Internet Trends 2018

INTERNET TRENDS 2018 Mary Meeker May 30 @ Code 2018 kleinerperkins.com/InternetTrends

Thanks Kleiner Perkins Partners Ansel Parikh & Michael Brogan helped steer ideas and did a lot of heavy lifting. Other contributors include: Daegwon Chae, Mood Rowghani, Eric Feng (E-Commerce) & Noah Knauf (Healthcare). In addition, Bing Gordon, Ted Schlein, Ilya Fushman, Mamoon Hamid, Juliet deBaubigny, John Doerr, Bucky Moore, Josh Coyne, Lucas Swisher, Everett Randle & Amanda Duckworth were more than on call with help. Hillhouse Capital Liang Wu & colleagues’ contribution of the China section provides an overview of the world’s largest market of Internet users. Participants in Evolution of Internet Connectivity From creators to consumers who keep us on our toes 24x7 + the people who directly help us prepare the report. And, Kara & team, thanks for continuing to do what you do so well. 2

Context We use data to tell stories of business-related trends we focus on. We hope others take the ideas, build on them & make them better. At 3.6B, the number of Internet users has surpassed half the world’s population. When markets reach mainstream, new growth gets harder to find - evinced by 0% new smartphone unit shipment growth in 2017. Internet usage growth is solid while many believe it’s higher than it should be. Reality is the dynamics of global innovation & competition are driving product improvements, which, in turn, are driving usage & monetization. Many usability improvements are based on data - collected during the taps / clicks / movements of mobile device users. This creates a privacy paradox... Internet Companies continue to make low-priced services better, in part, from user data. Internet Users continue to increase time spent on Internet services based on perceived value. Regulators want to ensure user data is not used ‘improperly.’ Scrutiny is rising on all sides - users / businesses / regulators. Technology-driven trends are changing so rapidly that it’s rare when one side fully understands the other...setting the stage for reactions that can have unintended consequences. And, not all countries & actors look at the issues through the same lens. We focus on trends around data + personalization; high relative levels of tech company R&D + Capex Spending; E-Commerce innovation + revenue acceleration; ways in which the Internet is helping consumers contain expenses + drive income (via on-demand work) + find learning opportunities. We review the consumerization of enterprise software and, lastly, we focus on China’s rising intensity & leadership in Internet-related markets. 3

INTERNET DEVICES + USERS = GROWTH CONTINUES TO SLOW 5

Global New Smartphone Unit Shipments = No Growth @ 0% vs. +2% Y/Y New Smartphone Unit Shipments vs. Y/Y Growth 1.5B 90% al b Glo ts, n 1.0B 60% h t me w p i o Sh Gr e Y n o Y/ h 0.5B 30% artp Sm ew N 0 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 Android iOS Other Y/Y Growth Source: Katy Huberty @ Morgan Stanley (3/18), IDC. 6

Global Internet Users = Slowing Growth @ +7% vs. +12% Y/Y Internet Users vs. Y/Y Growth 4B 16% 3B 12% al ob h Gl t w s, o 2B 8% Gr User Y Y/ net nter I 1B 4% 0 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 Global Internet Users Y/Y Growth Source: United Nations / International Telecommunications Union, USA Census Bureau. Internet user data is as of mid-year. Internet user data: Pew Research (USA), China Internet Network Information Center (China), Islamic Republic News Agency / InternetWorldStats / KP 7 estimates (Iran), KP estimates based on IAMAI data (India), & APJII (Indonesia). Note: Historical data (particularly in Sub-Saharan Africa) revised by ITU in 2017 to better account for dual-SIM subscriptions (i.e. two Internet subscriptions per single smartphone user).

Global Internet Users = 3.6B @ >50% of Population (2018) Internet Penetration 60% 49% al b o Gl , 40% n o ati etr 24% Pen et n 20% ter n I 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 Source: CIA World Factbook, United Nations / International Telecommunications Union, USA Census Bureau. Internet user data is as of mid-year. Internet user data: Pew Research (USA), China Internet Network Information Center (China), Islamic Republic News Agency / InternetWorldStats 8 / KP estimates (Iran), KP estimates based on IAMAI data (India), & APJII (Indonesia). Note: Historical data (particularly in Sub-Saharan Africa) revised by ITU in 2017 to better account for dual-SIM subscriptions (i.e. two Internet subscriptions per single smartphone user).

Internet Users… Growth Harder to Find After Hitting 50% Market Penetration 9

INTERNET USAGE = GROWTH REMAINS SOLID 10

Digital Media Usage @ +4% Growth... 5.9 Hours per Day (Not Deduped) Daily Hours Spent with Digital Media per Adult User 5.9 6 5.6 5.4 5.1 4.9 5 SA U 4.3 , 3.3 3.1 ay 4 2.8 3.7 D 2.6 2.3 1.6 er 3.2 p 0.8 3.0 t 3 2.7 0.4 0.3 en 0.3 Sp s 2 r u o 2.4 2.6 2.1 2.5 2.2 2.2 2.3 2.3 2.2 H 2.2 1 0.6 0.4 0.4 0.4 0.3 0.3 0.3 0.3 0.3 0.2 0 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Other Connected Devices Desktop / Laptop Mobile Source: eMarketer 9/14 (2008-2010), eMarketer 4/15 (2011-2013), eMarketer 4/17 (2014-2016), eMarketer 10/17 (2017). Note: Other connected devices include OTT and game consoles. Mobile includes smartphone and tablet. Usage includes both home and 11 work for consumers 18+. Non deduped defined as time spent with each medium individually, regardless of multitasking.

Internet Usage… How Much = Too Much? Depends How Time is Spent 12

INNOVATION + COMPETITION = DRIVING PRODUCT IMPROVEMENTS / USEFULNESS / USAGE + SCRUTINY 13

Innovation + Competition = Driving Product Improvements / Usefulness / Usage Devices Access Simplicity Payments Local Messaging Video Voice Personalization 14

Devices = Better / Faster / Cheaper Apple iPhone New Smartphone Shipments –ASP 2016 2017 $500 * al lob G $400 , SP A ‘Portrait’ Photos Face Tracking nt Water Resistant Full Device Display $300 me Wireless Charging Ship Google Android ne 2016 2017 $200 rtpho Sma $100 New Google Assistant ‘Lens’ Smart $0 Image Recognition ‘AI-Assisted’ 2007 2009 2011 2013 2015 2017 Photo Editing Always-On Display Source: Apple, Google, Katy Huberty @ Morgan Stanley, IDC. *ASP Based on Morgan Stanley’s new smartphone shipment breakdown by taking the midpoint of each $50 price band & assuming a $1,250 ASP for smartphones over $1,000. Note: Deloitte estimates that 15 120MM used smartphones were traded in 2016 and 80MM in 2015 which may further reduce smartphone costs to consumers as the ratio of used to new devices rises. Apple 2016 = iPhone 7 Plus, 2017 = iPhone X. Google 2016 = Pixel, 2017 = Pixel 2.

Access = WiFi Adoption Rising WiFi Networks 500MM 400MM al b o Gl 300MM ks, r o etw N 200MM i F i W 100MM 0 2001 2003 2005 2007 2009 2011 2013 2015 2017 Source: WiGLE.net as of 5/29/18. Note: WiGLE.net is a submission-based catalog of wireless networks that has collected >6B data points since launch in 2001. Submissions are not paired 16 with actual people, rather name / password identities which people use to associate their data.

Simplicity = Easy-to-Use Products Becoming Pervasive Messaging Commerce Media Telegram Square Cash Spotify Source: Telegram (5/18), Square (5/18), Spotify (5/18). 17

Payments = Digital Reach Expanding… Transactions by Payment Channel Thinking of your past 10 everyday transactions, how many were made in each of the following ways? In-Store In-Store 40% Other Online 15% Buy Buttons 9% Other Mobile Payments 8% P2P Transfer 7% 60% = Mobile Messenger Apps 7% Digital Other In-App Payments 4% QR Codes 4% Smart Home Device 3% Wearables / Contactless 2% Other 1% 0% 10% 20% 30% 40% 50% % of Global Responses (9/17) Source: Visa Innovations in a Cashless World 2017. Note: Full question was ‘Please think about the payments you make for everyday transactions (excluding rent, mortgage, or other larger, infrequent payments). Thinking of your past 10 everyday transactions, how many were made in each of the following ways?’, GfK 18 Research conducted the survey with n = 9,200 across 16 countries (USA, Canada, UK, France, Poland, Germany, Mexico, Brazil, Argentina, Australia, China, India, Japan, South Korea, Russia, UAE), between 7/27/17 – 9/5/17. All respondents do not work in Financial Services, Marketing, Marketing Research, Advertising, or Public Relations, own and currently use a smartphone, have a savings or checking account; own/use a computer or tablet, and own a credit or debit card.

…Payments = Friction Declining... China Mobile Payment Users 600MM na Chi , s r e400MM Us nt me y a P e 200MM Mobil 0 2012 2013 2014 2015 2016 2017 Source: China Internet Network Information Center (CNNIC). Note: User defined as active user of mobile- passed payment technology for everyday transactions, as well as more complex transactions, such as bill 19 paying in the relevant period. Includes all forms of transactions on mobile (e.g., QR codes, P2P, etc.)

…Payments = Digital Currencies Emerging Coinbase Users ) 4x 17 1/ o t d Indexe ( 3x Global , rs se U ed 2x er t egis R e as 1x oinb C January March May July September November 2017 Source: Coinbase. Note: Registered users defined as users that have an account on Coinbase. 20

Local = Offline Connections Driven by Online Network Effects Nextdoor Active Neighborhoods 200K SA 150K s, U d o o h r o b 100K h eig N e ctiv 50K A 0 2011 2012 2013 2014 2015 2016 2017 2018 Source: Nextdoor (5/18). Note: There are ~130MM households in USA. Nextdoor estimates that there are ~650 households per average neighborhood (~200K USA neighborhoods). 21

Messaging = Extensibility Expanding Messaging Messenger MAUs Tencent (2000  2018) 1.5B QQ WeChat 1.0B 0.5B 0 2011 2012 2013 2014 2015 2016 2017 WhatsApp Facebook Messenger WeChat Instagram Twitter Source: Facebook, WhatsApp, Tencent, Instagram, Twitter, Morgan Stanley Research. Note: 2013 data for Instagram & Facebook Messenger are approximated from statements made in early 2014. Twitter users excludes SMS fast followers. 22 MAUs (Monthly Active Users) are defined as users who log into a messenger on the web or through an application.

Video = Mobile Adoption Climbing... Mobile Video Usage 40 obal Gl , 30 inutes M ing w ie20 o V ide V obile10 M ly Dai 0 2012 2013 2014 2015 2016 2017 2018E Source: Zenith Online Video Forecasts 2017 (7/17). Note: Based on a study across 63 countries. The historical figures are taken from the most reliable third-party sources in each market including Nielsen 23 and comScore. The forecasts are provided by local experts, based on the historical trends, comparisons with the adoption of previous technologies, and their judgement.

…Video = New Content Types Emerging Fortnite Battle Royale Twitch Streaming Hours Most Watched Game on Twitch 18MM obal Gl , urs12MM Ho ming trea S ly 6MM Dai ge ra e v A 0 2012 2013 2014 2015 2016 2017 Source: Twitch (3/18). Note: Tyler “Ninja” Blevins Twitch stream has 7MM+ followers (#1 ranked) as of 5/29/18 based on Social Blade data. 24

Voice = Technology Lift Off… Google Machine Learning Word Accuracy 100% 95% 95% 90% Rate y c ura c c A ord 80% W 70% 2013 2014 2015 2016 2017 Google Threshold for Human Accuracy Source: Google (5/17). Note: Data as of 5/17/17 & refers to recognition accuracy for English language. Word error rate is evaluated using real world search data which is 25 extremely diverse & more error prone than typical human dialogue.

…Voice = Product Lift Off Amazon Echo Installed Base Amazon Echo Skills 30K 40MM 30MM ls USA il20K , k e Bas r of S d 20MM e mbe tall Nu Ins 10K 10MM 0 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 0 2015 2016 2017 2015 2016 2017 2018 Source: Consumer Intelligence Research Partners LLC (Echo install base, 2/18), Various media outlets including Geekwire, TechCrunch, and Wired (Echo skills, 3/18) 26

Innovation + Competition = Driving Product Improvements / Usefulness / Usage Devices Access Simplicity Payments Local Messaging Video Voice Personalization 27

Personalization = Data Improves Engagement + Experiences… Drives Growth + Scrutiny 28

Personal + Collective Data = Provide Better Experiences for Consumers… 2.2B 200MM 170MM 125MM Facebooks Pinterests Spotifys Netflixes Newsfeed Discovery Music Video Source: Facebook (5/18), Pinterest (5/18), Spotify (5/18), Netflix (5/18). Note: Facebook Q1:18 MAU (4/18), Pinterest MAU (9/17), Spotify Q1:18 29 MAU (5/18), Netflix Q1:18 global streaming memberships (4/18).

