number of servers(10),spend more money(5x)and substantially1.increase its compute speed as a result of the innovation(10 x)?This would representno constraints(i.e.,unlimited budgets,so demand dictated by available supply).Buy 1/10 the number of servers(1),maintaining the same l

73、evel of compute power2.as prior to the innovation and spending less money(0.5x)?This would representdemand as the constraint.Buy 1/5 the number of servers(2),thereby maintaining budget allocated for server3.purchases but getting a greater compute speed(2x)?This would represent budgetas the constrain

74、t.12 January 2025 10Goldman SachsGS SUSTAINConfidence in the above question is key to whether forecasting methodology should be weighted towards server-based(requiring nuance given varied and dynamic power consumption intensity and compute intensity)vs.demand based(forecasting compute power and powe

75、r intensity per unit of compute power).New generation AI servers consume more power and provide more compute speed,even as the power intensity has fallen meaningfully.There could be meaningful upside to our base case if appetite for purchase and utilization of servers is unconstrained.There could be

76、 downside to our base case if power efficiency is higher than expected or if power/compute speed efficiencies lead to fewer servers purchased than expected.We note corporate optimism that AI innovations can drive significant efficiencies to ultimately mitigate power demand and emissions growth.Overa

77、ll,it does not appear from our recent Exhibit 11:We have seen new AI innovations increase max power consumption per server but increase computing speed per server by an even greater level,representing a meaningful reduction in power intensity Recent evolution of NVIDIA server system specifications i

78、s indicative of increasing max power per server but with lower power intensity relative to computing speed(for training)Max power required(kW)Compute speed(petaFLOPS)56.5(1.30 kW/pFLOPS)10.2(0.32 kW/pFLOPS)3214.3(0.20 kW/pFLOPS)72NVIDIA DGX B2002x the power15x the compute speedNVIDIA DGX H1001.5x th

79、e power7x the compute speedNVIDIA DGX A1001 petaFLOPS=One quadrillion(one thousand trillion)operations per second Sour ce:NVIDIA,Goldman Sachs Global Invest ment Resear chExhibit 12:Extent of pent-up demand for AI server supply and voraciousness of technology capex budgets will be critical for pace

80、of AI power consumption Indicative scenario analysis of how demand vs.budget constraints could impact AI compute speed and power use Assumes power gener at ion,t r ansmission and int er connect ion ar e not a const r aint for indicat ive pur poses Sour ce:Goldman Sachs Global Invest ment Resear ch12

81、 January 2025 11Goldman SachsGS SUSTAINcorporate dialogues that efficiency gains are leading to a reduced need for data center infrastructure and rack space.Rather,we think the risk is to the unconstrained side until there is greater clarity on the impact/demand of AI solutions.12 January 2025 12Gol

82、dman SachsGS SUSTAINWill AI server shipments be constrained by data center capacity?New analysis from our Telecom Infrastructure team implies a tight market for data centers(DC)through 2026,then potential for modest loosening thereafter.Part of this growth will be fueled by a substantial shif t in w

83、orkload mix due to the rapid expansion of Generative AI,which our Telecom I nfrastructure team expects to grow from 9%of workloads(7.7 GW)in 1Q25 to 31%(35 GW)by 4Q2030.Overall DC workload demand is estimated to be 110 GW by 2030.Nonetheless,our Telecom I nfrastructure team expects DC supply will me

84、et these needs and that by the end of 2030 there will be 122 GW of capacity online,implying a 13%6-year CAGR.DC supply mix will skew even further to cloud workloads in 2030(70%vs 60%today)given then ongoing shif t to higher compute workloads.The sustainability of AI workload demand growth and cannib

85、alization of legacy DC workloads remain some of the largest variables impacting overall data center power demand estimates.Our Telecom I nfrastructure team estimates these factors can drive as much as 25 GW of demand variance vs.their base case,even by the end of 2027.The DC operator landscape outlo

86、ok appears more clear as our Telecom I nfrastructure team has highlighted that seven of the top ten companies with the most operational UPS power are also in the top ten companies with the most planned UPS power coming online over the next five years.I n ef f ect,the companies which currently lead t

87、he datacenter market are already invested heavily in maintaining their lead.For more details,please see our Telecom Infrastructure teams January 2025 report,We see a tightening market through 2026.Overall,this suggests that data center capacity growth we expect can accommodate our base case for powe

88、r demand outlook.We incorporate power usage ef f ectiveness(PUE)estimates based on US DOE historical data/outlook and apply our Telecom I nfrastructure teams views on data center vacancy rates.Based on this,our power demand estimates imply a rising utilization of what we expect will be contracted ca

89、pacity for data centers to around 90%-95%by decade end globally.Exhibit 13:AI will see the largest share of of DC demand growth over the next two years GS US DC baseline demand forecast-hyperscale+cloud,traditional,AI Exhibit 14:Cloud compute will continue to gain share of overall DC workload capaci

90、ty Forecast datacenter supply by workload type,2024-30 010,00020,00030,00040,00050,00060,00070,00080,000Hyperscale+CloudNon-Hyperscale/CloudAI0%10%20%30%40%50%60%70%80%010,00020,00030,00040,00050,00060,00070,00080,00090,000100,0004Q24E2Q25E4Q25E2Q26E4Q26E2Q27E4Q27ETotal-Hyperscale+CloudTotal-OtherHy

91、perscale+Cloud(%Total)Other(%Total)Sour ce:451,Goldman Sachs Global Invest ment Resear chSour ce:Goldman Sachs Global Invest ment Resear ch12 January 2025 13Goldman SachsGS SUSTAINExhibit 15:Our power demand base case implies rising utilization of contracted data center capacity based on our Telecom

92、 I nfrastructure teams outlook for data center capacity growth and vacancy rates Implied capacity utilization and occupancy rates of global contracted data center capacity 60%65%70%75%80%85%90%95%100%2020202120222023E 2024E 2025E 2026E 2027E 2028E 2029E 2030E%utilization%occupancySour ce:451,US DOE,

