《方舟投资(ARK):Big Ideas 2024-年度投研报告(英文版)(163页).pdf》由会员分享,可在线阅读,更多相关《方舟投资(ARK):Big Ideas 2024-年度投研报告(英文版)(163页).pdf(163页珍藏版)》请在三个皮匠报告上搜索。
1、1BIG IDEAS2024Annual ResearchReportFOR INFORMATIONAL PURPOSES ONLYBIG IDEAS 2024JANUARY 31,2024ARK Investment Management LLC.This is not a recommendation in relation to any named particular securities/cryptocurrencies and no warranty or guarantee is provided.Any references to particular securities/c
2、ryptocurrencies are for illustrative purposes only.There is no assurance that the Adviser will make any investments with the same or similar characteristics as any investment presented.The reader should not assume that an investment identified was or will be profitable.PAST PERFORMANCE IS NOT INDICA
3、TIVE OF FUTURE PERFORMANCE,FUTURE RETURNS ARE NOT GUARANTEED.2BIG IDEAS 2024:DISCLOSURERisks Of Investing In InnovationPlease note:Companies that ARK believes are capitalizing on disruptive innovation and developing technologies to displace older technologies or create new markets may not in fact do
4、 so.ARK aims to educate investors and seeks to size the potential investment opportunity,noting that risks and uncertainties may impact our projections and research models.Investors should use the content presented for informational purposes only,and be aware of market risk,disruptive innovation ris
5、k,regulatory risk,and risks related to certain innovation areas.Please read risk disclosure carefully.DISRUPTIVE INNOVATIONRAPID PACE OF CHANGEUNCERTAINTY AND UNKNOWNS EXPOSURE ACROSS SECTORS AND MARKET CAPRISK OF INVESTING IN INNOVATIONREGULATORY HURDLESCOMPETITIVE LANDSCAPEPOLITICAL OR LEGAL PRESS
6、URESources:ARK Investment Management LLC,2023.Aim to understand the regulatory,market,sector,and company risks.(See Disclosure Page)Aim for a cross-sector understanding of technology and combine top-down and bottom-up research.3BIG IDEAS 2024:INTRODUCTIONBig Ideas 2024ARK Invest proudly presents Big
7、 Ideas 2024:Disrupting the Norm,Defining the Future.A tradition since 2017,Big Ideas offers a comprehensive analysis of technological convergence and its potential to revolutionize industries and economies.ARK seeks to deliver long-term capital appreciation by investing in the leaders,enablers,and b
8、eneficiaries of disruptive innovation.With a belief that innovation is key not only to growth but also to resilience,ARK emphasizes the necessity of a strategic allocation to innovation in every investors portfolio.This approach aims to tap into the exponential growth opportunities often overlooked
9、in broad-based indices,while simultaneously providing a hedge against the risks posed by incumbents facing disruption.We hope you enjoy Big Ideas 2024.Disrupting The Norm,Defining The Future4Reusable Rockets143Technological Convergence5Artificial Intelligence19Bitcoin Allocation34Bitcoin In 202343Sm
10、art Contracts 53Digital Consumers64Digital Wallets75Precision Therapies87Multiomic Tools&Technology96Electric Vehicles104Robotics113Robotaxis122Autonomous Logistics1333D Printing153TABLE OF CONTENTS5 5Technological ConvergenceSources:ARK Investment Management LLC,2024.Forecasts are inherently limite
11、d and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.Brett WintonChief FuturistARK Venture Investment Committee MemberResearch By
12、:BIG IDEAS 20246According to ARKs research,convergence among disruptive technologies will define this decade.Five major technology platformsArtificial Intelligence,Public Blockchains,Multiomic Sequencing,Energy Storage,and Roboticsare coalescing and should transform global economic activity.Technolo
13、gical convergence could create tectonic macroeconomic shifts more impactful than the first and second industrial revolutions.Globally,real economic growth could accelerate from 3%on average during the past 125 years to more than 7%during the next 7 years as robots reinvigorate manufacturing,robotaxi
14、s transform transportation,and artificial intelligence amplifies knowledge worker productivity.Catalyzed by breakthroughs in artificial intelligence,the global equity market value associated with disruptive innovation could increase from 16%of the total*to more than 60%by 2030.As a result,the annual
15、ized equity return associated with disruptive innovation could exceed 40%during the next seven years,increasing its market capitalization from$19 trillion today to roughly$220 trillion by 2030.*Throughout this section,we include public blockchain value as part of all calculations and forecasts of“eq
16、uity market value.”Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered inve
17、stment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.CONVERGENCE7Five Innovation Platforms Are Converging And Defining This Technological EraSources:ARK Investment Management LLC,2024.This ARK analysis is based on a range
18、of underlying data from external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is no
19、t indicative of future results.Public BlockchainsUpon large-scale adoption,all money and contracts likely will migrate onto Public Blockchains that enableand verify digital scarcity and proof of ownership.The financial ecosystem is likely to reconfigure to accommodate the rise of Cryptocurrencies an
20、d Smart Contracts.These technologies increase transparency,reduce the influence of capital and regulatory controls,and collapse contract execution costs.In such a world,Digital Wallets would become increasingly necessary as more assets become money-like,and corporations and consumers adapt to the ne
21、w financial infrastructure.Corporate structures themselves may be called into question.Multiomic SequencingThe cost to gather,sequence,and understand digital biological data is falling precipitously.Multiomic Technologies provide research scientists,therapeutic organizations and health platforms wit
22、h unprecedented access to DNA,RNA,protein,and digital health data.Cancer care should transform with pan-cancer blood tests.Multiomic data should feed into novel Precision Therapies using emerging gene editing techniques that target and cure rare diseases and chronic conditions.Multiomics should unlo
23、ck entirely new Programmable Biology capabilities,including the design and synthesis of novel biological constructs with applications across industries,particularly agriculture and food production.Artificial IntelligenceComputational systems and software that evolve with data can solve intractable p
24、roblems,automate knowledge work,and accelerate technologys integration into every economic sector.The adoption of Neural Networks should prove more momentous than the introduction of the internet and potentially create 10s of trillion dollars of value.At scale these systems will require unprecedente
25、d computational resources,and AI-specific compute hardware should dominate the Next Gen Cloud datacenters that train and operate AI models.The potential for end-users is clear:a constellation of AI-driven Intelligent Devices that pervade peoples lives,changing the way that they spend,work,and play.T
26、he adoption of artificial intelligence should transform every sector,impact every business,and catalyze every innovation platform.Energy Storage Declining costs of Advanced Battery Technology should cause an explosion in form factors,enabling Autonomous Mobility systems that collapse the cost of get
27、ting people and things from place to place.Electric drivetrain cost declines should unlock micro-mobility and aerial systems,including flying taxis,enabling business models that transform the landscape of cities.Autonomy should reduce the cost of taxi,delivery,and surveillance by an order of magnitu
28、de,enabling frictionless transport that could increase the velocity of e-commerce and make individual car ownership the exception rather than the rule.These innovations combined with large-scale stationary batteries should cause a transformation in energy,substituting electricity for liquid fuel and
29、 pushing generation infrastructure towards the edge of the network.RoboticsCatalyzed by artificial intelligence,Adaptive Robots can operate alongside humans and navigate legacy infrastructure,changing the way products are made and sold.3D Printing should contribute to the digitization of manufacturi
30、ng,increasing not only the performance and precision of end-use parts but also the resilience of supply chains.Meanwhile,the worlds fastest robots,Reusable Rockets,should continue to reduce the cost of launching satellite constellations and enable uninterruptible connectivity.A nascent innovation pl
31、atform,robotics could collapse the cost of distance with hypersonic travel,the cost of manufacturing complexity with 3D printers,and the cost of production with AI-guided robots.CONVERGENCE8Converging Technologies Are Generating A Historic Technological WaveSources:ARK Investment Management LLC,2024
32、.This ARK analysis is based on a range of underlying sources,including Bekar et al.2017,which may be provided upon request.The chart uses GPT 4 prompting to survey a comprehensive list of general purpose technologies using the identification framework detailed therein.Where available,academic litera
33、ture is also used to assess attributable economic impact.A GPT-4 scoring rubric assesses technology-by-technology impacts.The impact measured directly is matched against the scoring to tune all scores to produce technology-by-technology estimates of economic impact(even when direct measures of econo
34、mic impact are unattainable).Consistent with General Purpose Technology theory,these technologies are assumed to go through a period of investment in which economic impact is negative before productivity advances begin to realize into economic data.All technologies are assumed to have the same diffu
35、sion and realization cycle.If recent technologies are assumed to diffuse more quickly,the current wave would appear steeper.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hol
36、d any particular security.Past performance is not indicative of future results.-23850505050505050520002005200252030Estimated