...Personal + Collective Data = Provide Better Experiences for Consumers 100MM+ 20% 100MM+ 17MM** Waze UberPOOL Share of All Snap Map Nextdoor Drivers Rides, Where Available* MAUs Recommendations Real-Time Real-Time Real-Time Often Real-Time Navigation Transportation Social Stories Local News Source: Facebook (5/18), Waze (2/18), Snap (5/18), Nextdoor (5/18) *Active Markets = Atlanta, Austin, Boston, Chicago, Denver, Las Vegas, Los Angeles, Miami, Nashville, New Jersey, New York City, Philadelphia, Portland, San Diego, San Francisco, Seattle, Washington D.C., Toronto, Rio de 30 Janeiro, Sao Paulo, Bogota, Guadalajara, Mexico City, Monterrey, Lima, Paris, London, Ahmedabad, Bangalore, Chandigarh, Chennai, New Delhi, Guwahati, Hyderabad, Jaipur, Kochi, Kolkata, Mumbai, Pune, & Sydney. **Refers to cumulative recommendations as of 11/17.

Privacy Paradox Internet Companies Making Low-Priced Services Better, in Part, from User Data Internet Users Increasing Time on Internet Services Based on Perceived Value Regulators Want to Ensure User Data is Not Used ‘Improperly’ 31

Rising User Engagement = Drives Monetization + Investment in Product Improvements... Facebook Annualized Revenue per Daily User $40 al b o e Gl u $34 ), en U $30 A ev R PD e R ag (A er $20 v ser A U e ed $16 v z i cti al u A $10 n y l n A ai D er p $0 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 2015 2016 2017 2018 Source: Facebook (4/18). Note: Facebook Daily Active Users (DAU) defined as a registered Facebook user who logged in and visited Facebook on desktop or mobile device, or took action to share content or activity with his or 32 her Facebook friends or connections via a third-party website that is integrated with Facebook, on a given day. ARPDAU calculated by dividing annualized total revenue by average DAU in the quarter.

...Rising Monetization + Data Collection = Drives Regulatory Scrutiny Data / Privacy Competition The European Data Protection Regulation will be Commission fines Google €2.42 billion for abusing applicable as of May 25th, 2018 in all member states dominance as search engine by giving illegal to harmonize data privacy laws across Europe. advantage to its own comparison shopping service. - European Union, 5/18 - European Commission, 6/17 Facebook’s collection & use of data from Commission approves acquisition of LinkedIn by third-party sources is abusive. Microsoft, subject to conditions. - German Federal Cartel Office, 12/17 - European Commission, 12/16 Safety / Content Taxes The Germany Network Enforcement Act will require Commissionfinds Luxembourg gave illegal tax for-profit social networks with >2MM registered benefits to Amazon worth around €250 million. users in Germany to remove unlawful content - European Commission, 10/17 within 24 hours of receiving a complaint. - German Federal Ministry of Justice & Consumer Protection, 10/17 Source: European Union (5/18), European Commission (6/17, 10/17, 33 12/16), Bundeskartellmt (German Competition Authority (12/17). German Federal Ministry of Justice and Consumer Protection (10/17).

Internet Companies = Key to Understand Unintended Consequences of Products... We’re an idealistic & optimistic company. For the first decade, we really focused on all the good that connecting people brings. But it’s clear now that we [Facebook] didn’t do enough. We didn’t focus enough on preventing abuse & thinking through how people could use these tools to do harm as well. - Mark Zuckerberg, Facebook CEO, 4/18 Source: Facebook (4/18). 34

…Regulators = Key to Understand Unintended Consequences of Regulation This month, the European Union will embark on an expansive effort to give people more control over their data online... As it comes into force, Europe should be mindful of unintended consequences & open to change when things go wrong. - Bloomberg Opinion Editorial, 5/8/18 Source: Bloomberg (5/18). 35

It’s Crucial To Manage For Unintended Consequences… But It’s Irresponsible to Stop Innovation + Progress 36

USA Internet Leaders = Aggressive + Forward-Thinking Investors for Years 37

Investment (Public + Private) Into Technology Companies = High for Two Decades Global USA-Listed Technology IPO Issuance & Global Technology Venture Capital Financing $200B 8,000 obal Gl ing, $150B 6,000 ite nc Fina mpos te a Co $100B 4,000 iv r Q P DA & S O NA IP $50B 2,000 hnology c e T $0 0 1990 1995 2000 2005 2010 2015 2018YTD* Technology Private Financing Technology IPO NASDAQ Source: Morgan Stanley Equity Capital Markets, *2018YTD figure as of 5/25/18, Thomson ONE. All global USA-listed technology IPOs over $30MM, data per Dealogic, Bloomberg, & Capital IQ. 2012: Facebook ($16B IPO) = 75% of 2012 38 IPO $ value. 2014: Alibaba ($25B IPO) = 69% of 2014 IPO $ value. 2017: Snap ($4B IPO) = 34% of 2017 $ value.

Technology Companies = 25% & Rising % of Market Cap, USA USA Information Technology % of MSCI Market Capitalization 40% 33% USA March, 2000 , on ati 30% 25% iz April, 2018 tal Capi et 20% ark M SCI M 10% of % IT 0% 1997 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 Source: FactSet, Katy Huberty @ Morgan Stanley. MSCI, Formerly Morgan Stanley Capital International = American provider of equity, fixed income, hedge fund stock market indexes, and 39 equity portfolio analysis tools. Data refers to MSCI’s index of USA publicly traded companies.

Technology Companies = 6 of Top 15 R&D + Capex Spenders, USA USA Public Company Research & Development Spend + Capital Expenditures (2017) Amazon Amazon +45% Y/Y Google / Alphabet Google / Alphabet +23% Intel Intel +11% Apple +5% Apple Microsoft Microsoft +6% AT&T -4% Verizon +1% Exxon Mobil -4% General Motors +5% Ford +5% Facebook +40% Facebook Chevron -26% Johnson & Johnson +12% General Electric +2% Merck +3% $0 $10B $20B $30B $40B 2017 R&D + Capex Capex R&D Source: SEC Edgar, Katy Huberty @ Morgan Stanley. Note: All figures are calendar year 2017. Amazon R&D = Tech & Content spend. General Motors does not include purchases of leased vehicles. AT&T capex does not include interest 40 during construction, just purchases of property, plant, & equipment. Verizon capitalizes R&D expense (i.e. reported as capex). General Electric R&D = GE funded, not government or customer. Bold indicates tech companies.

Technology Companies = Largest + Fastest Growing R&D + Capex Spenders, USA Research & Development Spend + Capital Expenditures – Select USA GICS Sectors $300B Technology +9% CAGR +18% Y/Y USA , x $200B Cape Healthcare* D + +4% CAGR R& +8% Y/Y $100B Discretionary 0% CAGR -22% Y/Y $0 2007 2009 2011 2013 2015 2017 Technology Healthcare Energy Materials Industrials Discretionary Staples Utilities Telecom Source: ClariFi, Katy Huberty @ Morgan Stanley. GICS = Global Industry Classification Standard, an industry taxonomy developed in 1999 by MSCI and Standard & Poor's (S&P) for use by the global financial community. CAGR = Compounded annual growth rate from 2007-2017. Note: Amazon, Netflix and Expedia removed from Discretionary 41 Sector & added to Technology. Discretionary includes companies that sell goods & services that are considered non-essential by consumers such as Starbucks (restaurants) & Nike (apparel). See appendix for detailed GICs definition. ClariFi does not have R&D or Capex data from Financial Services. *Healthcare Includes pharmaceutical companies.

Technology Companies = Rising R&D + Capex as % of Revenue…18% vs. 13% (2007) USA Technology Company Research & Development Spend + Capital Expenditures vs. % of Revenue $800B 20% 18% $600B 15% USA 13% x, ue en Cape $400B 10% Rev + of R&D % $200B 5% $0 0% 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 R&D + Capex % of Revenue Source: ClariFi, Katy Huberty @ Morgan Stanley. GICS = Global Industry Classification Standard, an industry taxonomy developed in 1999 by MSCI and Standard & Poor's (S&P) for use by the global financial community. Note: Amazon, Netflix and Expedia removed 42 from Discretionary Sector & added to Technology. Discretionary includes companies that sell goods & services that are considered non-essential by consumers such as Starbucks (restaurants) & Nike (apparel). See appendix for detailed GICs definition.

USA Tech Companies… Aggressive Competition + Spending on R&D + Capex = Driving Innovation + Growth 43

E-COMMERCE = TRANSFORMATION ACCELERATING 44

E-Commerce = Acceleration Continues @ +16% vs. +14% Y/Y, USA E-Commerce Sales + Y/Y Growth $500B 20% $400B 16% USA , s le th a $300B 12% S e Grow rc /Y Y $200B 8% mme Co - E $100B 4% $0 0% 2010 2011 2012 2013 2014 2015 2016 2017 E-Commerce Sales Y/Y Growth Source: St. Louis Federal Reserve FRED database. Note: Historic data (Pre-2016) adjusted / back-casted in 2017 by USA Census Bureau to better 45 align with Annual Retail Trade + Monthly Retail Trade Survey data.

E-Commerce vs. Physical Retail = Share Gains Continue @ 13% of Retail E-Commerce as % of Retail Sales 16% 12% SA U , Share 8% ce er omm C - E 4% 0% 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Source: USA Census Bureau, St. Louis Federal Reserve FRED database. Note: 13% = Annualized share. Penetration calculated by dividing 46 E-Commerce sales by “Core” Retail Sales (excluding food services, motor vehicles / auto parts, gas stations & fuel). All figures are seasonally adjusted.

Amazon = E-Commerce Share Gains Continue @ 28% vs. 20% in 2013 E-Commerce Gross Merchandise Value (GMV) – Amazon vs. Other Other 100% 2013 80% USA 2017 , e rc 60% mme Amazon Co - E 40% of re 2017 ha $129B GMV = 28% Share S 20% 2013 $52B GMV = 20% Share 0% $0 $100B $200B $300B $400B $500B GMV Source: St. Louis Federal Reserve FRED database, Brian Nowak @ Morgan Stanley (5/18). Morgan Stanley Amazon USA GMV estimates exclude in-store GMV and assume 90% of 47 North American GMV is USA. Market share calculated using FRED E-Commerce sales data.

E-Commerce = Evolving + Scaling 48

E-Commerce = Mobile / Interactive / Personalized / In-Feed + Inbox / Front-Doored Instacart Find Explore View + Share Pay Update Local Store Custom Savings Recommendations Seamlessly Source: Instacart (5/18) 49

E-Commerce = A Look @ Tools + Numbers… Payment Online Store Online Payment Fraud Prevention Purchase Financing Customer Support Finding Customers Delivering Product 50

Offline Merchants = Set Up Payment System… Square Estimated Active Sellers & Points of Sale (POS) Gross Payment Volume (GPV) 3MM $90B obal Gl , rs Software Services e ll 2MM $60B e S obal Payroll Loans e Gl , tiv V c A GP ted 1MM $30B tima s E Invoices Analytics 0 $0 2013 2014 2015 2016 2017 GPV Active Sellers Source: Square (5/18). Note: Active Sellers have accepted five or more payments using Square in the last 12 months. In 11/15 Square disclosed it had 2MM users and in 3/16 disclosed it was adding 100K sellers per quarter 51 – assuming seller trends remained constant, Square had approximately 2.8MM active sellers at the end of 2017. (~2.8MM = 2017E)

...Build Online Store… Shopify Active Merchants & Online Stores Gross Merchandise Volume (GMV) 900K $30B obal 600K $20B Gl , s nt obal ha Gl , rc e M GMV e 300K $10B tiv c A 0 $0 2013 2014 2015 2016 2017E GMV Merchants Source: Shopify, Brian Essex @ Morgan Stanley. Note: Active Merchants refers to merchants with an active Shopify subscription at the end of the relevant period. 2017 52 Active merchants and GMV are estimates based on periodic disclosures. (609K = 2017E)

…Integrate Online Payment System… Stripe Payment API Implementation Source: Stripe (5/18). 53

...Integrate Fraud Prevention… Merchants Signifyd Fraud Prevention 12K Increase Revenue obal 8K Gl , s nt Fast Decisions (milliseconds) ha rc e M e tiv4K c A Shift Liability 0 2015 2016 2017 Source: Signifyd (5/18). Note: Merchants refers to retailers using Signifyd services to monitor for fraud @ period end. (10K = 2017) 54

…Integrate Purchase Financing… Affirm Financing 1,200+ = Merchants $350 Source: Affirm (5/18). 55

…Integrate Customer Support… Intercom Customer Conversations Real-Time Support 500MM 400MM obal Gl d, 300MM tarte S tions a 200MM rs e nv Co 100MM 0 2013 2014 2015 2016 2017 2018 Source: Intercom (5/18). Note: Conversations started include messages initiated by business & customers. (500MM = 2018) 56

…Find Customers… Criteo Marketing Clients Customer Targeting 20K 15K obal Gl , 10K s nt Clie 5K 0 2013 2014 2015 2016 2017 Source: Criteo (5/18). Note: Clients defined as active clients @ relevant period end. (18K = 2017) 57

…Deliver Products to Customers Product Delivery Parcel Volume UPS + FedEx + USPS* 12B 10B 8B * USA , 6B me olu V 4B 2B 0 2012 2013 2014 2015 2016 2017 USPS UPS FedEx Source: UPS, FedEx, USPS, Caviar. *Combines USPS's domestic shipping & package services volumes, FedEx's domestic package volumes, and UPS's 58 domestic package volumes. All figures are calendar year end except FedEx which includes LTM figures ending November 30 due to offset fiscal year.