93、Goldman Sachs Global Invest ment Resear ch12 January 2025 14Goldman SachsGS SUSTAINWill data center capacity be constrained by power infrastructure?We continue to expect an all-in approach in sourcing power demand.Our conversations with renewable developers indicated that wind and solar could serve

94、roughly 80%of a data centers power demand if paired with storage,but some sort of baseload generation is needed to meet the 24/7 demand.Nuclear is the pref erred option to step in as the baseload generation,but given both the scarcity of existing unregulated nuclear plants and the dif ficulty of bui

95、lding a new one,we view natural gas as the more realistic option for most data centers.This creates a potential bottleneck for data centers,with many utilities estimating that you would have to wait until 2029 to receive a large gas turbine if you entered the queue now,shif ting the focus to smaller

96、 scale natural gas peakers.Our US Utilities team sees almost 50 GW of incremental US generation capacity likely through 2030.I n our original note discussing the rising power demand from data centers,we laid out our framework for identifying how much of the power demand from data centers needs to be

97、 met with incremental capacity vs existing capacity.Our methodology remains the same,with the conclusion that roughly 50%of data center demand needing to be met with new capacity,and that the generation capacity that is added will likely be 40%natural gas CCGTs,20%natural gas peakers,25%solar,and 15

98、%wind.Applying this to our demand estimates,it implies roughly 47 GWs of incremental generation capacity through 2030.In the US,we expect more than$700 bn of grid investment through 2030 in the US.The grid will also require significant investment,not solely related to data centers but the rise in po

99、wer demand does further accelerate the need for grid upgrades and grid expansion.Our US Utilities teams estimates imply roughly$720 bn of grid spend through 2030,with the bulk of the spend focused on distribution.However,we expect transmission spend to grow at a slightly faster rate than distributio

100、n as a significant Exhibit 16:We forecast roughly 47 GWs of incremental capacity related to data center power demand,with the bulk being natural gas capacity Overall net capacity additions through 2030 by source,MW Sour ce:Goldman Sachs Global Invest ment Resear ch12 January 2025 15Goldman SachsGS S

101、USTAINamount of high voltage transmission is needed to supply regions where data centers are being built out with suf ficient power to distribute.These transmission projects can take several years to permit,and then several more to build out,creating another potential bottleneck for data center grow

102、th if the regions are not proactive about this given the lead time.Exhibit 17:Our US Utilities teams forecasts imply$720 bn of grid capex through 2030,with transmission growing at a slightly faster rate than distribution Grid capex,$bn Sour ce:Company dat a,EEI,Goldman Sachs Global Invest ment Resea

103、r ch12 January 2025 16Goldman SachsGS SUSTAINWill power infrastructure be constrained by low-carbon optionality/cost?While we do not see a Green Premium for intermittent power in the US,our analysis suggests a Green Reliability Premium for low-carbon round-the-clock power solutions vs natural gas co

104、mbined cycle in the US.We expect the Green Reliability Premium to be greater in the US vs.other developed markets in part due to lower-cost natural gas in the US,even before taking into account country-level carbon prices in some other markets.Exhibit 18:We see a Green Reliability Premium to source

105、round-the-clock low-carbon solutions,even as intermittent solar/wind have lower levelized energy cost bined cycle natural gas in the US;renewables and nuclear options would see higher levelized costs of energy if I RA incentives are diluted/ended Levelized cost of energy of various fuel&technology c

106、ombinations to power new data centers,inclusive of assumed transmission and distribution(except intermittent solar/wind);call out boxes show LCOE under current IRA incentives;bracketed ranges reflect incremental LCOE in scenarios of pre-IRA/no-IRA incentives$25$26$52$71$75$77$84$86$94$95$107$125$3-$

107、16$6-$17$6-$24$6-$25$9$6-$25$3-$21$6-$24$9$9$9$0$20$40$60$80$100$120$140Wind(intermittent)Solar(intermittent)Gas CCGTSolar,Storage,NGPeaker orGrid100%Renewables(Solar,Wind&Storage)NuclearLarge-Scale(Onsite)100%Renewables(OffsiteSolar,Wind&Storage)Wind,Storage,NGPeaker orGridOffsite Solar,Storage,NGP

108、eaker orGridNuclear SMR(Onsite)NuclearLarge-Scale(Offsite)Nuclear SMR(Offsite)Levelized cost of energy(LCOE);$per MWhGreen Reliability Premium-No IncentiveGreen Reliability Premium-Pre-IRAGreen Reliability Premium-IRA IncentiveCCGT Benchmark*Wind and Solar (Init er mit t ent)assume no t r ansmission

109、;Wind is assumed t o be offsit e,solar/gas/nuclear is assumed onsit e unless not ed;100%r enewables assumes 5%power sour ced fr om nat ur al gas peaker s or gr id due t o day-t o-day int er mit t ency var iabilit y Sour ce:Goldman Sachs Global Invest ment Resear ch12 January 2025 17Goldman SachsGS S

110、USTAINWe expect hyperscalers will remain committed to pursuing low-carbon power solutions based on our analysis of Green Reliability Premiums,industry discussions and recent contracts.Our analysis updated for our latest data center power demand outlook and analyst expectations for hyperscaler EBI TD

111、A suggests that the capital requirements for Green Reliability Premiums to source data center power demand to be modest relative to the EBI TDA(2%-3%)and corporate returns of key hyperscalers(1 percentage point impact vs.32%average baseline estimates;median across all sectors is about 12%-13%).Exhib

112、it 19:Relative tradeoffs of various technologies which can all provide capacity for new data center-driven power demand Sour ce:Goldman Sachs Global Invest ment Resear ch12 January 2025 18Goldman SachsGS SUSTAINExhibit 20:The Green Reliability Premium in the US,conservatively applied to global data