37、 Economic Impact of General Purpose Technologies(Annual Percentage Point Additions to Real GDP Growth And Consumer Surplus)Integrated CircuitNuclear PowerContainerizationInternetCell phonesGPSThe WebPCsBiotechFiber opticsE-CommerceRenewables3D PrintingReusable RocketsAdaptive robotsAdvanced Batterie
38、sAutonomous MobilityCloud ComputingAIIntelligent DevicesMultiomic TechnologyPrecision TherapiesProgrammable BiologyDigital WalletsSmart ContractsCryptocurrenciesInternal Combustion EngineElectricityTelephoneRadioRefrigerationAir ConditioningChemicals&Synthetic MaterialsAutomobileAssembly LineTelevis
39、ionJet EngineRailroadsTelegraphPhotographyBicycleSteam Engine2025 F2030 FCONVERGENCE9AI Serves As The Central Technology CatalystHighestHighMidLowLowestThe Technology Convergence matrix illustrates the relationships between and among technologies.More detailed version of this graphic,including detai
40、led scoring information and justification available here.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes
41、only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.CryptocurrenciesSmartContractsDigitalWalletsPrecisionTherapiesMultiomicTechnologyProgrammableBiologyNeuralNetworksNext GenCloudInte
42、lligentDevicesAutonomousMobilityAdvanced BatteryTechnologyRenewableRocketsAdaptiveRobotics3DPrintingCrypto-currenciesSmart ContractsDigitalWalletsPrecision TherapiesMultiomicTechnologyNeuralNetworksNext GenCloudIntelligentDevicesAutonomousMobilityAdvancedBatteryTechnologyRenewableRocketsAdaptiveRobo
43、tics3DPrintingProgrammableBiologyConvergence ScoreTechnologyCatalyzing Technology10AI Is Accelerating Faster Than Forecasters AnticipatedSources:ARK Investment Management LLC,2024,based on data from Metaculus,including benchmark details,as of January 3,2024.Benchmark broadly requires the successful
44、passage of an adversarial two-hour Tuning test,broad success on a Q&A knowledge and logic benchmark,and the successful interpretation of and execution complex model car assembly instruction,all within a single system.Green lines are derived estimates for time to general purpose AI(strongly formulate
45、d)based upon forecasts for a weaker benchmark.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results
46、.Google demonstrates advanced conversational agent,LLaMda2ChatGPT launches to the publicGPT-4 launchesOpenAI announces GPT-302020224202520262027202820292030Number of YearsExpected Years Until Launch Of A General Artificial Intelligence System(Log Scale)Pre GPT-3 average80 years
47、If forecast error continuesIf forecast is well-tuned34 years18 years8 years50 yearsCONVERGENCE11Individual Technology Advances Can Coalesce And Cascade Into Massive New Market OpportunitiesAdvanced AI enables robotaxis to rely on fewer,less expensive sensors.Battery electric drivetrains reduce robot
48、axi operating costs by 60%.$0.31$0.12 Internal combustionElectricRobotaxi Operating Cost Per MileBy Drivetrain TypeNeural NetworksAdvanced Battery Technology+=The combination of AI and battery electric drivetrains enables robotaxi systems to scale.Autonomous MobilityIn addition to better batteries a
49、nd AI,general purpose robots will require better:Electric motorsPower electronicsSensorsPower-efficient computeAs robotaxis scale,the cost of each technology should decline according to its learning curve.Adaptive Robotics*Waymo manufacturing costs are estimated based upon public statements.Sources:
50、ARK Investment Management LLC,2024.This ARK analysis is based on a range of external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or
51、 hold any particular security.Past performance is not indicative of future results.50100150200WaymoTeslaThousands,$Robotaxi Manufacturing Costs(Per Vehicle,2024)*5 LIDARs,29 cameras,6 radars,8 ultrasonic sensors9 CamerasCONVERGENCE12The Impact Of These Technologies On The Economy Should Prove Dramat
52、ic*Adaptive Robotics,Autonomous Mobility,and AI Impact are ARK Invest estimates.AI estimate includes consumer surpluses that may not be captured in traditional economic statistics.IT productivity impact likely also undercounts consumer surplus.Industrial Robot and IT impact measures impact on US,Eur
53、ope,and Japanese economies.Steam Engine impact is measured against the UK economy.Sources:ARK Investment Management LLC,2024,based on data from Crafts 2004,OMahony et al.2009,and McKinsey Global Institute 2017.Forecasts are inherently limited and cannot be relied upon.For informational purposes only
54、 and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.*0%20%40%60%80%100%120%140%Industrial robots(1997 to 2007)Information Technology(1995 to 2005)Adaptive robotics(2023 to 2030)Autonomous
55、 Mobility(2023 to 2030)Steam engine(1830 to 1910)AI(2023 to 2030)Economic Impact of Select Major Technologies(Cumulative Increase In Real GDP Attributable to Technology After Introduction)Industrial Robots(1997 to 2007)Adaptive Robotics*(2023 to 2030)Steam Engine(1830 to 1910)CONVERGENCE13Technologi
56、cal Innovation Could Be Disruptive Enough To Dominate Global Equity Market Capitalizations2023Equity Market Cap EstimateNon-innovation$98 trillionDisruptive Innovation$19 trillionTotal$117 trillion2030Equity Market Cap ForecastNon-innovation$140 trillionDisruptive Innovation$220 trillionTotal$360 tr
57、illionAnnual GrowthForecast3%42%17%ArtificiaI Intelligence37%Energy Storage50%Public Blockchains48%Robotics78%Multiomic Sequencing39%Note:Forecasted numbers are rounded.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of external sources,including the World Federation
58、 of Exchanges and the MSCI ACWI IMI Innovation Index which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past perfo
59、rmance is not indicative of future results.CONVERGENCEPublic BlockchainsMultiomic SequencingEnergy StorageAIRobotics14Expectations For Public BlockchainsTechnology2040 PossibilitiesARKs 2030 Expectation of ProgressCryptocurrenciesCryptocurrencies have displaced most permission-based,centrally contro
60、lled monetary systems,enabling financial ecosystems to reformulate around a digital asset that can eliminate counterparty risk while continuing to facilitate transaction flows.The reformulation began at the edges of the traditional financial system in geographies with broken money systems and in mar
61、kets otherwise mis-served by traditional financial intermediaries.In developed markets,cryptocurrencies initially served as a store of value,providing little direct utility.Over time,the efficiencies of a truly neutral digital currency,primarily bitcoin,have prevailed over other financial architectu
62、res.Global money supply has grown in tandem with GDP,and cryptocurrencies now account for 10%of the total.Little of that value accrual is attributable to the direct displacement of money though there are instances in emerging markets.Much of the appreciation is a function of low single-digit percent
63、 allocations by institutional and high net worth individuals as well as corporate and nation-state treasuries.Cryptocurrencies continue to displace gold as a flight-to-safety asset,taking 40%share of the market.Utility use cases such as remittances and global settlements account for 10%and 5%of volu
64、mes,respectivelySmart ContractsMost contracts have migrated to open-source protocols that enable and verify digital scarcity and proof of ownership.Risk-sharing arrangements are more transparent,assets of all sorts are securitized,bought,and sold more easily,and counterparty risks have diminished su
65、bstantially.The importance of traditional financial intermediaries has dwindled,even as more human activity becomes commercialized.Decentralized protocols,enabled by balance-sheet-light digital wallet platforms,facilitate most traditional financial functions.Consumer internet services rely on busine
66、ss models enabled by digital asset ownership.Every corporate entity and every consumer has adapted as centralized corporate structures themselves are called into question.Global financial assets as percent of GDP have continued to increase,with less than 5%secured by smart contracting platformsa dyn
67、amic consistent with the adoption curve of dialup internet.At 1%,the gross take from tokenized assets on decentralized protocols is less than a third of the fees that traditional financial institutions extract.Application protocols,which pay a larger share of fees to incentivize network participants
68、,account for 75%of gross decentralized protocol revenues.The blended net take rate between application layer protocols and Level 1 protocols is roughly 60bps.Digital WalletsDigital wallets enable nearly every person with a connected device to transmit and receive money instantly,fundamentally transf
69、orming the through-flow of commercial and financial experiences.Digital wallets that facilitate wholesale pricing of financial services for individual users have disrupted retail banking relationships,fundamentally transforming consumer relationships with financial service providers.In addition to t
70、heir financial functions,digital wallets are distribution platforms for a variety of digital servicesfrom ride-hailing to e-commerceand are secure repositories for digital health and other sensitive data.Traditional financial service institutions and their associated payment processing value chains
71、have given way largely to internet-enabled digital wallets for most economic activity.Roughly 90%of smartphone users rely on digital wallets to some degree.The majority uses digital wallets as the front-end for more than half of meaningful financial functions.Digital wallet platform providers contin
72、ue to rely on traditional ecosystems to facilitate financial activities like lending but can extract lead generation fees of 5-20%for delivering customers to those institutions.They also can capture 3-10%commerce facilitation fees for e-commerce activity directed through their platforms.Sources:ARK
73、Investment Management LLC,2024.In the above table,we characterize the convergent technological capabilities that we believe may manifest by 2030 and 2050.We stress that these scenarios,written in the present tense,are possible outcomesnot assured outcomesand that the future may play out differently.