…E-Commerce = A Look @ Tools + Numbers Payment Online Store Online Payment Fraud Prevention Purchase Financing Customer Support Finding Customers Delivering Product 59

Building / Deploying Online Stores = Trend Evinced by Shopify Storefront Exchange Shopify Storefront Exchange (Launched 6/17) Loopies.com Source: Shopify (5/18) 60

Online Product Finding Evolution = Search Leads… Discovery Emerging Getting More... Data Driven / Personalized / Competitive 61

Product Finding = Often Starts @ Search (Amazon + Google...) Where Do You Begin Your Product Search? 49% Amazon 15% Other 36% Search Engine Source: Survata (9/17). Note: n = 2,000 USA customers. 62

Product Finding (Amazon) = Started @ Search...Fulfilled by Amazon Product Search 1-Click Purchasing Prime Fulfillment Sponsored Product Listings Voice Search + Fulfillment Source: The Internet Archive, Amazon. 63

Product Finding (Google) = Started @ Search...Fulfilled by Others Organic Search Paid Search Google Shopping Product Listing Ads Shopping Actions Source: The Internet Archive, Google. 64

Online Product Finding Evolution = Search Leads… Discovery Emerging Getting More... Data Driven / Personalized / Competitive 65

Product Finding (Facebook / Instagram) = Started @ Personalized Discovery in Feed Facebook Instagram Source: Facebook (5/18), Instagram (5/18). 66

Online Product Finding Evolution = Search Leads… Discovery Emerging Getting More... Data Driven / Personalized / Competitive 67

Google = Ad Platform to a Commerce Platform... Amazon = Commerce Platform to an Ad Platform 1997…2000 2018 AdWords Google Home Ordering Google 1-Click Checkout Sponsored Products Amazon Source: Advia (Google 2000 image), TechCrunch (2/17), Amazon (5/18). 68

E-Commerce-Related Advertising Revenue = Rising @ Google + Amazon + Facebook Google Amazon Facebook 3x = Engagement Increase $4B +42% Y/Y = >80MM +23% Y/Y = For Top Mobile Product Listing Ad* Ad Revenue SMBs with Pages Source: Google (7/17), Brian Nowak @ Morgan Stanley (Amazon Ad revenue estimate, 5/18), Facebook (4/18). *Google disclosed that the leftmost listing in a mobile product listing ad experiences 3x engagement. 69

Social Media = Enabling More Efficient Product Discovery / Commerce 70

Social Media = Driving Product Discovery + Purchases Social Media Driving …Social Media Discovery Product Discovery… Driving Purchases Facebook 78% Instagram 59% 44% 45% Pinterest 59% 55% = Bought Product Online After Social 11% Twitter 34% Media Discovery Snap 22% Bought Online Later Bought Online Immediately 0% 50% 100% Never Bought / Other % of Respondents that Have Discovered % of Respondents, USA (18-65 Years Old) Products on Platform, USA (18-34 Years Old) Source: Curalate Consumer Survey 2017 (8/17). Note: n = 1,000 USA consumers ages 18-65. Left chart question: ‘In the last 3 months, have you discovered any retail products that you were interested in buying on 71 any of the following social media channels?’ Right chart question: ‘What action did you take after discovering a product in a brand’s social media post?’ Never Bought / Other includes offline purchases made later.

Social Media = Share of E-Commerce Referrals Rising @ 6% vs. 2% (2015) Social / Feed Referrals to E-Commerce Sites 8% 6% 6% SA U 4% e, ar Sh 2% 2% 0% Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 2015 2016 2017 2018 Source: Adobe Digital Insights (5/18). Note: Adobe Digital Insights based on 50B+ online USA page visits since 2015. Data is collected on a per-visit basis across all internet 72 connected devices and then aggregated by Adobe. Data reflects 5/1/18 measurements.

Social Media = Helping Drive Growth for Emerging DTC Retailers / Brands Select USA Direct-to-Consumer (DTC) Brands – Revenue Ramp to $100MM Since Inception* $100MM $80MM A US $60MM nue, e Rev $40MM nnual A $20MM $0 0 1 2 3 4 5 6 7 Years Since Inception Source: Internet Retailer 2017 Top 1,000 Guide. *Data only for E-Commerce sales and shown in 2017 dollars. Chart includes pure-play E-Commerce retailers and evolved pure-play retailers. The Top 1,000 Guide uses a combination of 73 internal research staff and well-known e-commerce market measurement firms such as Compete, Compuware APM, comScore, ForeSee, Experian Marketing Services, StellaService and ROI Revolution to collect and verify information.

Social Media = Ad Engagement Rising…Facebook E-Commerce CTRs Rising Facebook E-Commerce CTRs (Click-Through Rates) 4% l loba 3% 3% R, G CT e rc 2% mme Co - 1% E 1% book e c Fa 0% Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 2016 2017 2018 Source: Facebook, Nanigans Quarterly Facebook Benchmarking Data. Note: Click-Through Rate is defined as the percentage of people visiting a web page who access a hyperlink text from a particular 74 advertisement. CTR figures based on $600MM+ of ad spend through Nannigan’s platform.

Return on Ad Spend = Cost Rising @ Faster Rate than Reach Facebook E-Commerce eCPM vs. CTR Y/Y Growth eCPM = +112% 120% In performance-based 80% [digital advertising] channels, CTR = +61% competition for top placement has obal Gl , reduced ROIs over the years & th40% been a source of margin pressure… Grow /Y - Glenn D. Fogel, CEO & President, Booking Holdings Y 0% Q3:17 Earnings Call, (11/17) -40% Q1 Q2 Q3 Q4 Q1 2017 2018 eCPM CTR Source: Booking Holdings Inc. (11/17), Nanigans Quarterly Facebook Benchmarking Data. Note: eCPMs are defined as the effective (blended across ad formats) cost per thousand ad impressions. Click-Through Rate is defined as the percentage of people visiting a web page who access a hyperlink text from a particular advertisement. CTR 75 figures based on $600MM+ of ad spend through Nanigan’s platform. In 2017, Booking Holdings spent $4.1B on online performance advertising which is primarily focused on search engine marketing (SEM) channels. The quote on the left relates to historical long-term ad ROI trends as competition across performance channels intensified.

Customer Lifetime Value (LTV) = Importance Rising as... Customer Acquisition Cost (CAC) Increases What Do You Consider To Be Important Ad Spending Optimization Metrics? Customer Lifetime Value 27% Impressions / Web Traffic 19% Brand Recognition & Lift 18% Closed-Won Business 15% Last-Click Attribution 13% Multi-touch Attribution 8% 0% 10% 20% 30% % of Respondents, Global Source: Salesforce Digital Advertising 2020 Report (1/18). Note: n = 900 full-time advertisers, media buyers, and marketers with the title of manager and above. Respondents are from companies in North America (USA, 76 Canada), Europe (France, Germany, Netherlands, UK, Ireland) and Asia Pacific (Japan, Australia, New Zealand) with each region having 300 participants. The survey was done online via FocusVision in 11/17.

Lifetime Value / Customer Acquisition Cost (LTV / CAC) = Increasingly Important Metric for Retailers / Brands Facebook Ad Analytics Tools LTV Integration Source: Facebook (3/17). 77

Data-Driven Personalization / Recommendations = Early Innings @ Scale 78

Evolution of Commerce Drivers (1890s -> 2010s) = Demographic -> Brand -> Utility -> Data 1890s - 1940s 1940s - 1990s 1990s - 2010s 2010s - … Demographic Brand Utility Data Catalogs DepartmentStores / E-Commerce – E-Commerce – Malls Transactional Personalized Limited product Rising product Massive product Curated product selection + selection + selection + 24x7 discovery + 24x7 shopping moments shoppingmoments shopping moments recommendations • Macy’s • Amazon • Sears Roebuck • Amazon • GAP • Facebook • MontgomeryWard • eBay • Nike • Stitch Fix Source: Eric Feng @ Kleiner Perkins Wikimedia, eBay, Stitch Fix. 79

Product Purchases = Many Evolving from Buying to Subscribing 80

Subscription Service Growth = Driven by... Access / Selection / Price / Experience / Personalization Online Subscription Services Subscribers Growth Representative Companies 2017 Y/Y Netflix Video 118MM +25% Amazon Commerce/ Media 100MM -- Spotify Music / Audio 71MM +48% Sony PlayStation Plus Gaming 34MM +30% Dropbox File Storage 11MM +25% The New York Times News / Media 3MM +43% Stitch Fix Fashion / Clothing 3MM +31% LegalZoom Legal Services 550K +16% Peloton Fitness 172K +173% Source: Netflix, Amazon, Spotify, Sony, Dropbox, The New York Times, Stitch Fix, LegalZoom, Peloton. Note: Netflix = global streaming memberships. The New York Times = digital subscribers. Sony PlayStation 81 Plus figures reflect FY, which ends March 31. Stitch Fix figures reflect FY, which ends January 31.

Free-to-Paid Conversion = Driven by User Experience... Spotify Subscribers @ 45% of MAUs vs. 0% @ 2008 Launch Spotify Subscribers % of MAU 75MM 60% U al A 50MM 40% M lob of G , % ers ers crib crib 25MM 20% Subs Subs 0 0% 2014 2015 2016 2017 Subscribers Subscribers % of MAU Source: Spotify (5/18). Note: MAU = Monthly Active Users. 82

Shopping = Entertainment… 83

Mobile Shopping Usage = Sessions Growing Fast Mobile Shopping App Sessions – Growth Y/Y Average = 6% Shopping Shopping 54% Music / Media / Entertainment 43% Business / Finance 33% Utilities / Productivity 20% News / Magazine 20% Sports -8% Photography -8% Personalization -8% Games -16% Lifestyle -40% -60% -40% -20% 0% 20% 40% 60% Session Growth Y/Y (Global, 2017 vs. 2016) Source: Flurry Analytics State of Mobile 2017 (1/18). Note: n = 1MM applications across 2.6B devices globally. Sessions defined as when a user opens an app. 84

Product + Price Discovery = Often Video-Enabled… YouTube Taobao Many USA Consumers View YouTube 1.5MM+ Active Before Purchasing Products Content Creators Source: YouTube (3/16, 5/18), Alibaba (3/18), Right image: South China Morning Post (2/18). Note: Many USA customers refers to data in a report published by Google, based on Google / Ipsos Connect, 85 YouTubeSports Viewers Study conducted on n = 1,500 18-54 year old consumers in the USA in 3/16.