113、center power demand growth in 2030 vs.2023,is modest in comparison to hyperscalers EBI TDA and corporate returns US Green Reliability Premium applied to global data center power demand growth in 2030 vs.2023 relative to 2026E and 2030E EBITDA and 2026E cash return on cash invested AverageBaseline le

114、velized cost of energy,US,$/MWh$52Green Reliable levelized cost of energy average,US,$/MWh$91Average US Green Reliability Premium,$/MWh$39Global data center power demand(TWh)20234112030E1,076Increase,2030E vs.2023665Required 2030E industry outlay if applying US Green Reliability Premium,$bn$26Hypers

115、caler total company EBITDA,2026E,$bn$764%of hyperscaler 2026E EBITDA3%Hyperscaler 2030E EBITDA(at 10%CAGR vs.2026E)$1,119%of hyperscaler 2030E EBITDA2%Hyperscaler average CROCI,2026E32%CROCI impact from industry 2030E Green Reliability Premium-1%Adj.CROCI if hyperscalers cover industry Green Reliabi

116、lity Premium31%Baseline sour ce is combined cycle nat ur al gas;Gr een Reliable sour ces ar e lar ge nuclear,SMR nuclear,solar +st or age+gas peaker,solar +offsit e wind+st or age.Please see t he sect ion lat er in t he r epor t for mor e det ails on our global dat a cent er power demand for ecast s

117、.Sour ce:Goldman Sachs Global Invest ment Resear chWe are in the early stages of nuclear renaissance in US and globally.Recent contracts for small modular reactors(SMRs)and larger-scale nuclear to source data center power demand growth,combined with increased country-level embrace of nuclear power a

118、ppears poised to drive a significant rampup of investment in the next 5 years and power in the 2030s.Lowering the capital costs of SMRs and accommodating nuclear expansion while minimizing impact to reliability/pricing elsewhere in the grid will be key for long-term competitiveness,in our view.Confi

119、dence in enriched uranium supply is a key challenge.Our Energy equity research team highlighted the three processes needed to supply nuclear plant feedstock:Uranium production,Uranium Conversion and Uranium Enrichment.Each is not done in the same place so could also require transportation.As shown i

120、n the below exhibit,Europe and the Americas source about 30%of uranium production,hold 56%of uranium conversion capacity,and hold 41%of uranium enrichment capacity.The key participants outside these countries are China/Russia/Kazakhstan,together representing about 46%of uranium supply,44%of uranium

121、conversion capacity and 59%of uranium enrichment capacity.12 January 2025 19Goldman SachsGS SUSTAINWe expect Big Techs all-in approach to low-carbon technology deployment will continue,supportive of upside for Green Capex.Our analysis suggests less variability in levelized cost of energy among low-c

122、arbon power solutions such as large-scale nuclear and solar/wind/energy storage.We also expect continued hyperscaler support for carbon capture and carbon removal.We expect natural gas-fired power use by data centers to rise with 60%of data center power demand from thermal sources(largely gas).Polic

123、y,technology and equipment lead times will likely help guide the split between combined cycle and peaker unit deployment.Exhibit 21:As momentum builds for nuclear capacity expansion,we expect greater focus on geographical exposure to Uranium not just production but conversion and enrichment as well

124、Kazakhstan38%Canada20%Namibia13%Australia9%Uzbekistan7%Russia5%China3%Niger2%Other3%France24%China24%Canada20%Russia20%US12%East59%West41%2023Uranium Production by Country2022Conversion Capacity by Country2022Enrichment Capacity by RegionSour ce:Wor ld Nuclear Associat ion,UxC,Goldman Sachs Global I

125、nvest ment Resear ch12 January 2025 20Goldman SachsGS SUSTAINWill AI server demand be constrained by AI results?As Artificial Intelligence investment increases,we expect the investment community will look to measure the impact of AI solutions.We see interest from Sustainable investors in understandi

126、ng how AI can accelerate progress across a range of sustainability objectives,comparing corporate value creation relative to investment and social value of potential AI solutions relative to the social cost of emissions.I n a previous report,AIs Sustainable Solutions,frames AI-related opportunities

127、across key sectors,including Human Capital,Healthcare,Education,Agriculture,and Climate(Exhibit 22).We identify SDGs in which we see the greatest potential for AI-impact and propose metrics that may help investors assess the value of AI-benefits relative to the negative value caused by higher emissi

128、ons and social risks.Healthcare:Accelerating discovery and care(SDG 3).Massive amounts ofnhealthcare data are produced every day from a range of diverse sources,providing arich opportunity for the application of AI-based technologies to drive innovationacross drug discovery&design,clinical trials,he

129、althcare analytics,tools&diagnostics,and personalized care.Metrics:Value for new drug/vaccine/therapeutic products linked to AIoacceleration,value for ef ficiency gains for swif ter drug development timelineto market,value for ef ficiencyExhibit 22:We see AI accelerating progress across a range of S

130、ustainable Development Goals(SDGs)AIs Sustainability opportunities:Where to watch,SDG crossover,and how to measureHealthcare:Accelerating discovery and care(SDG 3)Metrics:Value for new drug/vaccine/therapeutic products linked to AI acceleration,value for efficiency gains for swifter drug development

131、 timeline to market,value for efficiencyAgriculture:Improving yields and reducing waste(SDG 2).Metrics:Value of improved crop yield,value of reduced resource usage(water,fertilizer)Climate Solutions:Optimization and efficiency in power generation and physical assets(SDG 7,9).Metrics:Value of linked

132、power generation/utilization efficiency,value/level of reduction in emissions and emissions intensityHuman Capital:The opportunity and need for reskilling and upskilling(SDG 8).Metrics:Economic productivity,value of employees re-skilled/re-purposed for different roles,value of certifications earnedE

133、ducation:A step change in interactivity and personalization(SDG 4).Metrics:Value of linked improvement to student test scores,value of linked enablement of certifications/degrees earnedSour ce:Goldman Sachs Global Invest ment Resear ch12 January 2025 21Goldman SachsGS SUSTAINAgriculture:Improving yi