74、This ARK analysis is based on a range of external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Altho
75、ugh the scenarios described in the table below are written in present tense,they are forecasted,possible outcomes based on ARKs views.These possible outcomes may not be realized in the future due to a number of uncertainties.The information provided should not be considered investment advice and sho
76、uld not form the basis of any investment decision.CONVERGENCE15Expectations For Multiomic SequencingTechnology2040 PossibilitiesARKs 2030 Expectation of ProgressPrecision TherapiesTechnology enables the manipulation of molecular biological systems,catalyzing a new generation of more efficacious and
77、durable precision therapies.CRISPR-based gene-editing enables the manipulation of DNA directly with increasing specificity.RNA-acting therapeutic techniques restrict the area of DNA that can be transcribed into proteins.AI-advances enable the targeting of specific proteins that cause underlying diso
78、rders.These breakthroughs have shortened development timelines for and increased the efficacy of curative therapies that command higher prices than traditional therapies.Researchers are aiming to cure most rare diseases.Traditional health service spending declines,ceding economic terrain to molecula
79、r cures.Precision therapies make up 25%of newly released drugs.By improving the quality of life,lowering ancillary medical costs,and often effectively curing diseases,they command average price premiums of 7x relative to traditional drugs.Combined with expected improvements in R&D efficiencies,these
80、 drugs add 15%or$300 billion to drug revenues in 2030.Multiomic TechnologiesCatalyzed by the precipitous fall in sequencing costs,researchers and clinicians routinely collect patients epigenomic,transcriptomic,and proteomic data.With increasingly comprehensive digital health readouts from intelligen
81、t devices and emerging AI tools,they align this panoply of multiomic data to understand,predict,and treat disease.As a result,cancer care has transformed completely:multiomic technologies detect cancer at early stages,target treatment more precisely,and provide recurrence monitoring.Regular blood-ba
82、sed pan-cancer tests are a standard of care for patients in middle age.Multiomic technology has increased biotech R&D efficiency,as clinical trials target patient populations and measure outcomes more precisely and easily.Combined with AI,multiomic technology has transformed the relationship between
83、 patients and health systems.Digital health providers,diagnostic tool companies,and molecular testing companies are leading the charge.Legacy drug franchises and health service systems have lost their prominence.Wasteful healthcare spending declines as healthy lives extend.At full penetration,R&D ef
84、ficiency associated with drug development could double,thanks to AI-enhanced multiomic technology.By 2030,nearly all new drug development programs incorporate multiomics into preclinical R&D,and 50%incorporate AI into clinical programs.Realized returns on R&D have improved by 10%with line-of-sight t
85、o a near doubling of R&D returns by 2035.Early detection multi-cancer blood tests have become standard of care as they have cut cancer mortality by 25%for some age cohorts.In developed markets,30%of patients benefit from the new diagnostics regime.Programmable BiologyAI tools,improved genomic synthe
86、sis techniques,and scalable biological manufacturing techniques enable novel,lower cost biological constructs with predictable performance,powering a renaissance in agriculture and materials science.Programmable biology enables breakthroughs in materials science and bio-based fuels that increase foo
87、d production and reduce environmental externalities.Molecular biological primitives offer a substrate for new robust computation architectures.Still restricted to early stage and development projects,gene synthesis generates$10 billion in annual revenue.Programmable biology platforms capture 10%of p
88、recision therapy revenue.Those platforms generate another$30 billion in revenue with gross margins at 70%,EBITDA margins in the 35%range,and free cash flow margins at 20%.Sources:ARK Investment Management LLC,2024.In the above table,we characterize the convergent technological capabilities that we b
89、elieve may manifest by 2030 and 2050.We stress that these scenarios,written in the present tense,are possible outcomesnot assured outcomesand that the future may play out differently.This ARK analysis is based on a range of external sources,which may be provided upon request.Forecasts are inherently
90、 limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Although the scenarios described in the table below are written in present tense,they are forecasted,possible outcomes bas
91、ed on ARKs views.These possible outcomes may not be realized in the future due to a number of uncertainties.The information provided should not be considered investment advice and should not form the basis of any investment decision.CONVERGENCE16Expectations For Energy StorageTechnology2040 Possibil
92、itiesARKs 2030 Expectation of ProgressAutonomous MobilityRobots move people and parcels from place to place and have changed the economics of physical movement entirely.The cost of taxi,delivery,and observation have fallen by an order of magnitude.Traveling by robotaxi is the norm and owning a perso
93、nal vehicle the exception.Frictionless drone and robot delivery has catalyzed the velocity of ecommerce.The data generated by autonomous mobility systems provide pervasive,real-time insights into the state of the world.Consumers and businesses that harness autonomous mobility platforms are benefitti
94、ng,while prior incumbents in the automotive,logistics,retail,and insurance sectors have been upended.Autonomous robotaxis have transformed global transport,as point-to-point transportation is available in nearly every country at an average price of$.50 per mile.Given the compelling price-point and u
95、tility,robotaxis have traveled 13 trillion vehicle miles and are gaining traction.Autonomous robotaxi platforms charge platform fees or take-rates of 50%+,generate 50%operating margins,and give asset owner-operators the opportunity to generate reasonable rates of return on capital.The number of auto
96、nomous vehicles facilitating this travel is 100 million,and most of the incremental vehicle production is autonomous-capable.Advanced Battery SystemsDeclining battery costs have ignited a Cambrian explosion in mobility form factors,pushing electrical supply out to end-nodes on networks.Electric vehi
97、cles dominate transport as internal combustion dies.Micro-mobility and aerial systems that include flying taxis enable innovative business models that transform urban landscapes.All these innovations drive fundamental demand for electrical energy at the expense of liquid fuel.They also provide elect
98、rical energy more efficiently,reducing the vulnerability of grids,operational expenses,and the capital intensity of transmission and distribution.Oil demand is in decline,and traditional automotive manufacturers and suppliers have been displaced by a smaller number of vertically integrated technolog
99、y providers.As ridership shifts to electric autonomous platforms,the number of autonomous capable EVs sold annually is 74 million,accounting for most of the automotive market.At an average selling price of$20,000,EV manufacturers generate$1.4 trillion in annual revenue,20%gross margins,and 10%EBIT m
100、argins.With manufacturing consolidation,margins increase.Batteries account for 20%of the value of EVs.Much like that of EVs,battery manufacturing is capital-intensive and low-margin.Supplying the EV OEMs,battery manufacturers generate revenue of$300 billion per year.Stationary energy storage require
101、s a volume of batteries roughly equivalent to that consumed by EVs,generating another$300 billion in revenue.Sources:ARK Investment Management LLC,2024.In the above table,we characterize the convergent technological capabilities that we believe may manifest by 2030 and 2050.We stress that these scen
102、arios,written in the present tense,are possible outcomesnot assured outcomesand that the future may play out differently.This ARK analysis is based on a range of external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes
103、only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Although the scenarios described in the table below are written in present tense,they are forecasted,possible outcomes based on ARKs views.These possible outcomes may not be realized i
104、n the future due to a number of uncertainties.The information provided should not be considered investment advice and should not form the basis of any investment decision.CONVERGENCE17Expectations For Artificial IntelligenceTechnology2040 PossibilitiesARKs 2030 Expectation of ProgressNeural Networks
105、Fed by massive amounts of data,computational systems and software are solving previously unsolvable problems,automating knowledge work,and accelerating the integration of technology into all economic processes.As costs have plummeted,custom software is improving with every AI model enhancement and c
106、onnecting the world.Learning systems are blazingly fast,their impact as momentous as the introduction of the microprocessor,transforming every sector and region.The cost of training AI models has fallen more than 40,000-fold which,when combined with aggressive investments in AI hardware,has catapult
107、ed aggregate AI capability roughly 600,000-fold since 2023.Adopted by 50%of knowledge workers,AI software systems have improved their productivity by 9x on average.Consistent with other software products,enterprises pay 10%of the productivity increase to access the software.Next Gen CloudCloud tools
108、 train the AI models that dominate software stacks and the software connections that stitch together the AI-run world.The infrastructure-as-a-service providers,chip manufacturers,and tool-manufacturers that facilitate the training of neural networks have enjoyed a multi-decade demand cycle.Software
109、development has been democratized,and the companies providing API hooks that stitch together interoperable software layers experience unprecedented demand.AI hardware spend of$1.3 trillion supports$13 trillion in AI software sales and accommodates traditional software gross margins of 75%.Three type
110、s of customers support the demand for AI hardware-infrastructure-as-a-service providers,software companies,and AI foundation model providerswhich should generate 20%cashflow margins,consistent with those of chip manufacturers.Intelligent DevicesAI powers a new class of intelligent devices in the hom
111、e and on the go.Fixed internet-and AI-powered infrastructure exists in homes and other social environments,transforming distribution for all media providers.End-users interface with the world in completely new ways,and data on their consumption preferences spawn new business models and services.Comm
112、erce and wagering permeate entertainment experiences,enabling and catalyzing new advertising formats and content monetization.The show is the store.Linear TV is obsolete,as digital curation and direct consumer preference drive visual content.Linear content is ceding ground to interactive experiences
113、,sometimes subtly.AI-mediated glasses and headsets thread through the fabric of everyday life.Consumer spending on intelligent device hardware continues its uptrend to$60 per internet user per year.Time spent connected grows dramatically to half of waking leisure hours,or 20 trillion globally.Digita
114、l experiences continue to monetize at a discount to in-person experiences and yield$0.25 per hour spent online in revenue to platform providers.Between device spend and digital entertainment experiences,$5.4 trillion in revenue accrues to intelligent devices,entertainment,and social platforms.Advert
115、ising and commerce comprise 80%of that revenue.Sources:ARK Investment Management LLC,2024.In the above table,we characterize the convergent technological capabilities that we believe may manifest by 2030 and 2050.We stress that these scenarios,written in the present tense,are possible outcomesnot as
116、sured outcomesand that the future may play out differently.This ARK analysis is based on a range of external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recom
117、mendation to buy,sell,or hold any particular security.Although the scenarios described in the table below are written in present tense,they are forecasted,possible outcomes based on ARKs views.These possible outcomes may not be realized in the future due to a number of uncertainties.The information