…Product + Price Discovery = Often Social + Gamified Wish Pinduoduo Hourly Deals Refer Friends to 300MM+ Users Reduce Price Source: Wish (5/18), Pinduoduo (1/18), Right image: Walkthechat (1/18). Note: Wish user figures are cumulative users, not MAU. 86

Physical Retail Trending = Long-Term Growth Deceleration 87

Physical Retail = Long-Term Sales Growth Deceleration Trend Physical Retail Sales + Y/Y Growth, USA $4B 6% $3B 3% USA , s le a th S il ow r $2B 0% G Reta /Y l a Y ic s hy P $1B -3% $0 -6% 2000 2002 2004 2006 2008 2010 2012 2014 2016 Volume Growth Y/Y Source: St. Louis Federal Reserve FRED Database. Note: Physical Retail includes all retail sales excluding food services, motor vehicles / auto parts & fuel. 88

‘New Retail’ = Alibaba View from China 89

Alibaba = Building E-Commerce Ecosystem Born in China Source: Alibaba Investor Day (6/17). 90

Original Source

Alibaba & Amazon = Similar Focus Areas… Alibaba = Higher GMV…Amazon = Higher Revenue (2017) Alibaba Amazon $509B= Market Capitalization $783B = Market Capitalization $225B = GMV(E) +25% Y/Y $701B = GMV(E) +29%Y/Y $178B = Revenue +31% Y/Y $34B = Revenue +31% Y/Y 37% = Gross Margin 60% = Gross Margin $4B = Free Cash Flow $14B = Free Cash Flow 31% = Non-USA Revenue as % of Total** 8% = Non-China Revenue as % of Total** Tmall / Taobao / AliExpress / Online Lazada / Alibaba.com / Amazon.com Marketplace 1688.com / Juhuasuan / Daraz Physical Whole Foods / Amazon Go / Intime / Suning* / Hema Amazonbooks Retail Ant Financial* / Paytm* Amazon Payments Payments Youku / UCWeb / Alisports / Amazon Video / Amazon Digital Alibaba Music / Damai / Music / Twitch / Amazon Game Entertainment Alibaba Pictures* Studios / Audible Ele.Me (Local) / Koubei (Local) / Alimama / Alexa (IoT) / Ring (IoT) / (Marketing) / Cainiao (Logistics) / Autonavi Kindle + Fire Other (Mapping) / Tmall Genie (IoT) Devices (Hardware) Alibaba Cloud Amazon Web Services (AWS) Cloud Platform Source: Grace Chen (Alibaba) + Brian Nowak (Amazon) @ Morgan Stanley. *Alibaba has invested but does not have a majority ownership. **Alibaba Non- China revenue = Alibaba International Commerce revenue (AliExpress, Lazada, & Alibaba.com). Amazon Non-USA revenue = Retail sales of consumer 91 products & subscriptions through internationally-focused websites outside of North America. Note: All figures reflect calendar year 2017. Alibaba GMV includes Non-China GMV estimates, Y/Y Growth is FX adjusted using 6.76 RMB / USD average exchange rate for 2017. All figures refer to calendar year. Market cap as of 5/29/18. Amazon GMV includes in-store GMV. FCF = Cash flow from operations - stock-based compensation - capital expenditures.

Alibaba = ‘New Retail’ Vision Starts in China… …through technology & consumer insights, we [Alibaba] put the right products in front of right customers at the right time… our ‘New Retail’ initiatives are substantially growing Alibaba’s total addressable market in commerce… in this process of digitizing the entire retail operation, we are driving a massive transformation of the traditional retail industry. It is fair to say that our e-commerce platform is fast becoming the leading retail infrastructure of China. Since Jack Ma coined the term ‘New Retail’ in 2016, the term has been widely adopted in China by traditional retailers & Internet companies alike. New Retail has become the most talked about concept in business… Alibaba has three unique success factors that are enabling us to realize the New Retail vision. Alibaba CQ1:18 & CQ4:17 Earnings Calls, 5/4/18, 1/24/18 92

…Alibaba = ‘New Retail’ Vision Starts in China …Alibaba’s marketplace platforms handle billions of transactions each month in shopping, daily services & payments. These transactions provide us with the best insights into consumer behavior & shifting consumption trends. This puts us in the best position to enable our retail partners to grow their business. …Alibaba is a deep technology company. We contribute expertise in cloud, artificial intelligence, mobile transactions & enterprise systems to help our retail partners improve their businesses through digitization & operating efficiency. …Alibaba has the most comprehensive ecosystem of commerce platforms, logistics & payments to support the digital transformation of the retail sector. Alibaba CQ1:18 & CQ4:17 Earnings Calls, 5/4/18, 1/24/18. 93

…Alibaba = Extending Platform Beyond China Alibaba Non-China E-Commerce Highlights Selected Investment Revenue $3B 90% Company Country Category Type Date International Revenue = Daraz.pk Pakistan Marketplace M&A 5/18 8.4% vs. 7.9 Y/Y* ue en Tokopedia Indonesia Marketplace Equity 8/17 ev R $2B 60% h Paytm India Payments Equity 4/17 t w ro ommerce G Lazada Singapore Marketplace M&A 4/16 C Y / al Y $1B 30% on nternati I $0 0% 2012 2013 2014 2015 2016 2017 International Commerce Revenue Source: Alibaba, Pitchbook. *Percentages represent international commerce revenue proportion of total revenue. Note: All figures are calendar year. Revenue figures translated using the USD / CNY = 6.76, the average rate for 2017. Grey indicates 94 a majority control stake, all others are minority investments. Country based on headquarters, not countries of operation. Alibaba International Commerce revenue includes revenue generated from AliExpress, Lazada, and Alibaba.com.

INTERNET ADVERTISING = GROWTH CONTINUING... ACCOUNTABILITY RISING 95

Advertising $ = Shift to Usage (Mobile) Continues % of Time Spent in Media vs. % of Advertising Spending 50% / USA , 40% 7 dia 1 0 2 36% 36% in Me 30% ~$7B imending, 29% T Opportunity pe S dia 26% ing Me 20% of 20% rtis 18% e % dv A 13% 10% 9% 9% 4% 0% Print Radio TV Desktop Mobile Time Spent Ad Spend Source: Internet and Mobile advertising spend based on IAB and PwC data for full year 2017. Print advertising spend based on Magna Global estimates for full year 2017. Print includes newspaper and magazine. ~$7B opportunity 96 calculated assuming Mobile (IAB) ad spend share equal its respective time spent share. Time spent share data based on eMarketer (9/17). Arrows denote Y/Y shift in percent share. Excludes out-of-home, video game & cinema advertising.

Internet Advertising = +21% vs. +22% Y/Y Internet Advertising Spend $90B 30% $88 USA $73 nd, $60B 20% pe $60 S th ing $50 Grow rtis e $43 /Y $37 Y dv $32 A $30B 10% t $26 ne $23 Inter 0 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 Desktop Advertising Mobile Advertising Y/Y Growth Source: IAB / PWC Internet Advertising Report (5/18). 97

Advertisers / Users vs. Content Platforms = Accountability Rising... Many Americans Believe Fake News Is Sowing Confusion Pew Research Center, December 2016 Adweek, July 2017 The Wall Street Journal, February 2018 98

...Advertisers / Users vs. Content Platforms = Accountability Rising Content Initiatives Google / YouTube Facebook (Q1:18) 8MM = Videos Removed (Q4:17)… 583MM = Fake Accounts Removed… 81% Flagged by Algorithms… 99% Flagged Prior To User Reporting 75% Removed Before First View 21MM = Pieces of Lewd Content Removed… 2MM= Videos De-Monetized For 96% Flagged by Algorithms Misleading Content Tagging (2017) 3.5MM = Pieces of Violent Content Removed… 10K = Content Moderators (2018 Goal) 86% Flagged by Algorithms 2.5MM = Pieces of Hate Speech Removed… 38% Flagged by Algorithms +7,500 = Content Moderators… 3,000 Hired (5/17–2/18) Source: YouTube (5/18, 12/17), Facebook (Transparency Report: 5/18, 5/17, 2/18). Note: All Google content moderators represent full-time 99 hires but Facebook content moderators are not all full-time.

CONSUMER SPENDING = DYNAMICS EVOLVING… INTERNET CREATING OPPORTUNITIES 100

Consumers… Making Ends Meet = Difficult 101

Household Debt = Highest Level Ever & Rising… Change vs. Q3:08 = Student +126%...Auto +51%...Mortgage -4% Household Debt & Unemployment Rate $15T 15% $13.1T $12.7T $12T A 10% USA , $9T Rate Debt, US nt $6T me hold e 6% 5% us mploy 4% Ho e $3T Un $0 0% 2003 2005 2007 2009 2011 2013 2015 2017 Mortgage Student Loan Auto Credit Card Home Equity Revolving Other Unemployment Rate Source: Federal Reserve Bank of New York Consumer Credit Panel / Equifax, Quarterly Household Debt and Credit Report, Q4:17; St. Louis Federal Reserve FRED Database. 102

Personal Saving Rate = Falling @ 3% vs. 12% Fifty Years Ago… Debt-to-Annual-Income Ratio = Rising @ 22% vs. 15% Personal Saving Rate & Debt-to-Annual-Income* Ratio 25% Debt-to-Annual-Income* Ratio 20% 15% USA , o ati 10% R 5% Personal Saving Rate 0% 1968 1978 1988 1998 2008 2018 Source: St. Louis Federal Reserve FRED Database, USA Federal Reserve Bank. *Consumer debt-to-annual-income ratio reflects outstanding credit extended to individuals for household, family, and other personal expenditures, excluding loans secured by real 103 estate vs. average annual personal income. Personal saving rate is shown as a percentage of disposable personal income (DPI), frequently referred to as “the personal saving rate.” (i.e. the annual share of disposable income dedicated to saving)

Relative Household Spending = Shifting Over Past Half-Century 104

Relative Household Spending Rising Over Time = Shelter + Pensions / Insurance + Healthcare… Relative Household Spending 20% 17% 15% 15% SA U 12% , d 10% en 10% 8% Sp 7% 7% al u 5%5% n 5% n A 0% 1972 1990 2017 Total Expenditure $11K $31K $68K Source: USA Bureau of Labor Statistics (BLS), Consumer Expenditure Survey. *Pensions + Insurance includes deductions for private retirement accounts, social security, and life insurance. **Other Includes education and miscellaneous other expenses. Note: Results based on Surveys of American Urban & Rural Households 105 (Families & Single Consumers). 1972 data reflects non-annual survey conducted by BLS + Census Bureau to adjust CPI. 1990 and 2017 Data Based on Annual Survey performed by BLS + Census Bureau. Healthcare costs include insurance, drugs, out-of-pocket medical expenses, etc. 2017 = mid-year figures.

…Relative Household Spending Falling Over Time = Food + Entertainment + Apparel Relative Household Spending 20% 15% 15% 15% SA U , 12% d en 10% Sp 6% al u 5% 5% n 5% 4% 5% n A 3% 0% 1972 1990 2017 Total Expenditure $11K $31K $68K Source: USA Bureau of Labor Statistics (BLS), Consumer Expenditure Survey. *Pensions + Insurance includes deductions for private retirement accounts, social security, and life insurance. **Other Includes education and miscellaneous other expenses. Note: Results based on Surveys of American Urban & Rural Households 106 (Families & Single Consumers). 1972 data reflects non-annual survey conducted by BLS + Census Bureau to adjust CPI. 1990 and 2017 Data Based on Annual Survey performed by BLS + Census Bureau. Healthcare costs include insurance, drugs, out-of-pocket medical expenses, etc. 2017 = mid-year figures.

Food = 12% vs. 15% of Household Spending 28 Years Ago... 107

Grocery Price Growth = Declining Trend… Owing To Grocery Competition Grocery Price Change Y/Y & Market Share of Top 20 Grocers Top 20 Grocer Market Share 10% 75% 8% SA SA U U , 6% 50% Y e, Y/ ar e g Sh 4% t an h ke C ar 2% 25% M ce i Pr Grocery* Price Y/Y Change 0% -2% 0% 1990 1995 2000 2005 2010 2015 2017 Source: USDA Research Services, using data from the USA Census Bureau’s Annual Retail Trade Survey + Company Reports, USA Bureau of Labor Statistics (BLS). *Grocery Price growth refers to the growth in prices for “Food at Home” as reported by the USA Census Bureau. Note: Includes all food 108 purchases in CPI, other than meals purchased away from home (e.g., Restaurants). Grocery @ 56% of Food Spend in 2017 vs. 58% in 1990 per BLS.

Walmart = Helped Reduce Grocery Prices via Technology + Scale... per Greg Melich @ MoffettNathanson Walmart – Grocery Share 20% SA15% U e, ar 10% Sh y cer o 5% Gr 0% 1995 2000 2005 2010 2015 2017 By using technology to reduce inventory, expenses & shrinkage, we can create lower prices for our customers. - Walmart 1999 Annual Report Source: Greg Melich @ MoffettNathanson Note: Share reflects retail value of food for off-site 109 consumption sold across all Walmart properties.

E-Commerce = Helping Reduce Prices for Consumers 110

E-Commerce sales have risen rapidly over the past decade. Online prices are falling – absolutely & relative to – traditional inflation measures like the CPI. Inflation online is, literally, 200 basis points lower per year than what the CPI has been showing. To better understand the economy going forward, we will need to find better ways to measure prices & inflation. - Austan Goolsbee, Professor of Economics, University of Chicago Booth School of Business, 5/18 111

Consumer Goods Prices = Have Fallen… -3% Online & -1% Offline Over 2 1/4 Years per Adobe DPI… Consumer Prices For Matching Products - Online vs. Offline $1.02 A US Offline = -1% $1.00 ), 6 /1 /1 1 $0.98 d to e x e Online = -3% (Ind $0.96 s e ic r P $0.94 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 2016 2017 2018 Online Retail (DPI) Offline Retail Source: Adobe Digital Economy Project Note: Adobe Digital Economy Project measures prices and sales volume for 80% of online transactions at top 100 USA retailers (15B site visits & 2.2MM products) then calculates a Digital Price Index (DPI) using a Fisher 112 Ideal model. CPI calculates USA prices using a basket of 83K goods, tracked monthly, & applied to a Laspeyeres model. DPI Excludes Apparel. Austan Goolsbee serves as strategic advisor to Adobe DPI project.