134、elds and reducing waste(SDG 2).AI-enablednapplications stand to improve agricultural outcomes all the way from enhancedinsights into what to plant and when to improve logistics that reduce time-to-marketfor perishable goods.Precision agriculture data driven approaches to farming(crops&livestock),is

135、the broadest category where leveraging AI can help toimprove yields and reduce waste.However,beyond the myriad of AI applicationsthat fall under the precision ag umbrella,AI also stands to help improve agriculturalsupply chain saf ety,speed and transparency.Metrics:Value of improved crop yield,value

136、 of reduced resource usage(water,of ertilizer)Climate Solutions:Optimization and efficiency in power generation andnphysical assets(SDG 7,9).AI holds promise for being a contributor in mitigatingthe impact of climate change in a wide range of applications,including renewableenergy optimization(e.g.,

137、weather forecasting,operational scheduling,batterystorage optimization);physical asset and power use optimization(e.g.,dynamicheating/cooling,data center ef ficiency,manufacturing ef ficiencies);and climatemodeling.Metrics:Value of linked power generation/utilization ef ficiency,value/level oforeduc

138、tion in emissions and emissions intensityHuman Capital:The opportunity and need for reskilling and upskilling(SDG 8).nI ncreasing adoption of advanced technologies such as AI create a growing need foreducational platforms that enable reskilling and upskilling.We also anticipategreater demand for hum

139、an capital management tools that can help identify andmatch rapidly shif ting worker competencies with organizational needs(skillsmatching).Metrics:Economic productivity,value of employees re-skilled/re-purposed forodif f erent roles,value of certifications earnedEducation:A step change in interacti

140、vity and personalization(SDG 4).AI isncomplementing e-learning ef f ectiveness through leaps forward in personalizationand interactivity.Sof tware-based e-learning platforms can leverage AI to createcost-ef f ective custom study sequences to fit an individuals needs,track progress,and enable self-di

141、rected learning,while AI-powered chatbots are making the futureof interactivity a reality.Education is especially relevant in periods of technologicalacceleration in which workers must upskill,and job seekers must gain new skills.Metrics:Value of linked improvement to student test scores,value of li

142、nkednenablement of certifications/degrees earnedKey risks include both how AI models work and how theyre likely to be used.The first order of concerns related to AI center on how the models work,including informational accuracy,transparency,potential algorithmic bias,and intellectual property protec

143、tions.A next order of risk encompasses the weaponization of AI models,and includes privacy,academic integrity,disinformation,fraud and cyber attacks.I n each of these cases,AI enables potential threats to be propagated at new levels of speed,scale and customization.Potential impact on jobs is perhap

144、s a less immediate,but 12 January 2025 22Goldman SachsGS SUSTAINoverarching concern.Our economists estimate that roughly two-thirds of current jobs are exposed to some degree of AI automation,and that generative AI could substitute up to one-fourth of current work.The highest order of potential impa

145、cts includes more extreme tail risks of societal destabilization and disaster scenarios,such as the manipulation of elections,political repression,or automated warfare(Exhibit 23).What can be done to mitigate key risks?Creators of AI models will be pressed for transparency and accountability.This wo

146、uld include mechanisms to alert users to AI-generated content and authorship(watermarking),warning labels,and disclosures of how the systems are trained and how they work.Voluntary commitments from creators are already being made including the enablement of independent audits and third-party testing

147、,while corporate users are putting into place frameworks for ethical use of AI.Regulations will also play a central role in mitigating AI risks.With the EU AI Act,Europe was the first to pass a comprehensive AI-related legal framework.The AI Act proposes a risk-based framework that classifies AI sys

148、tems in four tiers of risk and lists prohibited technologies,disclosures and obligations of AI systems operators,and penalties for non-compliance.The US,UK,and China are among other nations that have proposed their own AI-focused rules and regulations.Exhibit 23:We frame four principal tiers of risk

149、s,ordered by time horizon and potential impact,and mitigants for each AI-related risks and mitigants Sour ce:Goldman Sachs Global Invest ment Resear ch12 January 2025 23Goldman SachsGS SUSTAINDisclosure Appendix Reg AC We,Brian Singer,CFA,Brendan Corbett,Carly Davenport,Alberto Gandolfi,James Schnei

150、der,Ph.D.,Brian Lee,CFA,Neil Mehta,Adam Wijaya,John Mackay,Eric Sheridan,Kash Rangan,Toshiya Hari,Derek R.Bingham,Evan Tylenda,CFA,Emma Jones,Madeline Meyer,Varsha Venugopal,Grace Chen,Xavier Zhang,Joshua M.Frantz,CFA,Michael Smith,John Miller,Tyler Bisset,CFA and Jaskaran Jaiya,hereby certify that

151、all of the views expressed in this report accurately reflect our personal views about the subject company or companies and its or their securities.We also certify that no part of our compensation was,is or will be,directly or indirectly,related to the specific recommendations or views expressed in t

152、his report.Unless otherwise stated,the individuals listed on the cover page of this report are analysts in Goldman Sachs Global I nvestment Research division.GS Factor Profile The Goldman Sachs Factor Profile provides investment context for a stock by comparing key attributes to the market(i.e.our c

153、overage universe)and its sector peers.The four key attributes depicted are:Growth,Financial Returns,Multiple(e.g.valuation)and I ntegrated(a composite of Growth,Financial Returns and Multiple).Growth,Financial Returns and Multiple are calculated by using normalized ranks for specific metrics for eac

154、h stock.The normalized ranks for the metrics are then averaged and converted into percentiles for the relevant attribute.The precise calculation of each metric may vary depending on the fiscal year,industry and region,but the standard approach is as follows:Growth is based on a stocks forward-lookin