118、provided should not be considered investment advice and should not form the basis of any investment decision.CONVERGENCE18Expectations For RoboticsTechnology2040 PossibilitiesARKs 2030 Expectation of ProgressReusable RocketsReusable rockets are inexpensive and have spawned new business models.Low-ea
119、rth orbit constellations connect every smartphone user on earth to a censor-resistant data feed.Hypersonic point-to-point travel is becoming a reality,disrupting long-haul flight,transforming military asset delivery,and shrinking global supply chains.Extra-planetary human exploration has begun rampi
120、ng.Led by SpaceXs Starship launch volumes,a 40,000 strong satellite network is in orbit,facilitating direct-to-satellite communications for nearly all smartphones and delivering broadband-type speeds to ships,RVs,airplanes,and rural residents in developed and developing countries.Given the relative
121、ease with which customers can be on-boardeda power outlet,an antenna,and a clear path to the skymost customers are engaged in an addressable market totaling$130 billion annually.Adaptive RoboticsAdaptive robots powered by artificial intelligence are transforming the economy.The cost of humanoid robo
122、ts that are backward-compatible with existing infrastructure has dropped below that of human manufacturing labor for many applications.Previously intractable household tasks are submitting to automation at price points that create compelling end-markets.Fleets of robots grow more performant with eve
123、ry AI software upgrade.A virtuous circle of fleet data generation and AI model training drives performance forward.Manufacturing productivity growth accelerates as a wider array of physical goods submit to technologically-driven cost declines.Robots continue to penetrate the service sector as well.T
124、he economy has entered a period of undeniable and unprecedented explosive growth.Adaptive robots have penetrated manufacturing processes enough to increase productivity by 15%,and annual unit sales of humanoid robots have grown to 10%of the number of humans in the manufacturing workforce.Less expens
125、ive robots in human form-factors have begun to populate households,particularly in developed countries.While still limited in capability,these robots address a third of household chores,their sticker prices justified by the time that household members save.Robot manufacturers enjoy margins at the hi
126、gher end of capital equipment suppliers,thanks to software.3D Printing3D printing has removed design barriers and reduced cost,weight,and time to production,dramatically transforming traditional manufacturing methods.Healthcare tools created with 3D printing are personalized and custom-made,resultin
127、g in better experiences for both patients and doctors.Lighter 3D-printed aerospace parts reduce global emissions and give flight to new aircraft both for earth and outer space.Replacement parts across industries are printed on demand at a fraction of previous costs,ultimately short-circuiting supply
128、-chain shortfalls.3D printing enables artificial intelligence to design parts once impossible to manufacture.3D printing continues to dominate the prototyping market and has penetrated substantial parts of the intermediate tooling market,enabling low-cost design iterations across injection molding a
129、nd metal casting applications.Most important to industry growth,3D printing has begun to see meaningful uptake into end-use applications across aerospace and automotive,markets that collectively sell more than$4 trillion in equipment per year.Across all industries,nearly$900 billion in end-use parts
130、 could adopt 3D printing,though that penetration remains in the teens.Sources:ARK Investment Management LLC,2024.In the above table,we characterize the convergent technological capabilities that we believe may manifest by 2030 and 2050.We stress that these scenarios,written in the present tense,are
131、possible outcomesnot assured outcomesand that the future may play out differently.This ARK analysis is based on a range of external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered inves
132、tment advice or a recommendation to buy,sell,or hold any particular security.Although the scenarios described in the table below are written in present tense,they are forecasted,possible outcomes based on ARKs views.These possible outcomes may not be realized in the future due to a number of uncerta
133、inties.The information provided should not be considered investment advice and should not form the basis of any investment decision.CONVERGENCE1919Artificial IntelligenceScaling Global Intelligence And Redefining WorkSources:ARK Investment Management LLC,2024.Forecasts are inherently limited and can
134、not be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.Frank Downing Director of Research,Next Generation InternetJozef SojaResearch Associa
135、te Research By:BIG IDEAS 202420With superhuman performance on a wide range of tests,AI models like GPT-4 should catalyze an unprecedented boom in productivity.Jolted by ChatGPTs“iPhone”like moment,enterprises are scrambling to harness the potential of artificial intelligence(AI).AI promises more tha
136、n efficiency gains,thanks to rapidly falling costs and open-source models.If knowledge worker productivity were to quadruple by 2030,as we believe is likely,growth in real GDP could accelerate and break records during the next five to ten years.Sources:ARK Investment Management LLC,2024.Forecasts ar
137、e inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.ARTIFICIAL INTELLIGENCE21Pre-ChatGPT AveragePost-ChatGPT
138、Average0204060800180Q421Q122Q222Q322Q422Q123Q223Q323Number of MentionsThe Number of AI Mentions Tripled On Earnings CallsAlphabetAppleAmazonMetaMicrosoft00708090100012345Monthly Active Users(Millions)YearsChatGPT Users Hit 100 Million Users In Two MonthsChatGPTWeChatTikTokInsta
139、gramYouTubeFaceBookChatGPT Delighted Consumers And Amazed EnterprisesBuilding on years of progress since Google invented transformer architecture in 2017,ChatGPT catalyzed the publics understanding of generative AI.No longer a tool just for developers,ChatGPTs simple chat interface enabled anyone sp
140、eaking any language to harness the power of large language models(LLMs).In 2023,enterprises scrambled to understand and deploy generative AI.*values between 0 and 100 million users are estimatesSources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of data sources,which are
141、 available upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.ARTIFICIAL INTELLIGEN
142、CE22AI Already Has Boosted Productivity SignificantlyCoding assistants like GitHub Copilot and Replit AI are early success stories that have boosted the productivity and job satisfaction of software developers.AI-powered assistants are increasing the performance of knowledge workers and,interestingl
143、y,benefiting underperforming workers relatively more than high performers.Without Gen AIWith Gen AITask Speed1.25xWithout Gen AIWith Gen AITask QualityTask Quality,Top 50th Percentile of WorkersTask Quality,Bottom 50th Percentile of Workers1.17x1.43xProductivity of Consultants Using Gen AI In 2023So
144、urces:ARK Investment Management LLC,2024.The data used to analyze productivity were collected from several different studies with varying numbers of participants and definitions of task quality.The sources used are DellAcqua et al.2023 and GitHub 2022.Forecasts are inherently limited and cannot be r
145、elied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.Without CopilotWith CopilotProductivity of Developers On Coding Tasks Using Github Copilot in
146、 20232.2xARTIFICIAL INTELLIGENCE230%10%20%30%40%50%60%70%80%90%100%USABO*Semifinal 2020GRE VerbalSAT EBRWGRE WritingAP Environmental ScienceUniform Bar ExamSAT MathAP US HistoryLSATAP US GovernmentAP Art HistoryAP StatisticsAP BiologyAP MacroeconomicsAP PsychologyAP MicroeconomicsGRE QuantitativeAP
147、ChemistryAP Physics 2AP World HistoryAMC 12AP Calculus BCAP English LanguageAP English LiteratureAMC 10Codeforces RatingPercentileGPT-3.5,GPT-4,and Claude 2 Results on Professional and Academic ExamsGPT-3.5GPT-4Claude 2GPT-4 VisionFoundation Models Are Improving Across DomainsWith larger training da
148、tasets and more parameters,GPT-4 outperforms GPT-3.5 significantly.Increasingly,foundation models are becoming“multimodal”supporting text,images,audio,and videoand are not only more dynamic and user friendly,but also more performant.*USA Biology Olympiad,a prestigious national competition testing hi
149、gh school students in biology.Sources:ARK Investment Management LLC,2024,based on data from OpenAI and Anthropic as of Jan.9,2024.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,
150、or hold any particular security.Past performance is not indicative of future results.ARTIFICIAL INTELLIGENCE24Text-To-Image Models Are Reinventing Graphic Design Eight years after researchers at the University of Toronto introduced the first modern text-to-image model,the output from image models no
151、w rivals that of professional graphic designers.A human designer can create an imagelike a herd of elephants walking across a green grass fieldin several hours for several hundred dollars.Text-to-image models can produce the same graphic in seconds for pennies.Professional apps like Adobe Photoshop
152、and consumer apps like Lensa and ChatGPT are integrating image models into their products and services.February 2016alignDRAWFebruary 2022Midjourney v1November 2022Midjourney v4December 2023Midjourney v6Sources:ARK Investment Management LLC,2024.Images sourced from Masimov et al.2016 and Midjourney.
153、Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.A herd of elephants walking across a green gr
154、ass fieldARTIFICIAL INTELLIGENCE25The Cost Of Authoring The Written Word Has CollapsedOver the past century,the cost of authoring written content has been relatively constant in real terms.During the past two years,as the writing quality of LLMs has improved,the cost has collapsed.Sources:ARK Invest
155、ment Management LLC,2024.This ARK analysis is based on a range of data sources as of Jan 9,2024,which are available upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,o
156、r hold any particular security.Past performance is not indicative of future results.ARTIFICIAL INTELLIGENCEGPT4 32k$0.16Median GRE Analytic WritingClaude 2$0.04Top Decile GRE Analytic Writing$0$0$1$10$100$1,0008765432196519
157、6876542007200192022Cost Per 1000 Words Written,2023 Dollars,Log ScaleThe Cost of Authoring Written ContentPost 1997 assumes constant words per employed writer over time$0.1026AI Training Performance Is Improving RapidlyOther Algorithmic Innovati
158、ons Llama2 suggests superior writing ability of LLMs is fundamentally driven by reinforcement learning from human feedback(RLHF)Optimized prompts can outperform human prompts by over 50%Speculative Decoding speeds up inference 2-3x on certain models Flash Attention 2 results in a 2.8x training speed
159、up in GPT modelsAI researchers are innovating across training and inference,hardware,and model designs to increase performance and lower costs.Model Training Performance GainsIncreaseDecreaseTotal2023 PerformanceMoores Law Predicted ImprovementNVIDIAs Outperformance of Moores LawChinchilla Optimal S
160、calingOther Software Innovations2024 Performance ForecastAlgorithmic OptimizationsMoores Law Accelerator Optimizations5xBase=1xSources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of data sources,including Benaich 2023,Touvron et al.2023,Yang et al.2023,Leviathan et al.20
161、22,and Dao 2023,which are available upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future resu
162、lts.ARTIFICIAL INTELLIGENCE27Training Costs Should Continue To Fall 75%Per YearAccording to Wrights Law,improvements in accelerated compute hardware should reduce AI-relative compute unit(RCU)production costs by 53%per year,while algorithmic model enhancements could lower training costs further by 4
163、7%per year.In other words,the convergence of hardware and software could drive AI training costs down by 75%at an annual rate through 2030.0.000 0.000 0.001 0.010 0.100 1.000 0 1 100 10,000 1,000,000TFS-Days*(Log Scale)Cumulative RCUs Produced(Millions)(Log Scale)AI Software Training Cost Using Neur
164、al NetworksEstimated Compute*TFS-Days is a measure of compute required to train a model.Wrights Law states that for every cumulative doubling of units produced,cost will fall by a constant percentage.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of data sources as
165、of Jan.9,2024,which are available upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future result
166、s.$0.01$0.10$1.00$10.00$100.00$1,000.00$10,000.00$100,000.00 0 1 100 10,000 1,000,000 100,000,000$/RCU(Log Scale)Cumulative RCUs Produced(Millions)(Log Scale)AI Training Hardware CostActual$/RCUPredicted$/RCUARTIFICIAL INTELLIGENCEActual Compute28GPT-3GPT-3GPT-3.5 TurboGPT-4-32kGPT-4 Turbo$-$0.01$0.