…Online vs. Offline Price Decline Leaders = TVs / Furniture / Computers / Sporting Goods per Adobe DPI Price Change, Y/Y (DPI vs. CPI), USA, 3/17-3/18 10% 0% (10%) (20%) DPI vs. CPI 5% 1% 1% 0% 0% 0% 0% -1% -1% -1% -2% -2% -3% -4% Difference (%) (30%) DPI CPI Source: Adobe Digital Economy Project Note: Adobe Digital Economy Project measures prices and sales volume for 80% of online transactions at top 100 USA retailers (15B site visits & 2.2MM products) then calculates a Digital Price Index (DPI) using a Fisher 113 Ideal model. CPI calculates USA prices using a basket of 83K goods, tracked monthly, & applied to a Laspeyeres model. DPI Excludes Apparel. Austan Goolsbee serves as strategic advisor to Adobe DPI project.

We've seen how technology can make online shopping more efficient, with lower prices, more selection & increased convenience. We are about to see the same thing happen to offline shopping. - Hal Varian, Chief Economist @ Google, 5/18 114

Relative Household Spending = How Might it Evolve? Shelter Spend = Rising Transportation Spend = Flat Healthcare Spend = Rising 115

Shelter as % of Household Spending = 17% vs. 12% (1972)... Largest Segment in % + $ Growth Relative Household Spending 20% 17% 15% 15% SA U 12% , d en 10% Sp al u n 5% n A 0% 1972 1990 2017 Total Expenditure $11K $31K $68K Source: USA Bureau of Labor Statistics (BLS), Consumer Expenditure Survey. *Pensions + Insurance includes deductions for private retirement accounts, social security, and life insurance. **Other Includes education and miscellaneous other expenses. Note: Results based on Surveys of American Urban & Rural Households 116 (Families & Single Consumers). 1972 data reflects non-annual survey conducted by BLS + Census Bureau to adjust CPI. 1990 and 2017 Data Based on Annual Survey performed by BLS + Census Bureau. Healthcare costs include insurance, drugs, out-of-pocket medical expenses, etc.. 2017 = mid-year figures.

Shelter… 117

USA Cities = Less Densely Populated vs. Developed World Population Density – Urban Areas* Top 10 ‘Advanced’ Economies**, 2014 4000 17x USA ) s a re A 3000 n rba (U 2 2000 9x r KM pe s 6x nt 5x 1000 ide ~3x ~3x ~2x Res ~2x 0 South Japan UK Italy Germany Spain France Australia USA Canada USA Korea Source: OECD, International Monetary Fund (IMF). *Urban areas defined as “Functional Urban Areas’ per OECD/EU with greater than 500K residents. **IMF determines ‘Advanced Economies’ designation using a 118 combination of GDP per Capita, Export Diversity, and integration into the global financial system.

USA Homes = Bigger vs. Developed World… Average Home Size* (Square Feet) – Select Countries USA Japan ~1,500 ~1,015 UK ~990 South Korea ~725 Source: Wikimedia, Japan Ministry of Internal Affairs, US Census Bureau, UK Office for National Statistics, Asian Development Bank Institute. *USA + Japan + UK = 2013. Korea = 2010, owing to lag in publication of government measured data. Note: Data reflects average occupied dwelling size across each nation. 119

…USA Homes = Getting Bigger...Residents Falling @ 2.5 vs. 3.0 (1972) Average New Home Square Footage & Residents 4K 4 SA U e, SA Residents U 3K 3 tag per Home o o me, F o e H ar u 2K er 2 p Sq ts New Home me en o d Square Footage H 1K 1 esi R ew N 0 0 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 Source: USA Census Bureau (6/17). Note: Data reflects newly built housing stock. Single Family homes includes newly built single family homes. Similar growth trends are seen across all housing units, as single-family homes are the majority of new USA housing stock. Average size of multifamily new 120 dwelling in USA = 1,095 square feet in 1999 (earliest data available), 1,207 square feet in 2016. Residents per household based on all households.

USA Office Space = Steadily Getting Denser / More Efficient Occupied Office Space per Employee – Square Feet 250 SA 195 U 200 180 ee, y o l 150 Emp r e p e 100 tag o o F e ar 50 u Sq 0 1990 1995 2000 2005 2010 2015 2017 Source: CoStart Portfolio strategy Analysis of USA Leased office space & USA Employment Figures (2017). 121

…Shelter... To Contain Spending… Consumers May Aim to Increase Utility of Space 122

Airbnb = Provides Income Opportunities for Hosts… Airbnb Guest Arrivals & Active Listings by Hosts 5MM Global Active Listings 90MM 6MM obal 60MM 4MM obal Gl , Gl ls a v rri tings, A t Lis s e 30MM 2MM tiv Gue c A 0 0 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Guest Arrivals Active Listings by Hosts Source: Airbnb, TechCrunch. Note: Airbnb disclosed in 2017 ~660K listings were in USA. A 2017 CBRE study of ~256K USA Airbnb listings + ~177K Airbnb hosts in Austin, Boston, Chicago, LA, Miami, Nashville, New Orleans, New York City, Oahu, 123 Portland, San Francisco, Seattle, & Washington D.C. found that 83% of listings are made by single-listing hosts, or are listings for spaces within otherwise occupied dwellings. This implies that there >500K individuals listing spaces on Airbnb in USA as of 2018.

…Airbnb Consumer Benefits = Can Offer Lower Prices for Overnight Accommodations Airbnb vs. Hotel –Average Room Price per Night $350 $306 $300 $240 $250 $220 $217 18 $193 $191 $187 $200 1/ $179 $167 ce, i $150 Pr $118 $114 $114 $110 $93 $92 $100 $65 $50 $0 New Sydney Tokyo London Toronto Paris Moscow Berlin York City Hotel Airbnb Source: AirDNA, HRS, Originally Compiled by Statista. Note: Hotel data based on HRS’s inventory of hotels. Euro prices converted to USD on 1/22/18. 124

Relative Household Spending = How Might it Evolve? Shelter Spend = Rising Transportation Spend = Flat Healthcare Spend = Rising 125

Transportation as % of Household Spending = 14% vs. 14% (1972)... #3 Segment of $ Spending Behind Shelter + Taxes Relative Household Spending 20% 16% 14% 15% 14% SA U , d en 10% Sp al u n 5% n A 0% 1972 1990 2017 Total Expenditure $11K $31K $68K Source: USA Bureau of Labor Statistics (BLS), Consumer Expenditure Survey. *Pensions + Insurance includes deductions for private retirement accounts, social security, and life insurance. **Other Includes education and miscellaneous other expenses. Note: Results based on Surveys of American Urban & Rural Households 126 (Families & Single Consumers). 1972 data reflects non-annual survey conducted by BLS + Census Bureau to adjust CPI. 1990 and 2017 Data Based on Annual Survey performed by BLS + Census Bureau. Healthcare costs include insurance, drugs, out-of-pocket medical expenses, etc.. 2017 = mid-year figures.

Transportation… To Contain Spending… Consumers Reducing Relative Spend on Vehicles + Increasing Utility of Vehicles 127

Transportation as % of Household Spending = Vehicle Purchase % Declining…Other Transportation % Rising Relative Household Spending – Transportation 60% Relative Transportation Spending = SA U , d en Vehicles Stay On Road Longer... Sp 40% @ 12 vs. 8 Years (1995) n o Average Car Lifespan tati r o sp 20% …Other Transportation Rising an r T +30% vs. 1995 e v Public Transit Usage ati 0% el R ~2x Y/Y (2017) 1972 1990 2017 Ride-Share Rides Vehicle Operation + Vehicle Purchases Maintenance* Other Gas + Oil Transportation Source: USA BLS Consumer Expenditure Survey. Vehicle Age = Bureau of Transportation Statistics + I.H.S. Public Transit Trips = American Public Transit Association Note: *Vehicle Operation + Maintenance includes Insurance, Repairs, Parking, and Other expenses. Other transportation includes all transportation outside of personal vehicles, including rise-sharing.. 128 Results based on Surveys of American Urban and Rural Households (Families and Single Consumers). 1972 data reflects non-annual survey conducted by BLS + Census Bureau to adjust CPI. 1990 and 2017 Data Based on Annual Survey performed by BLS + Census Bureau. Cars refers to all light vehicles (i.e. passenger cars + light trucks). Includes all actively driven cars. Public transit trips reflect unlinked rides (i.e. one full journey). Note: Ride Share Statistics based on Q1:16 and Q1:17 Estimates from Hillhouse Capital.

Uber = Can Provide Work Opportunities for Driver-Partners… Uber Gross Bookings & Driver-Partners 3MM Global Driver-Partners +50*% $45B 3MM $30B 2MM obal obal Gl Gl , , ings rtners a ok P - r $15B 1MM Bo e s Driv Gros $0 0 2012 2013 2014 2015 2016 2017 Gross Bookings Driver-Partners Source: Uber. *Approximately +50% Y/Y. Note: ~900K USA Driver-Partners. Note: As of Jan 2015, ~85% of Uber driver-partners drove for UberX – based on historical growth rates, it is estimated that >90% of USA Uber driver-partners drive for UberX. 129

…Uber Consumer Benefits = Lower Commute Cost vs. Personal Cars – 4 of 5 Largest USA Cities UberX / POOL vs. Personal Car* – Weekly Commute Costs 5 Largest USA Cities, 2017 $250 $218 $200 $181 st o $142 $150 C $130 $116 $96 eekly $100 $89 W $77 $65 $62 $50 $0 New York City Chicago Washington Los Angeles Dallas D.C. Personal Car Uber Source: Nerdwallet Study, March 2017. Washington D.C. included in Top 5 due to including of Baltimore MSA population. *Car commute costs include Gas (OPIS), Maintenance (Edmunds.com), Insurance (NerdWallet), & Parking (parkme.com). Note: Commute distances are from 2015 Brookings analysis. 130 Uber data is based on a suburbs-to-city-center trip mirroring average commute distance for a metro. Data collected at peak commute times in February 2017. Cheapest Option (UberX, UberPOOL, etc.) selected for Uber costs.

Relative Household Spending = How Might it Evolve? Shelter Spend = Rising Transportation Spend = Flat Healthcare Spend = Rising CREATED BY NOAH KNAUF @ KLEINER PERKINS 131

Healthcare as % of Household Spending = 7% vs. 5% (1972)... Fastest Relative % Grower Relative Household Spending 20% 15% SA U , d en 10% Sp 7% al u 5% 5% n 5% n A 0% 1972 1990 2017 Total Expenditure $11K $31K $68K Source: USA Bureau of Labor Statistics (BLS), Consumer Expenditure Survey. *Pensions + Insurance includes deductions for private retirement accounts, social security, and life insurance. **Other Includes education and miscellaneous other expenses. Note: Results based on Surveys of American Urban & Rural Households 132 (Families & Single Consumers). 1972 data reflects non-annual survey conducted by BLS + Census Bureau to adjust CPI. 1990 and 2017 Data Based on Annual Survey performed by BLS + Census Bureau. Healthcare costs include insurance, drugs, out-of-pocket medical expenses, etc.. 2017 = mid-year figures.