155、g sales growth,EBI TDA growth and EPS growth(for financial stocks,only EPS and sales growth),with a higher percentile indicating a higher growth company.Financial Returns is based on a stocks forward-looking ROE,ROCE and CROCI (for financial stocks,only ROE),with a higher percentile indicating a com

156、pany with higher financial returns.Multiple is based on a stocks forward-looking P/E,P/B,price/dividend(P/D),EV/EBI TDA,EV/FCF and EV/Debt Adjusted Cash Flow(DACF)(for financial stocks,only P/E,P/B and P/D),with a higher percentile indicating a stock trading at a higher multiple.The Integrated perce

157、ntile is calculated as the average of the Growth percentile,Financial Returns percentile and(100%-Multiple percentile).Financial Returns and Multiple use the Goldman Sachs analyst forecasts at the fiscal year-end at least three quarters in the future.Growth uses inputs for the fiscal year at least s

158、even quarters in the future compared with the year at least three quarters in the future(on a per-share basis for all metrics).For a more detailed description of how we calculate the GS Factor Profile,please contact your GS representative.M&A Rank Across our global coverage,we examine stocks using a

159、n M&A framework,considering both qualitative factors and quantitative factors(which may vary across sectors and regions)to incorporate the potential that certain companies could be acquired.We then assign a M&A rank as a means of scoring companies under our rated coverage from 1 to 3,with 1 represen

160、ting high(30%-50%)probability of the company becoming an acquisition target,2 representing medium(15%-30%)probability and 3 representing low(0%-15%)probability.For companies ranked 1 or 2,in line with our standard departmental guidelines we incorporate an M&A component into our target price.M&A rank

161、 of 3 is considered immaterial and therefore does not factor into our price target,and may or may not be discussed in research.Quantum Quantum is Goldman Sachs proprietary database providing access to detailed financial statement histories,forecasts and ratios.I t can be used for in-depth analysis o

162、f a single company,or to make comparisons between companies in dif f erent sectors and markets.Disclosures Distribution of ratings/investment banking relationships Goldman Sachs I nvestment Research global Equity coverage universe As of January 1,2025,Goldman Sachs Global I nvestment Research had in

163、vestment ratings on 3,021 equity securities.Goldman Sachs assigns stocks as Buys and Sells on various regional I nvestment Lists;stocks not so assigned are deemed Neutral.Such assignments equate to Buy,Hold and Sell for the purposes of the above disclosure required by the FI NRA Rules.See Ratings,Co

164、verage universe and related definitions below.The I nvestment Banking Relationships chart reflects the percentage of subject companies within each rating category for whom Goldman Sachs has provided investment banking services within the previous twelve months.Regulatory disclosures Disclosures requ

165、ired by United States laws and regulations See company-specific regulatory disclosures above for any of the following disclosures required as to companies ref erred to in this report:manager or co-manager in a pending transaction;1%or other ownership;compensation for certain services;types of client

166、 relationships;managed/co-managed public of f erings in prior periods;directorships;for equity securities,market making and/or specialist role.Goldman Sachs trades or may trade as a principal in debt securities(or in related derivatives)of issuers discussed in this report.The following are additiona

167、l required disclosures:Ownership and material conflicts of interest:Goldman Sachs policy prohibits its analysts,prof essionals reporting to analysts and members of their households from owning securities of any company in the analysts area of coverage.Analyst compensation:Analysts are paid in part b

168、ased on the profitability of Goldman Sachs,which includes investment banking revenues.Analyst as officer or director:Goldman Sachs policy generally prohibits its analysts,persons reporting to analysts or members of their households from serving as an of ficer,director or advisor of any company in th

169、e analysts area of coverage.Non-U.S.Analysts:Non-U.S.analysts may not be associated persons of Goldman Sachs&Co.LLC and therefore may not be subject to FI NRA Rule 2241 or FI NRA Rule 2242 restrictions on communications with subject company,public appearances and trading securities held by the analy

170、sts.Distribution of ratings:See the distribution of ratings disclosure above.Price chart:See the price chart,with changes of ratings and price targets in prior periods,above,or,if electronic format or if with respect to multiple companies which are the subject of this report,on the Goldman Sachs web

171、site at https:/ DistributionInvestment Banking RelationshipsBuyHoldSellBuyHoldSellGlobal48%34%18%64%57%43%12 January 2025 24Goldman SachsGS SUSTAINAdditional disclosures required under the laws and regulations of jurisdictions other than the United States The following disclosures are those required

172、 by the jurisdiction indicated,except to the extent already made above pursuant to United States laws and regulations.Australia:Goldman Sachs Australia Pty Ltd and its af filiates are not authorised deposit-taking institutions(as that term is defined in the Banking Act 1959(Cth)in Australia and do n

173、ot provide banking services,nor carry on a banking business,in Australia.This research,and any access to it,is intended only for“wholesale clients”within the meaning of the Australian Corporations Act,unless otherwise agreed by Goldman Sachs.I n producing research reports,members of Global I nvestme

174、nt Research of Goldman Sachs Australia may attend site visits and other meetings hosted by the companies and other entities which are the subject of its research reports.I n some instances the costs of such site visits or meetings may be met in part or in whole by the issuers concerned if Goldman Sa

175、chs Australia considers it is appropriate and reasonable in the specific circumstances relating to the site visit or meeting.To the extent that the contents of this document contains any financial product advice,it is general advice only and has been prepared by Goldman Sachs without taking into acc

176、ount a clients objectives,financial situation or needs.A client should,before acting on any such advice,consider the appropriateness of the advice having regard to the clients own objectives,financial situation and needs.A copy of certain Goldman Sachs Australia and New Zealand disclosure of interes

177、ts and a copy of Goldman Sachs Australian Sell-Side Research I ndependence Policy Statement are available at:https:/ information in relation to CVM Resolution n.20 is available at https:/ applicable,the Brazil-registered analyst primarily responsible for the content of this research report,as define