167、02$0.03$0.04$0.05$0.06$0.07$0.0811/18/20219/1/202211/6/20233/14/202311/6/2023Date of Price ChangeGPT-3 and GPT-4 API Inference CostsPer 1,000 Tokens92%Annualized Cost Decline86%Annualized Cost DeclineAs Production Use Cases Emerge,AI Focus Is Shifting To Inference CostsAfter focusing initially on LL
168、M training cost optimization,researchers now are prioritizing inference costs.Based on enterprise scale use cases,inference costs seem to be falling at an annual rate of 86%,even faster than training costs.Today,the inference costs associated with GPT-4 Turbo are lower than those for GPT-3 a year ag
169、o.GPT-4-32k:Context Window:32k TokensSpeed:12 Tokens/SecGPT-4 Turbo:Context Window:128k Tokens,4xSpeed:44 Tokens/Sec,4xSources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of data sources,including Patel and Kostovic 2023,and ARK Investment Management LLC 2023,which are a
170、vailable upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.ARTIFICIAL INTELLIGENCE
171、29The Open-Source Community Is Competing With Private ModelsChallenging closed-source models from OpenAI and Google,the open-source community and its corporate champion,Meta,are democratizing access to generative AI.On balance,the performance of open-source models is improving faster than that of cl
172、osed-source models,helped recently by models from China.Note:The charts trendlines are fit to the most performant open-or closed-source models on 5-Shot MMLU(Massive Multitask Language Understanding)at the time of their release.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on
173、 a range of data sources as of Jan.9,2024,which are available upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not
174、 indicative of future results.ARTIFICIAL INTELLIGENCEGPT-2 1.5B(OpenAI,Fine-Tuned)GPT-3(OpenAI,Fine-Tuned)Chinchilla 70B(Alphabet)PaLM 540B(Alphabet)Flan-PaLM(Alphabet)Claude 1.3(Anthropic)GPT-4(OpenAI)Flan-T5-XXL(Alphabet)LlaMA 65B(Meta)LlaMA 2 70B(Meta)Falcon 180(TII,UAE)Yi-34B(01.AI,China)Qwen-72
175、B(Alibaba,China)GPT-3.5(OpenAI)PaLM-2(Alphabet)Flan-PaLM 2(Alphabet)Claude 2(Anthropic)Grok-1(X.ai)Gemini Ultra(Alphabet)Mixtral 8x7B(Mistral)-0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1.010/27/20185/15/201912/1/20196/18/20201/4/20217/23/20212/8/20228/27/20223/15/202310/1/20234/18/2024Absolute Log Error M
176、MLU PerformanceOpen Source vs Private Models 5-Shot MMLU PerformancePrivateOpen SourcePrivate(Doesnt Outperform Previous Models on 5-Shot MMLU)Open Source(Doesnt Outperform Previous Models on 5-Shot MMLU)Average Human Performance3000708090100USABO*Semifinal 2020Uniform BarExamSATAdvancedS
177、ommelierWinoGrande(commonsense)ScoreSelect GPT-4 Benchmark ResultsHuman Avg.GPT-4Language Model Performance Advances Require Nuanced TechniquesGPT-4 performs significantly better than the average human on standardized education tests,from the SAT to the Advanced Sommelier exam.Yet,it lags human-leve
178、l capability in commonsense reasoning,as measured by WinnoGrande.Stanfords frameworkHolistic Evaluation of Language Models(HELM)is one of the most comprehensive,continuously updated evaluation methodologies,having tested over 80 models against a combination of 73 scenarios and 65 metrics.HELM Evalua
179、tion MetricsAccuracyComparison with ground truth dataCalibrationProbability distribution assessmentRobustnessStress testing with perturbed inputsFairnessPerformance across diverse groupsBiasAnalysis of decision patterns for skewToxicityDetection rate of harmful contentEfficiencyResource usage during
180、 task execution*USA Biology Olympiad,a prestigious national competition testing high school students in biology.Sources:ARK Investment Management LLC,2024,based on data from Life Architect 2023 and Bomasani et al.2023 as of Jan.9,2024.Forecasts are inherently limited and cannot be relied upon.For in
181、formational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.ARTIFICIAL INTELLIGENCE330GPT-3Training TokensLlama 2Training TokensGPT-4Training TokensTokens Post
182、ed On XAnnual EstimateSpoken LanguageTokensAnnual EstimateTokens(Trillions)Leading LLM Training Set vs.Language Token StockGPT-3Training TokensLlama 2Training TokensGPT-4Training TokensSpoken Language TokensAnnual EstimateTokens Posted On XAnnual EstimateUntapped Data Sources30 quadrillion words spo
183、ken annuallySpeech-to-text tools that capture the estimated 80+trillion words spoken daily.Synthetic data that augments primary data.Autonomous taxis,trucks,drones,and other robots that generate large volumes of physical world data.Sources:ARK Investment Management LLC,2024.This ARK analysis is base
184、d on a range of data sources as of Jan 9,2024,which are available upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is
185、 not indicative of future results.40 Quadrillion TokensComputing power and high-quality training data appear to be the primary contributors to model performance.As models grow and require more training data,will a lack of fresh data cause model performance to plateau?Epoch AI estimates that high-qua
186、lity language/data sources like books and scientific papers could be exhausted by 2024,though a larger set of untapped vision data still exists.Will LLMs Run Out Of Data,Limiting Their Performance?ARTIFICIAL INTELLIGENCE320%5%10%15%20%25%Business EmailEmail MarketingIT Service ManagementCRMIT Incide
187、ntResponseSmart TransportationAnalyticsCloud-Based SecurityPlatforms%of Value CapturedTake-Rate Of Notable Enterprise Software SolutionsCustomized AI Offerings Should Enjoy More Pricing Power As open-source alternatives emerge and costs decline,software vendors tailoring AI to end-use applications s
188、hould be able to monetize them more readily.Conversely,simple generative AI applications are likely to commoditize rapidly.Horizontal,Commoditized Tools 1 Year Ago(%)2023$42,225$427.7523.270.2%51.714.82022$16,553$380.7254.366.5%43.314.1Transaction Count5(Non-Inscriptions Related,Thousands)367.5256.2
189、00500600Apr 22Jul 22Oct 22Jan 23Apr 23Jul 23Oct 23Exahashes/sBitcoins Hash Rate2,A Proxy for Network Security,Hit An All-Time High In 20231 The on-chain volume-weighted average price of the market,calculated by aggregating the value of all bitcoins in circulation at the time when they las
190、t moved.Also known as realized price or realized cap.2 The estimated computational power mining within and providing security to the Bitcoin network.3 Number of addresses in the Bitcoin network with a balance larger that zero.4 Bitcoin supply last moved 155 days ago or more,the threshold at which th
191、e possibility of a bitcoin remaining unmoved increases drastically.5 Number of transactions between two addresses of the Bitcoin network.Sources:ARK Investment Management LLC,2024,based on data from Glassnode as of December 31,2023.For informational purposes only and should not be considered investm
192、ent advice or a recommendation to buy,sell,or hold any particular security or cryptocurrency.Past performance is not indicative of future results.BITCOIN IN 202350CME*Surpassed Binance As The Worlds Largest Bitcoin Futures Exchange After outpacing the CME,FTXs market share of open interest collapsed
193、 in late 2022.CMEs open interest surpassed Binances for the first time.As the demand for more regulated and secure infrastructure increased following the contagion in 2022,bitcoins market dynamics shifted more to the US.*Short for Chicago Mercantile Exchange.Sources:ARK Investment Management LLC,202
194、4,based on data from Glassnode as of December 31,2023.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security or cryptocurrency.Past performance is not indicative of future results.$0$1$2$3$4$5Sep 22Oct 22Nov 22De
195、c 22Jan 23Feb 23Mar 23Apr 23May 23Jun 23Jul 23Aug 23Sep 23Oct 23Nov 23Dec 23$BillionsBitcoin Futures Open Interest Hit a Record$4.5 Billion on the CMECMEBinanceFTXBITCOIN IN 202351Bitcoin Is Evolving Into A Reliable Risk-Off AssetBitcoin operates on a decentralized network,independent of any single
196、entity,government,or central bank.Its distributed,open-source nature protects it against arbitrary asset seizure and counterparty risk.Despite its short-term volatility,bitcoin has delivered significant long-term price appreciation.By design,scarcity increases the probability of capital preservation
197、.Bitcoins historically low correlation with traditional asset classes is increasing its role as a source of diversification.Adding a non-correlated asset to portfolios potentially increases returns per unit of risk and provides a buffer against market downturns.Global investors can access and trade
198、bitcoin 24/7,which is increasingly important in times of risk-off uncertainty.Bitcoins supply will be capped at 21 million coins.As with gold,scarcity characterizes bitcoins role as a safe-haven asset.Safety&Capital PreservationDiversificationLong-TermInvestment HorizonLiquidity&AccessibilityInflati
199、on HedgeWith increasing macroeconomic uncertainty and less trust in traditional”flights to safety,”bitcoin has become a viable alternative.Evaluating Bitcoin As A Risk-Off AssetSources:ARK Investment Management LLC,2024.Information as of December 31,2023.For informational purposes only and should no
200、t be considered investment advice or a recommendation to buy,sell,or hold any particular security or cryptocurrency.Past performance is not indicative of future results.BITCOIN IN 202352Major Catalysts Await Bitcoin In 2024Bitcoin HalvingThe Bitcoin halving occurs approximately every 4 years,cutting
201、 the reward for mining new bitcoin blocks in half.Historically,each halving event has coincided with the beginnings of a bull market.Expected in April 2024,this halving will reduce bitcoins inflation rate from 1.8%to 0.9%.Bitcoin Spot ETF LaunchOn January 11,2024,the launch of spot bitcoin ETFs set
202、the stage for Bitcoins growth,by offering investors a more direct,regulated,and liquid way to gain exposure.Bitcoin spot ETFs are traded on major stock exchanges,allowing investors to buy and sell shares through their existing brokerage accounts,and should reduce the learning curve and operational c
203、omplexities associated with direct investments in bitcoin.Institutional AcceptanceThanks to its continued resilience and performance,the shift in perception of bitcoinfrom a speculative instrument to a strategic investment in a diversified portfolioshould characterize its evolution in 2024.Exemplify
204、ing this evolution,Larry Fink,CEO of BlackRock,has shifted his stance from bitcoin skepticism to its potential as a flight to quality.Regulatory DevelopmentsThe bankruptcies of FTX and Celsius have advanced the push for more transparent and open global crypto regulation,including the potential passa
205、ge of a US bill establishing a regulatory framework for cryptocurrencies,and the implementation of Europes Markets in Crypto-Assets(MiCA)regulation,which mandates licensing for crypto wallet providers and exchanges in the EU.0520212027Units Of Bitcoin(Million)Bitcoins Circulating SupplyBT
206、C Supply(units)BTC Supply CapExpected BitcoinIssuanceSources:ARK Investment Management LLC,2024,based on data from Glassnode as of December 31,2023.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommend
207、ation to buy,sell,or hold any particular security or cryptocurrency.Past performance is not indicative of future results.BITCOIN IN 20235353SmartContractsPowering The Internet-NativeFinancial SystemSources:ARK Investment Management LLC,2024.Forecasts are inherently limited and cannot be relied upon.