Healthcare Spending = Increasingly Shifting to Consumers… 133

USA Healthcare Insurance Costs = Rising for All… Consumers Paying Higher Portion @ 18% vs. 14% (1999)… Annual Health Insurance Premiums vs. Employee Contribution $8K 20% 18% r SA fo U , 14% age $6K 15% miums er re ov P C e le ribution nc nt ng $4K 10% ura Si Co d e Ins e re sore mploy lthca E $2K 5% Spon % Hea ee nnual A Employ $0 0% 1999 2001 2003 2005 2007 2009 2011 2013 2015 2017 Premiums % Employee Contribution Source: Kaiser Family Foundation Employer Health Benefits Survey (9/17). Note: n = 2,000 private, non-federal businesses with at least 3 employees. Employers are asked for full person costs of healthcare coverage and the employee contribution. 134

…USA Healthcare Deductible Costs = Rising A Lot… Employees @ >$2K Deductible = 22% vs. 7% (2009) Annual Deductibles vs. % of Covered Employees with >$2K Deductibles $1,500 30% ith e w ble s 22% cti ee a Singl loy in Dedu $1,000 20% Emp, USA 2K ge >$ ng era ith mo A s Enrolledw Cov n ee ble le $500 7% 10% Pla cti ge Sing Dedu era l of Employ ua % Cov nn A $0 0% 2006 2008 2010 2012 2014 2016 Annual Deductible Among Covered Employees % of Employees Enrolled in a Plan with >$2K Deductible Source: Kaiser Family Foundation Employer Health Benefits Survey (9/17). Note: n = 2,000 private, non-federal businesses with at least 3 employees. Employers are asked for full person costs of healthcare coverage and the employee contribution. 135

When Consumers Start Spending More They Tend To Pay More Attention to Value + Prices… Will Market Forces Finally Come to Healthcare & Drive Prices Lower for Consumers? 136

Healthcare Patients Increasingly Developing Consumer Expectations… Modern Retail Experience Digital Engagement On-Demand Access Vertical Expertise Transparent Pricing Simple Payments 137

Healthcare Consumerization… Modern Retail Digital Healthcare On-Demand Experience Management Pharmacy One Medical Oscar Capsule Office Locations Memberships Unique Conversations 300K 30K 80 s s s ship ion e t ice er 150K 15K f40 sa f iqu O er emb Unv n M 0 0 0 Co 2014 2016 2018 2014 2015 2017 2016 2017 Source: One Medical, Web.Archive.org, Oscar, Capsule. Note: Oscar data as of the first month of each year based on enrollments timing. 138

…Healthcare Consumerization Women’s Healthcare Transparent Simplified Specific Solutions Pricing Healthcare Billing Nurx Dr. Consulta Cedar Medical Interactions* Patients % of Collections** 100K 1,000K 100% s s n s ion t nt ctio 50K ie500K ac 50% t lle er a P Co Int f o 0 0 0% % 2016 2017 2018 2013 2015 2017 0 31 60 91 Days Source: Nurx, Dr. Consulta, Cedar. *Medical interactions include prescriptions, lab orders, & messages from MDs / RNs. **Cedar data represents the % of total collections using Cedar 139 over time at a multispecialty group with 450 physicians and an ambulatory surgical center.

Consumerization of Healthcare + Rising Data Availability = On Cusp of Reducing Consumer Healthcare Spending? 140

WORK = CHANGING RAPIDLY… INTERNET HELPING, SO FAR… 141

Technology Disruption = Not New...But Accelerating 142

Technology Disruption = Not New… New Technology Proliferation Curves* 100% 75% USA n, 50% io dopt A 25% 0% 1900 1915 1930 1945 1960 1975 1990 2005 2017 Grid Electricity Radio Refrigerator Automatic Transmission Color TV Shipping Containers Microwave Computer Cell Phone Internet Social Media Usage Smartphone Usage Source: ‘Our World In Data’ collection of published economics data including Isard (1942), Grubler (1990), Pew Research, USA Census Bureau, and others. *Proliferation defined by share of households using a particular technology. In the case 143 of features (e.g., Automatic Transmission), adoption refers to share of feature in available models.

…Technology Disruption = Accelerating…Internet > PC > TV > Telephone New Technology Adoption Curves 90 Dishwasher 75 Radio USA Telephone Air Conditioning 60 ion, Washer Electricity Refrigerator dopt A 45 % Television 5 Car Microwave til 2 30 Personal Computer Un rs a Mobile Phone e 15 Y Internet 0 1867 1887 1907 1927 1947 1967 1987 2007 2017 Source: The Economist (12/15), Pew Research Center (1/17), Asymco (11/13). Note: Starting years based on invention year of each consumer product. 144

Technology Disruption Drivers = Rising & Cheaper Compute Power + Storage Capacity... Storage Price vs. Hard Drive Capacity $1,000 of Computer Equipment 10000GB $10,000,000 1.E+10 1.E+09 Pentium II PC $1,000,000 1.E+08 1000GB 1.E+07 ity $100,000 c 1.E+06 Sun 1 100GB 1.E+05 ond $10,000 c Capa e 1.E+04 S ge 10GB r 1.E+03 $1,000 ra BINAC pe 1.E+02 to per GB IBM 1130 S 1.E+01 $100 e ce 1GB tions 1.E+00 Pri Driv ula1.E-01 $10 c 0.1GB 0GB 1.E-02 Cal Hard $1 1.E-03 IBM Tabulator 1.E-04 0.01GB 0GB $0 $0.1 1.E-05 Analytical Engine 1.E-06 0.001GB 0GB $0 $0.01 1.E-07 1956 1987 2017 1900 1925 1950 1975 2000 2025 Price Per GB Capacity Source: John McCallum @ IDC, David Rosenthal @ LOCKSS Program – Stanford): Kryder’s Law. Time + Ray Kurzweil analysis of multiple sources, including Gwennap (1996), Kempt (1961) and others. Note: All figures shown on logarithmic scale. 145

...Technology Disruption Drivers = Rising & Cheaper Connectivity + Data Sharing Internet + Social Media – Global Penetration 60% 49% 50% 40% obal 33% Gl 30% 24% tration, ne e 20% P 14% 10% 0% 2009 2010 2011 2012 2013 2014 2015 2016 2017 Internet Social Media Source: United Nations / International Telecommunications Union, USA Census Bureau. Internet user data is as of mid-year Internet user data: Pew Research (USA), China Internet Network Information Center (China), Islamic Republic News Agency / InternetWorldStats / KP estimates (Iran), KP 146 estimates based on IAMAI data (India), & APJII (Indonesia). Population sourced from Central Intelligence Agency database. eMarketer estimates for Social Media users based on number of active accounts, not unique users. Penetration calculated as a % of total population based on the CIA database.

New Technologies = Created / Displaced Jobs Historically 147

New Technologies = Job Concerns / Reality Ebb + Flow Over Time 1920 1940 1960 1980 2000 2020 Source: New York Times, 2/26/1928, article by Evans Clark. Originally sourced from Louis Anslow, “Robots have been about to take all the jobs for more than 200 years,” Timeline, 5/7/16. The New York Times, 2/24/1940, article by Louis Stark. Originally sourced from Louis Anslow, “Robots have been about to take all the jobs 148 for more than 200 years,” Timeline, 5/7/16. The New York Times, 5/4/1962, article by Milton Bracker. New York Times, 9/3/1940, article by Harley Shaiken. Originally sourced from Louis Anslow, “Robots have been about to take all the jobs for more than 200 years,” Timeline, 5/7/16. 2017 Article = The New York Times.

New Technologies = Aircraft Jobs Replaced Locomotive Jobs... Locomotive vs. Aircraft Jobs 400K 300K A US 200K obs, J 100K 0 1950 1960 1970 1980 1990 2000 2010 2015 Locomotive Jobs - Engineers / Operators / Conductors Aircraft Jobs - Pilots / Mechanics / Engineers Source: ITIF analysis of IPUMS data (Atkinson + Wu); St. Louis Federal Reserve FRED Database. Note: IPUMS data tracks historical employment (since 1950) using 2010 Census occupational codes. (7140:Aircraft Mechanics + Service Technicians; 9030: Aircraft Pilots + Flight Engineers; 9200: 149 Locomotive Engineers + Operators; 9230: Railroad Brake, Signal, + Switch Operators; 9240: Railroad Conductors + Yardmasters).

…New Technologies = Services Jobs Replaced Agriculture Jobs … Agriculture vs. Services Jobs 25MM 20MM 15MM USA obs, 10MM J 5MM 0 1900 1910 1920 1930 Agriculture Jobs - Farming / Forestry / Fishing / Hunting Services Jobs - Business / Education / Healthcare / Retail / Government / Other Services Source: Growth & Structural Transformation – Herrendorf et al. (NBER, 2013) Services includes all non-farm jobs except goods-producing industries such as natural resources / mining, construction and manufacturing. 150

…Agriculture = <2% vs. 41% of Jobs in 1900 Agriculture vs. Services Jobs 150MM 120MM 90MM SA U 60MM Jobs, 1900 Agriculture = 41% of Jobs 30MM 2017 Agriculture = <2% of Jobs 0 1900 1915 1930 1945 1960 1975 1990 2005 2017 Agriculture Jobs - Farming / Forestry / Fishing / Hunting Services Jobs - Business / Education / Healthcare / Retail / Government / Other Services Source: St. Louis Federal Reserve FRED Database, Growth & Structural Transformation – Herrendorf et al. (NBER, 2013), Bureau of Labor Statistics Note: Pre-1948 Agriculture data = Herrendorf et al. Post 1948 = Bureau of Labor Statistics. Pre-1939 151 services data = Herrendorf et al. Post 1939 services data = Bureau of Labor Statistics. Services includes all non-farm jobs except excluding Goods-Producing industries such as Natural Resources / Mining, Construction and Manufacturing.

70 Years = New Technology Concerns Ebb / Flow... GDP Rises...Unemployment Ranges 2.9 - 9.7% Real GDP vs. Unemployment Rate, USA $18T 30% $12T 20% Rate DP nt G l me y Rea plo 5.8% $6T 10% 70-Year Average Unem 3.9% Unemployment Rate (4/18) $0 0% 1929 1939 1949 1959 1969 1979 1989 1999 2009 2017 Real GDP Unemployment Rate Source: St. Louis Federal Reserve FRED Database, Bureau of Economic Analysis, BLS. Note: Real GDP based on chained 2009 dollars. Unemployment rate = annual average. 152

Will Technology Impact Jobs Differently This Time? Perhaps...But It Would Be Inconsistent With History as... New Jobs / Services + Efficiencies + Growth Typically Created Around New Technologies 153

Job Market = Solid Based on Traditional High-Level Metrics, USA 154

Unemployment @ 3.9% = Well Below 5.8% Seventy Year Average Unemployment Rate 30% 20% SA U ate, R ment y 10% nemplo U Average = 5.8% 0% 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 2018 Source: St Louis Federal Reserve FRED Database, Bureau of the Budget (1957). Note: Unemployment rate calculated by diving the total workforce by the total number of unemployed people. People are classified as unemployed if they do 155 not have a job, have actively looked for work in the prior 4 weeks and are currently available for work.

Consumer Confidence = High & Rising… Index @ 100 vs. 87 Fifty-Five Year Average Consumer Confidence Index (CCI) 120 100 USA ), 80 87= 4 6 9 55-Year Average 1 to 60 d e 40 (Index I CC 20 0 1952 1957 1962 1967 1972 1977 1982 1987 1992 1997 2002 2007 2012 2017 Source: St. Louis Federal Reserve FRED Database. Note: Indexed to Q1:66 = 100. Consumer Confidence Index (Michigan Consumer Sentiment Index) is a broad measure of American consumer sentiment, as 156 measured through a 50-question telephone survey of at least 500 USA residents each month.

Job Openings = 17 Year High… @ 7MM…~3x Higher vs. 2009 Trough Job Openings* – USA 7MM 6.6MM Job Openings (3/18) 6MM 1.4MM = Professional Services + Finance 5MM 1.3MM = Healthcare + Education * 1.2MM = Trade / Transportation / Utilities 4MM nings 879K = Leisure / Hospitality 3MM ob Ope J 661K = Mining / Construction / Manufacturing 2MM 622K = Government 1MM 486K = Other 0 2000 2005 2010 2015 2018 Source: St Louis Federal Reserve FRED Database. *A job opening is defined as a non-farm specific position of employment to be filled at an establishment. Conditions include the following: there is work available for 157 that position, the job could start within 30 days, and the employer is actively recruiting for the position.

Job Growth = Stronger in Urban Areas Where 86% of Americans Live Job / Population Growth – Urban vs. Rural (Indexed to 2001) 120 Jobs = +19% Population = +15% 110 Jobs = +4% Population = +4% 100 90 2001 2006 2011 2016 Urban Rural Source: USDA ERS, BLS. Note: LAUS county-level data from BLS are aggregated into urban (metropolitan/metro) and rural (nonmetropolitan / non-metro), based on the Office of Management and Budget's 2013 metropolitan classification. Metro areas 158 defined as counties with urban areas >50K in population and the outlying counties where >35% of population commutes to an urban center for work. ’Rural’ data reflects total non-metro employment, where population has been declining since 2011.

Labor Force Participation @ 63% = Below 64% Fifty-Year Average...~3.5MM People Below Average* Labor Force Participation Rate** 90% 80% 70% USA , 60% 64% = Rate 50 Year Average tion 50% ipa rtic 40% a P e 30% bor Forc20% La 10% 0% 1900 1910 1920 1930 1940 1950 1960 1970 1980 1990 2000 2010 Source: St Louis Federal Reserve FRED Database, BLS. *In March 2018, ~161.8MM Americans were in the labor force (62.9% participation). Participation @ 50-year average of 64.3% would imply a labor force of 165.3MM. The labor force participation rate is defined as the section of working population in the age group of 16+ in the 159 economy currently employed or seeking employment. **For data from 1900-1945 the labor force participation rate includes working population over the age of 10.