178、d in Article 20 of CVM Resolution n.20,is the first author named at the beginning of this report,unless indicated otherwise at the end of the text.Canada:This information is being provided to you for information purposes only and is not,and under no circumstances should be construed as,an advertisem

179、ent,of f ering or solicitation by Goldman Sachs&Co.LLC for purchasers of securities in Canada to trade in any Canadian security.Goldman Sachs&Co.LLC is not registered as a dealer in any jurisdiction in Canada under applicable Canadian securities laws and generally is not permitted to trade in Canadi

180、an securities and may be prohibited from selling certain securities and products in certain jurisdictions in Canada.I f you wish to trade in any Canadian securities or other products in Canada please contact Goldman Sachs Canada I nc.,an af filiate of The Goldman Sachs Group I nc.,or another registe

181、red Canadian dealer.Hong Kong:Further information on the securities of covered companies ref erred to in this research may be obtained on request from Goldman Sachs(Asia)L.L.C.India:Further information on the subject company or companies ref erred to in this research may be obtained from Goldman Sac

182、hs(I ndia)Securities Private Limited,Research Analyst-SEBI Registration Number I NH000001493,951-A,Rational House,Appasaheb Marathe Marg,Prabhadevi,Mumbai 400 025,I ndia,Corporate I dentity Number U74140MH2006FTC160634,Phone+91 22 6616 9000,Fax+91 22 6616 9001.Goldman Sachs may beneficially own 1%or

183、 more of the securities(as such term is defined in clause 2(h)the I ndian Securities Contracts(Regulation)Act,1956)of the subject company or companies ref erred to in this research report.I nvestment in securities market are subject to market risks.Read all the related documents carefully before inv

184、esting.Registration granted by SEBI and certification from NI SM in no way guarantee performance of the intermediary or provide any assurance of returns to investors.Goldman Sachs(I ndia)Securities Private Limited I nvestor Grievance E-mail:india-client-.Compliance Of ficer:Anil Rajput|Tel:+91 22 66

185、16 9000|Email:.Japan:See below.Korea:This research,and any access to it,is intended only for“prof essional investors”within the meaning of the Financial Services and Capital Markets Act,unless otherwise agreed by Goldman Sachs.Further information on the subject company or companies ref erred to in t

186、his research may be obtained from Goldman Sachs(Asia)L.L.C.,Seoul Branch.New Zealand:Goldman Sachs New Zealand Limited and its af filiates are neither“registered banks”nor“deposit takers”(as defined in the Reserve Bank of New Zealand Act 1989)in New Zealand.This research,and any access to it,is inte

187、nded for“wholesale clients”(as defined in the Financial Advisers Act 2008)unless otherwise agreed by Goldman Sachs.A copy of certain Goldman Sachs Australia and New Zealand disclosure of interests is available at:https:/ reports distributed in the Russian Federation are not advertising as defined in

188、 the Russian legislation,but are information and analysis not having product promotion as their main purpose and do not provide appraisal within the meaning of the Russian legislation on appraisal activity.Research reports do not constitute a personalized investment recommendation as defined in Russ

189、ian laws and regulations,are not addressed to a specific client,and are prepared without analyzing the financial circumstances,investment profiles or risk profiles of clients.Goldman Sachs assumes no responsibility for any investment decisions that may be taken by a client or any other person based

190、on this research report.Singapore:Goldman Sachs(Singapore)Pte.(Company Number:198602165W),which is regulated by the Monetary Authority of Singapore,accepts legal responsibility for this research,and should be contacted with respect to any matters arising from,or in connection with,this research.Taiw

191、an:This material is for ref erence only and must not be reprinted without permission.I nvestors should carefully consider their own investment risk.I nvestment results are the responsibility of the individual investor.United Kingdom:Persons who would be categorized as retail clients in the United Ki

192、ngdom,as such term is defined in the rules of the Financial Conduct Authority,should read this research in conjunction with prior Goldman Sachs research on the covered companies ref erred to herein and should ref er to the risk warnings that have been sent to them by Goldman Sachs I nternational.A c

193、opy of these risks warnings,and a glossary of certain financial terms used in this report,are available from Goldman Sachs I nternational on request.European Union and United Kingdom:Disclosure information in relation to Article 6(2)of the European Commission Delegated Regulation(EU)(2016/958)supple

194、menting Regulation(EU)No 596/2014 of the European Parliament and of the Council(including as that Delegated Regulation is implemented into United Kingdom domestic law and regulation following the United Kingdoms departure from the European Union and the European Economic Area)with regard to regulato

195、ry technical standards for the technical arrangements for objective presentation of investment recommendations or other information recommending or suggesting an investment strategy and for disclosure of particular interests or indications of conflicts of interest is available at https:/ which state

196、s the European Policy for Managing Conflicts of I nterest in Connection with I nvestment Research.Japan:Goldman Sachs Japan Co.,Ltd.is a Financial I nstrument Dealer registered with the Kanto Financial Bureau under registration number Kinsho 69,and a member of Japan Securities Dealers Association,Fi

197、nancial Futures Association of Japan Type I I Financial I nstruments Firms Association,The I nvestment Trusts Association,Japan,and Japan I nvestment Advisers Association.Sales and purchase of equities are subject to commission pre-determined with clients plus consumption tax.See company-specific di

198、sclosures as to any applicable disclosures required by Japanese stock exchanges,the Japanese Securities Dealers Association or the Japanese Securities Finance Company.Ratings,coverage universe and related definitions Buy(B),Neutral(N),Sell(S)Analysts recommend stocks as Buys or Sells for inclusion o

199、n various regional I nvestment Lists.Being assigned a Buy or Sell on an I nvestment List is determined by a stocks total return potential relative to its coverage universe.Any stock not assigned as a Buy or a Sell on an I nvestment List with an active rating(i.e.,a stock that is not Rating Suspended