208、For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.Research By:Frank DowningDirector of Research,Next Generation InternetBIG IDEAS 202454Deployed on public
209、 blockchains,smart contracts offer a global,automated,and auditable alternative to rent-seeking intermediaries and legacy financial infrastructure.In the aftermath of the“crypto crisis”in 2022,several digital asset solutions gained traction,including stablecoins,tokenized treasury funds,and scaling
210、technologies.According to ARKs research,as the value of on-chain financial assets increases,the market value associated with decentralized applications could scale 32%at an annual rate,from$775 billion in 2023 to$5.2 trillion in 2030.Public blockchains are digital asset ledgers openly available for
211、participants to access and are not controlled by a single entity.Smart Contracts are programs that exist on a blockchain and execute computer code when specific conditions are met.Sources for stablecoin usage,treasury issuance,and core development are provided in the corresponding slides that follow
212、.Sources:ARK Investment Management LLC,2024,based on a range of external sources,which are available upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any part
213、icular security.Past performance is not indicative of future results.SMART CONTRACTS55Smart Contracts Are The Foundation Of The Internet Financial SystemIn their infancy,smart contracts are powering a novel financial system that is native to the internet.Ignited by Ethereum,the largest smart contrac
214、t blockchain,multiple networks are supporting on-chain activity and vying for market share.NOTE:Networks represented are smart contract Layer 1 blockchains with$10 million in 2023 transaction fees.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data fro
215、m external sources,which are available upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future r
216、esults.Transaction FeesTop 6 Smart Contract Networks,2023EthereumTronBNB ChainAvalancheSolanaPolygon PoSSmart ContractNetworkMarket Value2023e Price Performance2023Ethereum$274 billion+90%BNB Chain$49 billion+28%Solana$44 billion+924%Avalanche$14 billion+254%Tron$9 billion+120%Polygon PoS$9 billion+
217、28%$3.7BillionSMART CONTRACTS56Stablecoins Highlight The Value Proposition Of Smart ContractsGiven hyperinflation in emerging markets and an increase in global instability,the demand for stablecoins offering digital access to the US dollar is soaring.During the past three years,the number of daily a
218、ctive stablecoin addresses globally has increased at an annual rate of 93%,from 171 thousand to 1.2 million.In 2023,stablecoin transfer volumes surpassed those of Mastercard.$2$9$10$15$0$2$4$6$8$10$12$14$16PaypalMastercardStablecoinsVisaTotal Transfer Volume,2023(Trillions)NOTE:Stablecoin Daily Acti
219、ve Addresses are averaged for each month displayed in chart.Transfer volume estimates are used where Q4 2023 data is not yet available at time of publication.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which are available
220、upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.USD-200,000 400,000 600,000 800,
221、000 1,000,000 1,200,000 1,400,000Jan-21Apr-21Jul-21Oct-21Jan-22Apr-22Jul-22Oct-22Jan-23Apr-23Jul-23Oct-23Stablecoin Daily Active AddressesTronBNB ChainEthereumAvalanche C-ChainPolygon PoSSolanaETH L2sSMART CONTRACTS57Traditional Financial Assets Are Moving On-ChainTokenization allows treasurers to t
222、rack,trade,and collateralize funds more easily on public blockchains than in traditionalfinancial markets.In 2023,tokenized treasury funds jumped more than 7-fold to$850 million.Early funds launched on the Stellar blockchain,but Ethereum became the largest market for tokenized treasuries in 2023.$-$
223、100$200$300$400$500$600$700$800$9001-Jan-231-Feb-231-Mar-231-Apr-231-May-231-Jun-231-Jul-231-Aug-231-Sep-231-Oct-231-Nov-231-Dec-231-Jan-24MillionsValue Of Tokenized Treasury FundsStellarEthereumPolygonSolanaSources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlyi
224、ng data from external sources,which are available upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative
225、of future results.SMART CONTRACTS58PoS MergeWithdrawals Enabled-5 10 15 20 25 30 35Dec-20Mar-21Jun-21Sep-21Dec-21Mar-22Jun-22Sep-22Dec-22Mar-23Jun-23Sep-23Dec-23Number Of Ether(Millions)Staked Ether LaunchFailed validator upgradeBlock propogation bugTransaction spamTransaction spamSoftware bugNode m
226、isconfigurationDeduplication error0500300350Mar-20Jun-20Sep-20Dec-20Mar-21Jun-21Sep-21Dec-21Mar-22Jun-22Sep-22Dec-22Mar-23Jun-23Sep-23Dec-23Number Of DaysSolana Network UptimeDevelopers Refined Protocols During The Bear MarketIn the face of crises and their aftermath in 2022,core develope
227、rs advanced technical roadmaps and hardened protocols to support the next bull market.Ethereum moved successfully to Proof-of-Stake(PoS)*consensus,and Solana hit a new record for continuous uptime.*Proof-of-Stake is a method of securing public blockchains,in which network participants who wish to va
228、lidate transactions on the network pledge or“stake”their assets at risk of loss if they fail to operate within the networks rules.Chart data end 12/31/23.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which are available upon
229、 request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.SMART CONTRACTS59Layer 2 Networks Ha
230、ve Scaled Transactions In The Ethereum EcosystemSince early 2021,more than 20 Layer 2(L2)*networks have launched,enabling Ethereum to scale average daily transactions at lower fees by 4x.Despite their early success,most L2 networks are controlled centrally.The proliferation of L2s has complicated us
231、er and developer experiences.*L2 networks aggregate transactions and settle the resulting state changes to a base-layer smart contract network like Ethereum,typically at higher throughput and lower cost compared to the base network.L2 transaction count is based on data available on Artemis Dashboard
232、:Arbitrum,Base,Linea,Optimism,Polygon zkEVM,Scroll,StarkNet,zkSync Era,Zora Network.Chart data end 12/31/23.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which are available upon request.Forecasts are inherently limited and
233、cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.012345Qtr12019Qtr2Qtr3Qtr4Qtr12020Qtr2Qtr3Qtr4Qtr12021Qtr2Qtr3Qtr4Qtr12022Qtr2Qtr3
234、Qtr4Qtr12023Qtr2Qtr3Qtr4Transaction Count(Millions)Average Daily TransactionsEthereum MainnetLayer 2 NeworksSMART CONTRACTS60Lower Costs Are Boosting On-Chain Engagement As transaction costs have declined,on-chain engagementas measured by the ratio of daily active addresses(DAUs)to monthly active ad
235、dresses(MAUs)has increased.Note:DAU/MAU traditionally refers to a measure of unique users.For this analysis,we are using a measure of unique addresses as an approximation for users.They are correlated but not equivalent.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range
236、 of underlying data from external sources,which are available upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not
237、 indicative of future results.$6.28$0.27$0.13$0.16$0.003$0.28$0.00$0.01$0.10$1.00$10.000%2%4%6%8%10%12%EthereumOptimismArbitrumBaseSolanazkSyncAverage Fee(2023)DAU As A Percent Of MAUEngagement Relative To Transaction FeesDAU/MAUSMART CONTRACTS61Monolithic Chains Like Solana Offer An Alternative To
238、Vertical ScalingSmart contract network designs offer tradeoffs.By prioritizing base-layer decentralization,the Ethereum ecosystem became more complex as it scaled.By prioritizing scalability in a single layer,Solana maintained a simple architecture for users and app developers and has gained tractio
239、n.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which are available upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a
240、recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.L1 EthereumL1Solana+Simplifies the environment for developers and users-Raises L1 validation costs+Maximizes composability and interoperability-Potentially requires L2s to maximally scale+
241、Lowers fees and increases throughput for base layer transactions-Requires apps depend L1 execution environmentL2OptimismL2BaseDAPP 1DAPP 2DAPP 3DAPP 1DAPP 2DAPP 3+Minimizes L1 validation cost-Requires asset bridging between L1 and L2,fragmenting liquidity+Supports multiple approaches for scaling,enc
242、ouraging flexibility&innovation-Increases complexity for developers and users+Leverages the network effect and liquidity advantages of Ethereum mainnet-Introduces additional reliability and security considerations across L2sL2ArbitrumVertical ScalingHorizontal ScalingSMART CONTRACTS62Smart Contracts
243、 Could Collapse The Cost Of Financial ServicesSources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which are available upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and shou
244、ld not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.20002020Global Financial Services Revenue20002020TrillionsValue Of Global Financial Assets 3.6x3.3x$510T$140T$17T$5T3.3%agg.take rateActivityCos
245、t To CurrentSystemCost LoweringSolutionKnow Your Customer Verification$1,500-$3,000+per individual verificationUnified digital identity verifiable across institutionsNasdaq Listing Fee$270k per listing+$52-$180k annuallyDirect DEX listing with global distributionGlobal Anti-Money Laundering Complian
246、ce$274 billion cost to the global financial system annuallyAuditable provenance of funds on global ledgerEconomic Impact of Financial Regulatory ComplianceSMART CONTRACTSThe value of financial assets globally ballooned from$140 trillion in 2000 to$510 trillion in 2020,thanks to a combination ofgloba
247、l economic growth,increased financialization,and expanding equity multiples.The operating cost of the global financial system increased in tandem with the value of financial assets.At$20 trillion in total annual revenue,the aggregate financial services industrys take rate has been 3.3%relative to th
248、e value of all financial assets.Smart contracts could lower this drag on the economy materially.63Smart Contract Networks Could Generate Fees Of$450 Billion In 2030$775$5,300$-$1,000$2,000$3,000$4,000$5,000$6,00020232030EstimateTrillionsSmart Contract Protocol Market Value37%CAGR$1$20$11$8$450$-$50$
249、100$150$200$250$300$350$400$450$500202020224202520262027202820292030BillionsGross Smart Contract Fee RevenueActualsForecast78%CAGR1SMART CONTRACTSSmart contracts could facilitate the origination,ownership,and management of on-chain assets for a fraction of traditional financial costs.If f