Most Common Activities For Many Who Don’t Work* = Leisure / Household Activities / Education Males* (Ages 25-54) – Daily Time Use Not In Labor Force In Labor Force Watching TV +3 Hours Watching TV Other Socializing, Relaxing, Leisure Other Socializing, Relaxing, Leisure +0.7 Other (Including Sleep) +0.6 Other (Including Sleep) Household Activities & Services Household Activities & Services +0.5 Education +0.3 Education +0.01 Caring for Non-Household Members Caring For Non-Household Members Caring for Household Members -0.02 -5 Work 0 2 4 6 8 10 12 14 Hours per Day, USA, 6/16 Source: 2014 American Time Use Survey, CEA calculations, BLS. Note: Prime-age males defined as men between the ages of 25-54. Daily hours may not add up to 24 since some individuals do not report all time spent. 160 Household activities include cleaning, cooking, yardwork & home maintenance not related to caregiving.

Job Expectations = Evolving 161

Most Desired Non-Monetary Benefit for Workers = Flexibility per Gallup Would You Change Jobs to Have Access To… 75% 61% 54% 53% 51% 51% 48% ) 50% 7 1 0 40% (2 A 35% US , re ha25% S 0% Health Monetary Paid Flexible Pension Paid Profit Working Insurance Bonuses Vacation Schedule Leave Sharing From Home Source: Gallup 2017 State of the American Workplace Note: *Flexible schedule defined as ability to choose own hours of work. Gallup developed State of the American Workplace using data collected from more than 195,600 USA employees via the Gallup Panel and Gallup Daily tracking in 2015 162 and 2016, and more than 31 million respondents through Gallup's Q12 Client Database. First launched in 2010, this is the third iteration of the report.

Technology = Makes Freelance Work Easier to Find… Freelance Workforce = 3x Faster Growth vs. Total Workforce Has Technology Has Made It Workforce Growth – Easier To Find Freelance Work? Freelance vs. Total 100% 110% USA , 108% +8.1% ly USA e th, itiv os 77% 106% ow r P G 75% e ing rc 104% fo 69% pond ork +2.5% W Res 102% % 50% 100% 2014 2015 2016 2017 2014 2015 2016 2017 Total Freelance Source: ‘Freelancing in America: 2017’ survey conducted by Edelman Research, co-commissioned by Upwork and Freelancers Union. Note: Survey conducted 7/17-8/17, n = 2,173 Freelance Employees who 163 have received payment for supplemental temporary, or project-oriented work in the past 12 months.

On-Demand Jobs = Big Numbers + High Growth Increasingly Filling Needs for Workers Who Want Extra Income / Flexibility... Have Underutilized Skills / Assets 164

On-Demand Workers = 5.4MM +23%, USA per Intuit On-Demand Platform Workers, USA 8MM 6.8* 6MM 5.4 SA U 3.9 s, 4MM ker r o 2.4 W 2MM 0 2015 2016 2017 2018E Source: Intuit (2017/2018). *2018 = Forecast from 2017 data. Preliminary 2018 results appear to be in line with forecast as of 5/16/18. Note: On-demand workers defined as online marketplace workers including transportation and/or logistics for people or products, online 165 talent marketplaces, renting out space. Providing other miscellaneous consumer and business services (e.g. TaskRabbit, Gigwalk, Wonolo, etc.). Workers defined as ‘active’ employees that have done ’significant’ on-demand work within the preceding 6 months.

On-Demand Jobs = >15MM Applicants on Checkr Platform Since 2014, USA Checkr Background Check On-Demand Applicants – Top 100 Metro Areas, USA Source: Checkr (2018) 166

On-Demand Jobs = Big Numbers + High Growth Real-Time Internet-Enabled Platforms Marketplaces Uber @ 3MM Driver-Partners Etsy @ 2MM Sellers $45B 3MM $4B 2MM s e l a obal obal S e $3B Gl Gl s , $30B 2MM s obal obal ngs, Gl Gl ner handi, , )$2B 1MM rt a rcS rs ooki P e e Airbnb @ 5MM Listings l B - l $15B 1MM r M GM e e ( s S v s $1B 90MM 6MM ri l D Gros Gros obal oba l $0 0 $0 0 Gl G , , 2014 2015 2016 2017 2014 2015 2016 2017 s l60MM 4MM a ngs Gross Bookings Driver-Partners GMS Sellers v i t rri s A Li t s e 30MM 2MM v DoorDash @ 200K Dashers i Upwork @ 16MM Freelancers t c Gue A 250K 20MM 0 0 obal200K 2014 2015 2016 2017 2018 obal Gl 15MM , Gl Guest Arrivals Active Listings s , 150K rs her s 10MM a nce 100K a D l e Fre5MM 50K 0K 0 2014 2015 2016 2017 2014 2015 2016 2017 Lifetime Dashers Freelancers Uber Source: Uber Note: ~900K USA Uber Driver-Partners. As of 1/15, based on historical growth rates, it is estimated that >90% of USA Uber driver-partners drive for UberX. DoorDash Source: DoorDash. Note: Lifetime Dashers defined as the total number of people that have dashed on the platform, most of which are still active. Etsy Source: Etsy. Note: In 167 2017, 65% of Etsy Sellers were USA-based (1.2MM). Upwork Source: Upwork. Airbnb Source: Airbnb, Note: Airbnb disclosed in 2017 that ~660K of their listings were in USA. A 2017 CBRE study of ~256K USA Airbnb listings + ~177K Airbnb hosts in Austin, Boston, Chicago, LA, Miami, Nashville, New Orleans, New York City, Oahu, Portland, San Francisco, Seattle, & Washington D.C. found 83% of hosts are single-listing hosts / non-full-home hosts. This implies >500K USA hosts.

On-Demand Jobs = Big Numbers + High Growth Filling Needs for Workers Who Want Extra Income / Flexibility... Have Underutilized Skills / Assets 168

On-Demand Work Basics + Benefits = Extra Income + Flexibility, USA per Intuit Extra Income Flexibility 37% = Run Own Business 71% = Always Wanted To Be Own Boss 33% =Use Multiple On-Demand Platforms 46% =Want To Control Schedule Basics 26% = Employed Full-Time (W2 Wages) 19% = Responsible for Family Care 14% = Employed Part-Time (W2 Wages) 9% = Active Student 5% = Retired 57% = Earn Extra Income 91% =Control Own Schedule 21% =Make Up For Financial Hardship 50% = Do Not Want Traditional Job 19% = Earn Income While Job Searching 35% = Have Better Work / Life Balance Benefits $34 Average Hourly Income 11 Average Weekly Hours With Primary On-Demand Platform $12K Average Annual Income 37Average Weekly Hours of Work 24% Average Share of Total Income (All Types / Platforms) Source: Intuit, 2017 Note: Intuit partnered with 12 On-Demand Economy platforms which provided access to their provider email lists. (n = 6,247 respondents who had worked on-demand within the past 6 months). The survey 169 focused on online talent marketplaces. Airbnb and other online capital marketplaces were not included.

On-Demand Platform Specifics… 170

Uber = 3MM Global Driver-Partners +~50% Y/Y (2017) Uber Driver-Partners (USA = 900K)… $21 = Average Hourly Earnings 17 = Average Weekly Hours 30 = Average Trips Per Week Basics Motivations 80% =Had Job Before Starting Uber 91% =Earn Extra Income 72% =Not Professional Driver 87% =Set Own Hours 71% = Increased Income Driving Uber 85% = Work / Life Balance 66%= Have Other Job 74% =Maintain Steady Income 32% = Earn Income While Job Searching Source: Drivers + Basics = Hall & Krueger (2016) ‘An Analysis Of The Labor Market For Uber’s Driver-partners In The United States.’ Other = Cook, Diamond, Hall, List, & Oyer (2018) ‘The Gender Earning Gap in the Gig Economy’ Note: % Statistics 171 based on 12/14 survey of Uber Drivers in 20 markets that represent 85% of all USA Uber Driver-Partners

Etsy = 2MM Global Active Sellers +9% (Q1) Etsy Sellers (USA = 1.2MM)… $1.7K = Annualized Gross Merchandise Sales (GMS) per Seller $3.4B = Annualized GMS +20% (Q1) 99.9% = USA Counties with Etsy Seller(s) Basics Motivations 97% = Operate @ Home 68% = Creativity Provides Happiness 87% = Identify as Women 65% = Way to Enjoy Spare Time 58% = Sell / Promote Etsy Goods Off Etsy.com 51% = Have Financial Challenges 53% = Started Their Business on Etsy 43% = Flexible Schedule 49% = Use Etsy Income for Household Bills 30% = Use Etsy Income for Savings 32% = Etsy Sole Occupation 32% = Have Traditional Full-Time Job 28% = Operate From Rural Location 27% = Have Children @ Home 13% = Etsy Portion of Annual Household Income Source: “Etsy SEC filings + “Crafting the future of work: the big impact of microbusinesses: Etsy Seller Census 2017” Published by Etsy. Survey measured 4,497 USA-based sellers 172 on Etsy’s marketplace In 2017, 65% of Etsy Sellers were USA-based (1.2MM).

Airbnb = 5MM Global Active Listings (5/18) Airbnb Hosts (USA Listings = 600K+)… $6,100 = Average Annual Earnings per Host Sharing Space 97% = Price of Listing Kept by Hosts (9/17) 43% = Airbnb Income Used for Rent / Mortgage / Home Improvement Basics Motivations 57% = Use Earnings to Stay in Home 80%+ = Share Home in Which They Live 36% = Spend >30% of Total Income on Housing 60%+ = ‘Superhosts’ Who Identify as Women 12%= Avoided Eviction / Foreclosure 29% = Not Full-Time Employed Owing to Airbnb Earnings 18% = Retirees Source: Average Earnings + Foreclosure Avoidance = ‘Introducing The Living Wage Pledge” Airbnb (9/17), Superhost Gender Identity = ‘Women Hosts & Airbnb” (3/17), Employment Status & Earning Usage = ‘2017 Seller Census Survey’ (5/18). Note: A Superhost is an 173 Airbnb host with a 4.8+ rating, 90% response rate, 10+ stays/year, and 0 cancellations. Superhosts are marked as such on Airbnb.com

No [Uber] driver-partner is ever told where or when to work. This is quite remarkable – an entire global network miraculously ‘level loads’ on its own. Driver-partners unilaterally decide when they want to work and where they want to work. The flip side is also true – they have unlimited freedom to choose when they do NOT want to work… The Uber Network…is able to elegantly match supply & demand without ‘schedules’ & ‘shifts’… That worker autonomy of both time & place simply does not exist in other industries. - Bill Gurley – The Thing I Love Most About Uber – Above the Crowd, 4/18 174

On-Demand + Internet-Related Jobs = Scale Becoming Significant 175

DATA GATHERING + OPTIMIZATION = YEARS IN MAKING… INCREASINGLY GLOBAL + COMPETITIVE 176

Data Gathering + Optimization = Accelerates With Computer Adoption... Mainframes (Early 1950s*)… * In 1952 IBM launched the first fully electronic data processing system, the IBM 701. 177

Data Gathering + Optimization (1950s ) = Enabled by Mainframe Adoption… Mainframe Shipment Value & Units $16B 16K USA , $12B 12K e lu USA a , V t n e Units m $8B 8K e ip Sh e nfram m ai M fra$4B 4K in a M $0 0 1960 1965 1970 1975 1980 1985 1990 Shipment Value Annual Mainframe Shipments Source: W. Edward Steinmueller: The USA Software Industry: An Analysis and Interpretive History (3/95). 178

…Data Gathering + Optimization (1950s ) = Government Mainframe Deployment… 1955 1960 1965 Social Security NASA IRS Calculate Benefits for Calculate Real-Time Calculate / Store 15MM Recipients (62MM Now) Orbital Determination 55MM Records (126MM Now) th Source: Social Security Administration (75 Anniversary Retrospective), NASA – ‘Computers in Spaceflight’, CNET – “IRS Trudges on With Aging Computers” (5/08). Note: Social Security includes Americans receiving retirement benefits, old-age / survivors insurance, unemployment 179 benefits, or disability benefits. Tax records includes include total households since all are required to file taxes regardless of amount owed.