200、,Not Rated,Coverage Suspended or Not Covered),is deemed Neutral.Each region manages Regional Conviction lists,which are selected from Buy rated stocks on the respective regions I nvestment lists and represent investment recommendations focused on the size of the total return potential and/or the lik

201、elihood of the realization of the return across their respective areas of coverage.The addition or removal of stocks from such Conviction lists are managed by the I nvestment Review Committee or other designated committee in each respective region and do not represent a change in the analysts invest

202、ment rating for such stocks.Total return potential represents the upside or downside dif f erential between the current share price and the price target,including all paid or anticipated dividends,expected during the time horizon associated with the price target.Price targets are required for all co

203、vered stocks.The total return potential,price target and associated time horizon are stated in each report adding or reiterating an I nvestment List membership.Coverage Universe:A list of all stocks in each coverage universe is available by primary analyst,stock and coverage universe at https:/ Rate

204、d(NR).The investment rating,target price and earnings estimates(where relevant)are not provided or have been suspended pursuant to Goldman Sachs policy when Goldman Sachs is acting in an advisory capacity in a merger or in a strategic transaction involving this company,when 12 January 2025 25Goldman

205、 SachsGS SUSTAINthere are legal,regulatory or policy constraints due to Goldman Sachs involvement in a transaction,when the company is an early-stage biotechnology company,and in certain other circumstances.Rating Suspended(RS).Goldman Sachs Research has suspended the investment rating and price tar

206、get for this stock,because there is not a suf ficient fundamental basis for determining an investment rating or target price.The previous investment rating and target price,if any,are no longer in ef f ect for this stock and should not be relied upon.Coverage Suspended(CS).Goldman Sachs has suspende

207、d coverage of this company.Not Covered(NC).Goldman Sachs does not cover this company.Not Available or Not Applicable(NA).The information is not available for display or is not applicable.Not Meaningful(NM).The information is not meaningful and is therefore excluded.Global product;distributing entiti

208、es Goldman Sachs Global I nvestment Research produces and distributes research products for clients of Goldman Sachs on a global basis.Analysts based in Goldman Sachs of fices around the world produce research on industries and companies,and research on macroeconomics,currencies,commodities and port

209、folio strategy.This research is disseminated in Australia by Goldman Sachs Australia Pty Ltd(ABN 21 006 797 897);in Brazil by Goldman Sachs do Brasil Corretora de T tulos e Valores Mobilirios S.A.;Public Communication Channel Goldman Sachs Brazil:0800 727 5764 and/or .Available Weekdays(except holid

210、ays),from 9am to 6pm.Canal de Comunicao com o Pblico Goldman Sachs Brasil:0800 727 5764 e/ou .Horrio de funcionamento:segunda-f eira sexta-f eira(exceto f eriados),das 9h s 18h;in Canada by Goldman Sachs&Co.LLC;in Hong Kong by Goldman Sachs(Asia)L.L.C.;in I ndia by Goldman Sachs(I ndia)Securities Pr

211、ivate Ltd.;in Japan by Goldman Sachs Japan Co.,Ltd.;in the Republic of Korea by Goldman Sachs(Asia)L.L.C.,Seoul Branch;in New Zealand by Goldman Sachs New Zealand Limited;in Russia by OOO Goldman Sachs;in Singapore by Goldman Sachs(Singapore)Pte.(Company Number:198602165W);and in the United States o

212、f America by Goldman Sachs&Co.LLC.Goldman Sachs I nternational has approved this research in connection with its distribution in the United Kingdom.Goldman Sachs I nternational(“GSI”),authorised by the Prudential Regulation Authority(“PRA”)and regulated by the Financial Conduct Authority(“FCA”)and t

213、he PRA,has approved this research in connection with its distribution in the United Kingdom.European Economic Area:GSI,authorised by the PRA and regulated by the FCA and the PRA,disseminates research in the following jurisdictions within the European Economic Area:the Grand Duchy of Luxembourg,I tal

214、y,the Kingdom of Belgium,the Kingdom of Denmark,the Kingdom of Norway,the Republic of Finland and the Republic of I reland;GSI -Succursale de Paris(Paris branch)which is authorised by the French Autorit de contrle prudentiel et de resolution(“ACPR”)and regulated by the Autorit de contrle prudentiel

215、et de resolution and the Autorit des marches financiers(“AMF”)disseminates research in France;GSI -Sucursal en Espaa(Madrid branch)authorized in Spain by the Comisin Nacional del Mercado de Valores disseminates research in the Kingdom of Spain;GSI -Sweden Bankfilial(Stockholm branch)is authorized by

216、 the SFSA as a“third country branch”in accordance with Chapter 4,Section 4 of the Swedish Securities and Market Act(Sw.lag(2007:528)om vrdepappersmarknaden)disseminates research in the Kingdom of Sweden;Goldman Sachs Bank Europe SE(“GSBE”)is a credit institution incorporated in Germany and,within th

217、e Single Supervisory Mechanism,subject to direct prudential supervision by the European Central Bank and in other respects supervised by German Federal Financial Supervisory Authority(Bundesanstalt fr Finanzdienstleistungsaufsicht,BaFin)and Deutsche Bundesbank and disseminates research in the Federa

218、l Republic of Germany and those jurisdictions within the European Economic Area where GSI is not authorised to disseminate research and additionally,GSBE,Copenhagen Branch filial af GSBE,Tyskland,supervised by the Danish Financial Authority disseminates research in the Kingdom of Denmark;GSBE-Sucurs

219、al en Espaa(Madrid branch)subject(to a limited extent)to local supervision by the Bank of Spain disseminates research in the Kingdom of Spain;GSBE-Succursale I talia(Milan branch)to the relevant applicable extent,subject to local supervision by the Bank of I taly(Banca dI talia)and the I talian Comp

220、anies and Exchange Commission(Commissione Nazionale per le Societ e la Borsa“Consob”)disseminates research in I taly;GSBE-Succursale de Paris(Paris branch),supervised by the AMF and by the ACPR disseminates research in France;and GSBE-Sweden Bankfilial(Stockholm branch),to a limited extent,subject t