250、inancial assets were to migrate to blockchain infrastructure at a rate similar to the adoption of the internet,and the take rates associated with decentralized financial services were a third those of traditional financial services,smart contracts could generate annual fees of more than$450 billion
251、and create more than$5 trillion in market value,increasing at compound annual rates of 78%and 32%,respectively,through 2030.12020-2021 data approximated using top 20 all-time fee generating protocols from Token Terminal Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range
252、 of underlying data from external sources,which are available upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not
253、 indicative of future results.6464Digital ConsumersSources:ARK Investment Management LLC,2024.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past
254、 performance is not indicative of future results.Transitioning Toward Digital LeisureBIG IDEAS 2024Nicholas Grous Associate Portfolio Manager Andrew Kim Analyst Research By:65Connected TV(CTV)Advertising should grow 17%at a compound annual rate,from$25 billion in 2023 to$73 billion in 2030.Social Co
255、mmerce should grow 32%at an annual rate,from$730 billion today to over$5 trillion in 2030.Sports Betting should remain turbocharged by the legalization of online/mobile betting.AI-assisted Video Game Creation is the new wave in gaming,building on user-generated platforms like Roblox,which has hosted
256、 more than 470 million experiences globally52x the combined number of PC,consoles,and mobile games.AI-enabled Hardware could redefine personal wearable computing,especially if virtual reality(VR)continues to face challenges.According to ARKs research,spending on digital leisure should take share fro
257、m physical options and grow 19%at an annual rate during the next seven years,from$7 trillion in 2023 to$23 trillion in 2030.Several trends are accelerating the shift to digital leisure:DIGITAL CONSUMERSSources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying dat
258、a from external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of f
259、uture results.66Artificial Intelligence Could Lower The Average Workweek And Stimulate Digital ConsumptionDuring the 80 years between the Second Industrial Revolution through the end of World War II,labor hours per worker decreased 0.5%at an annual rate globally.Generative AI could lower labor hours
260、 per worker by 1.3%on average,from 5.0 hours per day in 2022 to 4.5 hours in 2030.*As a result,consumers might devote more time to online entertainment,potentially increasing the share of total waking hoursspent online from 40%in 2023 to 49%in 2030.*To calculate global daily working hours,we divide
261、total annual hours of labor per worker by the total days of the year.*The chart illustrating daily allocation of online vs.offline time captures total daily waking hours,including those allocated to labor or education.The chart captures hours generated by internet users only.Sources:ARK Investment M
262、anagement LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,se
263、ll,or hold any particular security.Past performance is not indicative of future results.0.01.02.03.04.05.06.07.08.09.010.0019902030Daily HoursGlobal Labor Hours Per Worker Per Day*2nd Industrial RevolutionGenerative AI Revolution-0.5%per year-1.3%per year31%39%49%0%20%40%60%80%100%2010202
264、02030Share Of Waking HoursGlobal Online And Offline Time*OnlineOfflineForecastForecastDIGITAL CONSUMERS67Streaming TV Is Displacing Linear TVIn just two years,streamings share of overall TV consumption increased more than 10 percentage points to 39%as of July 2023,surpassing the shares of cable and
265、broadcast,respectively.Ad spend on connected TV(CTV)is following eyeballs and is likely to grow 17%in real terms at an annual rate,from$25 billion in 2023 to$73 billion in 2030.If so,ad spend on CTV should surpass that on linear TV in 2027.*The share of streaming,cable,and broadcast do not add up to
266、 100%,as we exclude the portion of consumption that Nielsen categorizes as“Other,”which includes time spent on unmeasured sources like video-on-demand(VOD),audio streaming,gaming,and other device use.*We define linear TV as traditional TV delivered via cable,satellite,or over-the-air.We define conne
267、cted TV as streamed TV delivered over-the-top through smart TVs,streaming media devices,video game consoles,and other modern hardware.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,including Nielsen,Insider Intelligence,and M
268、AGNA Global,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future result
269、s.$-$20$40$60$80203E24E25E26E27E28E29E30E$BillionsUS TV Ad Spend*Connected TVLinear TV0%5%10%15%20%25%30%35%40%45%May-21Jul-21Sep-21Nov-21Jan-22Mar-22May-22Jul-22Sep-22Nov-22Jan-23Mar-23May-23Jul-23Sep-23Nov-23Share Of US TV TimeUS TV Viewership Share*StreamingCableBroadcastDec-23DIGITAL
270、CONSUMERS68Pinterest0.4BSocial Commerce Merchants Can Sell to AnyoneAnytime,AnywhereSocial media platforms are increasing their monetization of global audiences with e-commerce.Thanks to omnichannel solutionsboth physical and digitalsocial commerce could grow 32%at an annual rate,from$730 billion to
271、day to over$5 trillion in 2030.YouTube2.8B MAUs*Instagram1.5BTikTok1.1BShopifyBusiness Launch Omnichannel Selling Payment Processing MarketingAnalytics and Management Logistics&Shipping Business Funding*We estimate each platforms monthly active users(MAUs)across its iOS and Android mobile apps.Sourc
272、es:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,including Sensor Tower,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered inves
273、tment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.Facebook2.2BWhatsApp1.8B0%5%10%15%20%25%30%$-$2$4$6$8$10$12$14$16$182018 1920212223E 24E 25E 26E 27E 28E 29E 30EShare Of E-commerceGross Merchandise Value($Trillions)Glob
274、al Social Commerce SalesARK ForecastSocial Commerce(LHS)Traditional E-commerce(LHS)Social Commerce Share(RHS)DIGITAL CONSUMERS69Mobile Sports Betting Continues To Grow And Consolidate In The USThanks to legalization and consumer adoption,the winners in online sports betting are pulling away from the
275、 pack.As online sports betting surged 35%during 2023,DraftKings and FanDuel offered superior user experiences that helped take share from other sportsbooks.DraftKings and FanDuel grew their share of national deposits to 75%in 2023,while the long tail of sportsbooks lost 8 percentage points of share.
276、Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,including Yipit Data,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered in
277、vestment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.0%25%50%75%100%$-$100$200$300$400$500203E 24E 25E 26E 27E 28E 29E 30E%of Total Volume$BillionsUS Online Sports Betting VolumeLegalized States(LHS)Future Leg
278、alizations(LHS)Online Penetration(RHS)124%CAGR22%CAGR0%20%40%60%80%100%Jan-20Apr-20Jul-20Oct-20Jan-21Apr-21Jul-21Oct-21Jan-22Apr-22Jul-22Oct-22Jan-23Apr-23Jul-23Oct-23Sportsbook ShareNational Deposit Market ShareDraftKings and FanDuelAll Other SportsbooksDIGITAL CONSUMERS70$0.0016$0.07$0.15$0.23$1.4
279、0$0.001$0.010$0.100$1.000$10.000Net Monetization Per HourGross Platform Monetization Rates*Text-based AIStreamed AudioVideo GamesStreamed VideoDating AppsOnline Experiences Are Becoming More Immersive And MonetizableHistory suggests that deeper immersion leads to higher monetization.After computer g
280、raphics expanded the market beyond text-based adventure games in the 1980s,gaming revenue soared 19%at an annual rate,from$6 billion in 1985 to$24 billion in 1993.Now,multimodal AItext,images,audio,and videoare creating more immersive and interactive experiences that should expand the market.*”Text
281、games”refer to both text-based and spreadsheet-based games.“All other games”exclude arcade game releases.Gaming revenue captures PC and console gaming revenue only*We estimate various platforms ability to monetize on direct consumer spend only.*Revenue figures have been inflation-adjusted to 2023 US
282、 Dollars.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment adv
283、ice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.$-$5$10$15$20$25$30$35$40$450%20%40%60%80%100%385878990305Gaming Revenue($2023 Billions)*Share Of Games ReleasedVideo Games Evolution*Text Games(LHS)All Othe
284、r Games(LHS)Gaming Revenue(RHS)$10$1$0.1$0.01$0.001DIGITAL CONSUMERS71Atari introduces the Atari 2600.Apple launches the first iPhone.Generative AI lowers the cost of UGC.1.E-041.E-031.E-021.E-011.E+001.E+011.E+021.E+031975 7889902 05 0811141720Cumulative Number(Log Scale)Number Of Video
285、Game ReleasesTraditional GamesMobile GamesRoblox Experiences*We normalize the cost of 3D asset generation by each models CLIP R-Precision scores.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,including Nichol et al.2022,which
286、 may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.Thanks To AI-As
287、sisted Creation,Gamers Could Become DevelopersAI-assisted game creation on user-generated content(UGC)platforms could cause an explosion in gaming content.According to our research,after normalizing for output quality,the cost of generating a single 3D asset has dropped 99%at an annual rate to less
288、than$0.06 since 2021.AI should democratize content creation and accelerate the growth in UGC.Roblox already has delivered more than 470 million experiences globally,52x the combined number of PC,console,and mobile app games.$1E-02$1E-01$1E+00$1E+01$1E+02$1E+03Oct-21Feb-22May-22Aug-22Dec-22Mar-23Doll
289、ar Cost per 3D AssetDate of PublicationCost Decline In Generative AI For 3D Assets*99%Annualized Cost Decline2259$1,000$100$10$1$0.1$0.01(Log Scale)DIGITAL CONSUMERS72The Market For Virtual Reality Is NascentDespite significant headset improvements,including Apples Vision Pro,d
290、evelopers have not flocked to support virtual reality(VR).Without compelling use cases,adoption has been slow.Meta Quest,for example,is offering only 2,200 appsa fraction of the 553,000 the iPhone boasted five years after its launch.As a result,Meta has sold only 27 million Quest units,18%of the 146
291、 million iPhones Apple sold cumulatively five years after launch.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational p
292、urposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.020406080012345Cumulative Units(Millions)Years After LaunchApple iPhone vs.Meta QuestHeadset ShipmentsApple iPho
293、neMeta Quest01101001,00012345Units(Thousands)(Log Scale)Years After LaunchiOS vs.Meta Quest AppsiOS AppsMeta Quest AppsDIGITAL CONSUMERS73AI Could Redefine Personal ComputingAs computing transitioned,hardware cycles compressed from 35 years for the personal computer to 20 years for the smartphone,ca
294、using the consolidation of software players.More rapid adoption of AI-enabled hardware could accelerate the consolidation of software providers.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,including DeGusta 2012,which may b
295、e provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.20 35 010203040AI-Ena
296、bledHardwareSmartphoneComputerYearsTime To Penetrate 75%Of The US Population?Software ExpansionOperating System ConsolidationWindows,Commodore 64 OS,Atari TOS,Amiga,Linux,MacOSWindows,MacOS,LinuxBB OS,Windows Phone,iOS,Symbian,Android,Palm OSiOS,Android?GPT-4,Claude 2,Mixtral,Llama 2,GrokDIGITAL CON
297、SUMERS74Digital Consumption Is Outpacing Economic GrowthGlobally,consumers spent 20%of their$34 trillion leisure budget on digitally-facilitated goods and services in 2023.Based on the shift toward digital leisure,real digital revenue*could increase 16%at an annual rate during the next seven years,f
298、rom$1.8 trillion to$5 trillion,and account for 43%of all leisure spending in 2030.*We define digital platform revenue as the gross revenues of US sports betting,global video game software and services,global digital video,and global digital audio.We also include net e-commerce platform revenue and n
299、et NFT creator fees and platform revenue on a global basis.*Direct includes spend across e-commerce,video game software,digital video,digital audio,NFTs,and US mobile sports betting.Indirect includes spend across all digital ads.Sources:ARK Investment Management LLC,2024.This ARK analysis is based o
300、n a range of underlying data from external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past perform
301、ance is not indicative of future results.0%5%10%15%20%25%30%35%40%45%50%$-$5$10$15$20$252018 19202122 23E 24E 25E 26E 27E 28E 29E 30EShare of Leisure Spend$TrillionsGlobal Digital Leisure Spend*Direct(LHS)Indirect(LHS)Share of Total Leisure Spend(RHS)$-$1$2$3$4$5$6203E24E25E26E27E28E29E30
302、E$TrillionsGlobal Digital Platform RevenueDirectIndirectDIGITAL CONSUMERS7575DigitalWalletsSources:ARK Investment Management LLC,2024.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,s
303、ell,or hold any particular security.Past performance is not indicative of future results.Closing The Loop With Two-Sided NetworksAndrew KimAnalystResearch By:BIG IDEAS 202476Vertical software refers to a suite of solutions tailored to the needs of specific industries.Leading vertical software platfo
304、rms are expanding rapidly into financial services for consumers and merchants.With two-sided networks,such software could facilitate closed loop transactions from consumer to merchant,merchant to employee,and employee to merchant.ARK believes that digital wallets on these platforms will enable fully
305、 closed payment ecosystems.Block,Shopify,and Toast are compelling platforms likely to use digital wallets as the nucleus of their consumer,merchant,and employee ecosystems.According to our research,closed loop consumer payments,merchant banking,and employee payroll/payments could increase their reve
306、nues by 22-33%at an annual rate during the next seven years,from$7 billion in 2023 to$27-$50 billion in 2030.*In this exercise,we forecast the core revenues of Block,Shopify,and Toast to grow 22%at an annual rate over the next seven years.Summing the mentioned revenue opportunities on top of our cor
307、e revenues forecast increases the annual growth rate from 22%to 33%over the next seven years.We primarily model Blocks historical revenue and future revenue opportunity across its Square merchant ecosystem and do not incorporate Cash App or Afterpay revenues independent of Square.Sources:ARK Investm
308、ent Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to b
309、uy,sell,or hold any particular security.Past performance is not indicative of future results.DIGITAL WALLETS77Vertical Software Platforms Are Consolidating Financial ServicesIn addition to enabling core business operations,vertical software providers like Block,Shopify,and Toast are consolidating fi
310、nancial services for merchants.With digital wallets at their core,and partnering with sponsor banks and fintech companies oractivating their own banking charters,vertical platforms should eliminate myriad merchant interactions with less efficient legacy financial institutions.Savings*We consider Blo
311、cks Cash App and Toasts MyToast mobile app as consumer digital wallets,and we consider Shopifys Shop mobile app and Toasts Toast Takeout mobile app as digital wallets in their early stages.*We consider Shopify Balance as both a checking and savings vehicle for merchants.*Given xtraCHEFs invoice auto
312、mation features,we believe Toast will soon offer direct bill pay on its platform.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For
313、 informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.ToastShopifyBlockChecking*Debit CardCredit CardWorking Capital FinancingPayrollBill Pay*Digital WalletConsu
314、mer*MerchantDIGITAL WALLETS78Vertical Software Platforms Are Consolidating Consumer ServicesVertical software platforms are not only enabling vast merchant networks but also building consumer networks using digital wallets.By scaling merchant and consumer networks simultaneously,vertical software pl
315、atforms are becoming operating systemsfor these two-sided networks.*We consider Blocks Cash App and Toasts MyToast mobile app as consumer digital wallets,and we consider Shopifys Shop mobile app and Toasts Toast Takeout mobile app as digital wallets in their early stages.*We consider the Toast Pay C
316、ard a form of consumer debit cards.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be
317、 considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.Debit Card*ToastShopifyBlockSavingsBuy Now Pay LaterPersonal LendingCheckingDigital WalletE-commercePaymentsConsumer*MerchantLoyaltyDIGITAL WALLETS79Two-
318、Sided Networks Can Close The Financial Loop Between Consumers And MerchantsClosed-loop payment ecosystems incorporate in-network money transfers in three ways:from consumers to merchants,from merchants to employees,and from employeescum consumersto merchants.To build these payment ecosystems,platfor
319、ms must have:1)large and engaged two-sided networks,2)end-to-end visibility over merchant operations and finances,and 3)vertical industry expertise.Sources:ARK Investment Management LLC,2024.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be
320、considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.Vertical Software PlatformConsumerMerchantsEmployeesIn-Network Consumer Digital WalletIn-Network Merchant Digital WalletIn-Network Payroll Service Provide
321、r$DIGITAL WALLETS80Steps:3PSP Take Rate:2.3%Steps:9PSP Take Rate:1.1%Digital Wallets Are Likely To Disintermediate Consumer-To-Business(C2B)Payment EcosystemsTransactions funded with digital wallet balances bypass banks and card networks,saving interchange fees for payment facilitators,merchants,and
322、 consumers.In ARKs view,vertical software platforms with scaled consumer and merchant ecosystems will leverage digital wallets to facilitate closed-loop transactions.*Payment processes and associated fee estimates are rendered for illustrative purposes only.Sources:ARK Investment Management LLC,2024
323、.This ARK analysis is based on a range of underlying data from external sources,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any par
324、ticular security.Past performance is not indicative of future results.BEFORE:C2B Card Payment AuthorizationAFTER:C2B Closed-Loop Payment AuthorizationConsumerPayment Service ProviderIssuing BankCard NetworkMerchantAcquiring Bank123456789Issuing BankCard NetworkAcquiring BankConsumer Digital WalletPa
325、yment Service ProviderMerchantIn-NetworkIn-Network23FundingSource1Transaction AuthorizationTransaction SettlementDIGITAL WALLETS81Closed-Loop Payment Volume In The US Could Increase 24-Fold By 2030According to ARKs research,total C2B digital wallet payments should increase 20%at an annual rate durin
326、g the next seven years,from$2 trillion in 2023 to$7 trillion in 2030.As a percent of the total,closed-loop payments should increase from 4%to 25%,taking the payments revenue forecast for Blocks Square,Shopify,and Toast from$3.5 billion to$21 billion,a 29%annualized rate of gain.*We define closed-loo
327、p transactions as any consumer-to-buyer(C2B)digital wallet transaction that does not involve third-party issuers or card networks except for digital wallet balance top-ups.*Closed-loop payment revenue is represented on a gross basis and will be shared between the software platforms and all enabling
328、market participants such as other fintechs or sponsoring financial institutions.We primarily model Blocks historical revenue and future revenue opportunity across its Square merchant ecosystem and do not incorporate Cash App or Afterpay revenues independent of Square.Sources:ARK Investment Managemen
329、t LLC,2024.This ARK analysis is based on a range of underlying data from external sources,including Worldplay 2019,2020,2021,2022,2023,which may be provided upon request.Forecasts are inherently limited and cannot be relied upon.For informational purposes only and should not be considered investment
330、 advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.0%5%10%15%20%25%30%$-$1$2$3$4$5$6$7$82018 19202122 23E 24E 25E 26E 27E 28E 29E 30EPercent of Personal ConsumptionC2B Volume($Trillions)US Digital Wallet Transaction Volume*Ope
331、n LoopClosed Loop%of PCE(RHS)20%CAGR$-$5$10$15$20$25201920212223E24E25E26E27E28E29E30EC2B Payment Revenue($Billions)US Payment Revenue For Blocks Square,Shopify,And ToastNet Payment RevenueClosed Loop Payment Revenue Opportunity29%CAGRDIGITAL WALLETS82Steps:3+2=5PSP Take Rate:2.3%2.4%Steps:9+7=16PSP
332、 Take Rate:1.1%*Payment processes and associated fee estimates are rendered for illustrative purposes only.Sources:ARK Investment Management LLC,2024.This ARK analysis is based on a range of underlying data from external sources,which may be provided upon request.Forecasts are inherently limited and
333、 cannot be relied upon.For informational purposes only and should not be considered investment advice or a recommendation to buy,sell,or hold any particular security.Past performance is not indicative of future results.BEFORE:Status Quo Merchant SettlementAFTER:Closed-Loop Merchant SettlementMerchantPayment Service ProviderConsumer Bank AccountCard NetworkMerchant Bank AccountAcquiring Bank10Issui