221、o local supervision by the Swedish Financial Supervisory Authority(Finansinpektionen)disseminates research in the Kingdom of Sweden.General disclosures This research is for our clients only.Other than disclosures relating to Goldman Sachs,this research is based on current public information that we

222、consider reliable,but we do not represent it is accurate or complete,and it should not be relied on as such.The information,opinions,estimates and forecasts contained herein are as of the date hereof and are subject to change without prior notification.We seek to update our research as appropriate,b

223、ut various regulations may prevent us from doing so.Other than certain industry reports published on a periodic basis,the large majority of reports are published at irregular intervals as appropriate in the analysts judgment.Goldman Sachs conducts a global full-service,integrated investment banking,

224、investment management,and brokerage business.We have investment banking and other business relationships with a substantial percentage of the companies covered by Global I nvestment Research.Goldman Sachs&Co.LLC,the United States broker dealer,is a member of SI PC(https:/www.sipc.org).Our salespeopl

225、e,traders,and other prof essionals may provide oral or written market commentary or trading strategies to our clients and principal trading desks that reflect opinions that are contrary to the opinions expressed in this research.Our asset management area,principal trading desks and investing busines

226、ses may make investment decisions that are inconsistent with the recommendations or views expressed in this research.The analysts named in this report may have from time to time discussed with our clients,including Goldman Sachs salespersons and traders,or may discuss in this report,trading strategi

227、es that ref erence catalysts or events that may have a near-term impact on the market price of the equity securities discussed in this report,which impact may be directionally counter to the analysts published price target expectations for such stocks.Any such trading strategies are distinct from an

228、d do not af f ect the analysts fundamental equity rating for such stocks,which rating reflects a stocks return potential relative to its coverage universe as described herein.We and our af filiates,of ficers,directors,and employees will from time to time have long or short positions in,act as princi

229、pal in,and buy or sell,the securities or derivatives,if any,ref erred to in this research,unless otherwise prohibited by regulation or Goldman Sachs policy.The views attributed to third party presenters at Goldman Sachs arranged conf erences,including individuals from other parts of Goldman Sachs,do

230、 not necessarily reflect those of Global I nvestment Research and are not an of ficial view of Goldman Sachs.Any third party ref erenced herein,including any salespeople,traders and other prof essionals or members of their household,may have positions in the products mentioned that are inconsistent

231、with the views expressed by analysts named in this report.This research is not an of f er to sell or the solicitation of an of f er to buy any security in any jurisdiction where such an of f er or solicitation would be illegal.I t does not constitute a personal recommendation or take into account th

232、e particular investment objectives,financial situations,or needs of individual clients.Clients should consider whether any advice or recommendation in this research is suitable for their particular circumstances and,if appropriate,seek prof essional advice,including tax advice.The price and value of

233、 investments ref erred to in this research and the income from them may fluctuate.Past performance is not a guide to future performance,future returns are not guaranteed,and a loss of original capital may occur.Fluctuations in exchange rates could have adverse ef f ects on the value or price of,or i

234、ncome derived from,certain investments.Certain transactions,including those involving futures,options,and other derivatives,give rise to substantial risk and are not suitable for all investors.I nvestors should review current options and futures disclosure documents which are available from Goldman

235、Sachs sales representatives or at https:/ and 12 January 2025 26Goldman SachsGS SUSTAINhttps:/www.fiadocumentation.org/fia/regulatory-disclosures_1/fia-uniform-futures-and-options-on-futures-risk-disclosures-booklet-pdf-version-2018.Transaction costs may be significant in option strategies calling f

236、or multiple purchase and sales of options such as spreads.Supporting documentation will be supplied upon request.Differing Levels of Service provided by Global Investment Research:The level and types of services provided to you by Goldman Sachs Global I nvestment Research may vary as compared to tha

237、t provided to internal and other external clients of GS,depending on various factors including your individual pref erences as to the frequency and manner of receiving communication,your risk profile and investment focus and perspective(e.g.,marketwide,sector specific,long term,short term),the size

238、and scope of your overall client relationship with GS,and legal and regulatory constraints.As an example,certain clients may request to receive notifications when research on specific securities is published,and certain clients may request that specific data underlying analysts fundamental analysis

239、available on our internal client websites be delivered to them electronically through data f eeds or otherwise.No change to an analysts fundamental research views(e.g.,ratings,price targets,or material changes to earnings estimates for equity securities),will be communicated to any client prior to i

240、nclusion of such information in a research report broadly disseminated through electronic publication to our internal client websites or through other means,as necessary,to all clients who are entitled to receive such reports.All research reports are disseminated and available to all clients simulta

241、neously through electronic publication to our internal client websites.Not all research content is redistributed to our clients or available to third-party aggregators,nor is Goldman Sachs responsible for the redistribution of our research by third party aggregators.For research,models or other data

242、 related to one or more securities,markets or asset classes(including related services)that may be available to you,please contact your GS representative or go to https:/.Disclosure information is also available at https:/ or from Research Compliance,200 West Street,New York,NY 10282.2025 Goldman Sa

243、chs.You are permitted to store,display,analyze,modify,reformat,and print the information made available to you via this service only for your own use.You may not resell or reverse engineer this information to calculate or develop any index for disclosure and/or marketing or create any other derivati

244、ve works or commercial product(s),data or of f ering(s)without the express written consent of Goldman Sachs.You are not permitted to publish,transmit,or otherwise reproduce this information,in whole or in part,in any format to any third party without the express written consent of Goldman Sachs.This

245、 foregoing restriction includes,without limitation,using,extracting,downloading or retrieving this information,in whole or in part,to train or finetune a machine learning or artificial intelligence system,or to provide or reproduce this information,in whole or in part,as a prompt or input to any such system.12 January 2025 27Goldman SachsGS SUSTAIN