…Data Gathering + Optimization (1950s ) = Business Mainframe Deployment 1955 1965 1975 Banks Insurance Credit Cards Bank of America Aetna Visa Process Checks Optimize Insurance Policies Manage Merchant Network Telecom Airlines Retail Bell Labs / AT&T American Airlines Walmart Optimize Telephone Switching Process Transactions / Data Track Inventory / Logistics Hospitals Tulane Medical School System Manage Patient Data Source: Bank of America, IBM, Computer World (9/85), Network Computing (3/04), Computer History Museum, Walmart Museum. Note: Banks (1952): Bank of America adopted ‘Electronic Recording Method of Accounting’ system developed by Stanford Research Institute. Telecom (1955): Bell Labs 180 installed the IBM 650 to facilitate engineering for complex automated telephone switching systems. Hospitals (1959): Tulane Medical School System installed the IBM 650 to process medical record data. Airlines (1962): IBM computers integrated into SABRE system. Insurance (1965): Aetna installed IBM’s 360 to automate policy creation. Retail (1972): Walmart established a data processing facility. Credit Cards (1973): Year IBM partnered with Visa.

...Data Gathering + Sharing + Optimization = Accelerates With Computer Adoption... Consumer Mobiles + The Cloud (2006)... 181

Computing Big Bangs = Cloud (2006) + Consumer Mobile (2007)… 2006 2007 Amazon AWS Apple iPhone Until now, a sophisticated & Why run such a sophisticated scalable data storage infrastructure operating system on a mobile has been beyond the reach device? Well, because it’s of small developers. got everything we need. - Amazon S3 Launch FAQ, 2006 - Steve Jobs, iPhone Launch, 2007 Source: Wikimedia, Apple, Amazon, Steve Jobs Photo by Tom Coates. 182

…Computing Big Bangs = Cloud (2006) + Consumer Mobile (2007) Amazon AWS –# of Services Apple iOS – # of Apps 2018 2018 2MM+ Apps 140+ Services 2008 <5,000 Apps 2006 1 Service 2006 2008 2010 2012 2014 2016 2018 2008 2010 2012 2014 2016 2018 Source: Amazon, The Internet Archive. Apple; AppleInsider. Note: Based on Apple releases. Includes all iPhone/iPad/Apple TV applications available for download. Data as of 5/18. 183

...Computing Big Bangs Volume Effects = Cloud Compute Cost Declines Continue -11% vs. -10% Y/Y... AWS Compute Cost + Growth* $0.5 0% $0.4 -10% e tanc$0.3 -20% nge Ins r Cha /Y t Pe $0.2 -30% s Y Co $0.1 -40% $0 -50% 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 2018 Cost Y/Y Change Source: The Internet Archive. *Cost data reflects price of ‘current generation’ m.large on-demand Linux instance in USA- East Virginia (m1.large = 2008-2013, m3.large = 2014-2015, m4.large = 2016-2017, m5.large = 2018). m.large chosen as 184 a representative instance of general purpose compute; pricing does not account for increasing instance performance.

...Computing Big Bangs Volume Effects = Cloud Revenue Re-Accelerating +58% vs. +54% Q/Q Cloud Service Revenue –Amazon + Microsoft + Google $10B 100% $8B 75% nue e th $6B Rev 50% Grow /Y obal$4B Y Gl 25% $2B $0 0% Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 Q1 2015 2016 2017 2018 Amazon AWS Microsoft Azure Google Cloud Source: Amazon AWS = Company filings, Microsoft Azure = Keith Weiss @ Morgan 185 Stanley (4/18), Google Cloud = Brian Nowak @ Morgan Stanley (5/18). Note: Google Cloud revenue excluded in Y/Y growth rate calculation due to limited quarterly estimates.

Data Gathering + Sharing + Optimization (2006 ) = Enabled by Consumer Mobile Adoption... Smartphone Shipments 1.5B 1.0B nts me Ship al lob 0.5B G 0 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 Source: Morgan Stanley (Katy Huberty, 3/18), IDC. 186

...Data Gathering + Sharing + Optimization (2006 ) = Enabled by Social Media Adoption… Time Spent on Social Media Messages per Day 150 60B ) 135 inutes (M dia e 100 40B M 90 l Day r ia pe oc s ge on S a s s e ime T M 50 20B ly Dai obal Gl 0 0 2012 2013 2014 2015 2016 2017 Telegram Line WeChat Whatsapp (2/16) (10/17) (11/17) (7/17) Source: Global Web Index (9/17), Telegram (2/16), Line (10/17), WeChat (11/17), Whatsapp (7/17). Note: Per Global Web Index, social media time spent for Internet users aged 16-64. n = 61,196 187 (2012), 156,876 (2013), 168,046 (2014), 198,734 (2015), 211,024 (2016), 178,421 (2017).

...Data Gathering + Sharing + Optimization (2006 ) = Enabled by Sensor Pervasiveness... MEMS Sensor / Actuator Shipments Sensors + Data = In More Places Visual Shared Home 15B Navigation Transportation Temperature Google Maps Mobike Nest s 10B nt hipme S Predictive Fitness Precision Maintenance Tracking Cooking obal Samsara Motiv Joule Gl 5B 0 2012 2013 2014 2015 2016 2017 Source: IC Insights (2018), Google Maps, Mobike, Nest, Samsara, Motiv, Joule. Note: MEMS sensors and actuators includes all MEMS-based sensors (e.g., Accelerometers, Gyroscopes, etc.), but does not include optical 188 sensors, like CMOS image sensors, also includes actuators made using MEMs processes, per IC Insights.

...Data Gathering + Sharing + Optimization (2006 ) = Ramping @ Torrid Pace Information Created Worldwide (per IDC) 2025E 163ZB, 36% 180 Amount, % Structured 160 140 2020E 47 ZB, 16% 120 B) (Z 100 tes by a 80 tt Ze 2015 60 12 ZB, 9% 2010 40 2 ZB, 9% 2005 0.1 ZB 20 0 2004 2006 2008 2010 2012 2014 2016 2018E 2020E 2022E 2024E Source: IDC Data Age 2025 Study, sponsored by Seagate (4/17). Note: 1 petabyte = 1MM gigabytes, 1 zeta byte = 1MM petabytes. The grey area in the graph represents data generated, not stored. Structured data indicates data that 189 has been organized so that it is easily searchable and includes metadata and machine-to-machine (M2M) data.

Data = Can Be Important Driver of Customer Satisfaction 190

USA Internet Data Leaders = Relatively High Customer Satisfaction American Customer Satisfaction Index (ASCI) Scores (Internet Data Companies >$100B Market Capitalization, 5/18, USA) 77 = Q4:17 USA Average Amazon (E-Comemrce) Amazon E-Commerce 85 Google / Alphabet (Search) Google (Alphabet) Search 82 Facebook (Social Media) Facebook / Instagram Social Media 72 Netflix (Video) Netflix* Video 79 Booking Holdings (Lodging Inventory) Booking.com (Priceline) Lodging Inventory 78 60 65 70 75 80 85 90 2017 ASCI Score Source: American Customer Satisfaction Index (ASCI). *Netflix data from 2016, as ASCI score was not tracked in 2017. Instagram / Facebook average score used as ‘Facebook’ score. Priceline.com used as ‘Booking Holdings’ score. Note: ASCI is a tool first developed by The University 191 of Michigan to measure consumer satisfaction with various companies, brands, and industries. ACSI surveys 250K USA customers annually via email, responses to weighted questions are used to create a cross-industry score on a scale of 0-100. Top 2017 Score = 87 (Chick-fil-A).

Google Personalization = Queries… Drive Engagement + Customer Satisfaction Query Growth (2015 -2017) Data-Driven Personalization 1000% 900% 800% 2017) - 15 20 ( al 600% lob G , h t w o r G 400% y er u Q le g o 200% o G 65% 60% 0% __ Should I __ For Me __? Near Me* Source: Google (5/18). Note: Google queries only personalized for geo-location data. *Reflects mobile queries, where location data is readily available / important. 192

Spotify Personalization = Preferences… Drive Engagement + Customer Satisfaction User Preferences Spotify Daily Engagement 50% 44% al 37% lob G U, 25% A / M U DA 0% 2014 2015 2016 2017 Unique Artist Listening Data-Driven Personalization o t ed 120 112 isten al L lob 68 80 ist t r , G A r e Use40 iqu er P Un ly h t 0 n o 2014 2015 2016 2017 M Source: Spotify, Benjamin Swinburne @ Morgan Stanley (4/18) Note: Monthly unique artists listened to per user as of 5/18. 193

Toutiao Personalization = Interests… Drive Engagement + Customer Satisfaction MAUs Data-Driven Personalization 250MM al200MM b lo G150MM s, U 100MM A M 50MM 0 2015 2016 2017 Minutes Spent per Day 80 60 es t 40 inu M 20 0 Snapchat Instagram* Toutiao Main Page – User B Main Page – User A (5/18) (8/17) (5/18) Source: Toutiao (5/18), Snap (5/18), Instagram (8/17). *Instagram data reflects time spent by users under the age 194 of 25, assumed to be representative of all Instagram users.

Data = Improves Predictive Ability of Many Services 195

Data Volume = Foundational to Algorithm Refinement + Artificial Intelligence (AI) Performance… Object Detection - Performance vs. Dataset Size Google Research & Carnegie Mellon, 2017 ) 95]40 . @[.5, P (mA 35 n o Precisi n 30 ctio ete D ect j 25 Ob 10MM 30MM 100MM 10 100 300MM Example Images in Training Dataset Source: Revisiting Unreasonable Effectiveness of Data in Deep Learning Era – Sun, Shrivastava, Singh, & Gupta, 2017 Note: Chart reflects object detection performance when initial checkpoints are pre-trained on different subsets of JFT-300M tagged image 196 dataset. X-axis is the data size in log-scale, y-axis is the detection performance in mAP@[.5,.95] on “COCO minival” testing set.

…Data Volume = Foundational to Tool / Product Improvement... Artificial Intelligence (AI) Predictive Capability AWS ‘Data Flywheel’ – Amazon Rekognition* More Data Better Analytics Pricing Accuracy More Uploads = Lower Average Price Regular Improvements More Customers Better Products Features Customers Regular Improvements Large / Small Enterprises + Public Agencies Source: Amazon Artificial Intelligence on AWS Presentation (6/17). *Amazon Rekognition enables users to detect objects, people, text, scenes, and activities in their photos and videos using machine learning. 197

Artificial Intelligence (AI) Service Platforms for Others = Emerging from Internet Leaders 198

Amazon = AI Platform Emerging from AWS… Enabling Easier Data Processing / Collection for Others… Amazon AWS AI Services / Infrastructure Rekognition Image Recognition AI Hardware – Scalable GPU Compute Clusters ComprehendLanguage Processing SageMaker Machine Learning Framework Source: Amazon. AWS = Amazon Web Services. 199

...Google = AI Platform Emerging from Google Cloud… Enabling Easier Data Processing / Collection for Others Google Cloud AI Services / Infrastructure Google Cloud Vision API AI Hardware – Tensor Processing Units Dialogflow Conversational Platform Cloud AutoML – Custom Models Source: Google 200

AI in Enterprises = Small But Rapidly Rising Spend Priority… Per Morgan Stanley CIO Survey (4/18 vs 1/18) Which IT Projects Will See The Largest Spend Increase in 2018? 10% . U . E + A S U s, t en d n o 5% p es R O I C f o are h S 0% Networking Equipment Artificial Intelligence Hyperconverged Infrastructure January 2018 April 2018 Source: AlphaWise, Morgan Stanley Research. Note: n = 100 USA / E.U. CIOs. Note: Full Question Text = ‘Which three External IT Spending projects will see the largest percentage increase in spending in 2018?’ 201

AI is one of the most important things humanity is working on. It is more profound than electricity or fire… We have learned to harness fire for the benefits of humanity but we had to overcome its downsides too. …AI is really important, but we have to be concerned about it. - Sundar Pichai, CEO of Google, 2/18 Source: CNBC (2/18). 202

Data Sharing = Creates Multi-Faceted Challenges 203

Data + Consumers = Love-Hate Relationship Source: Cartoonstock, Artist: Roy Delgado 204

Most Online Consumers Share Data for Benefits… USA Consumers per Deloitte 79% Willing to Share Personal Data For ‘Clear Personal Benefit’ >66% Willing To Share Online Data With Friends & Family Source: USA Consumer Data = Deloitte To share or not to share (9/17) Note: n = 1,538 USA customers surveyed in cooperation with SSI in 2016. 205

…Most Online Consumers Protect Data When Benefits Not Clear Consumers Taking Action To Address Data Privacy Concerns Deleted / Avoided Certain Apps 64% Adjusted Mobile Privacy Settings 47% Disabled Cookies 28% Didn't Visit / Closed Certain Websites 27% Closely Read Privacy Agreements 26% Did Not Buy Certain Product 9% 0% 25% 50% 75% % of Respondents that Took Action in the Last 12 Months Due to Data Privacy Concerns, USA Source: Deloitte To share or not to share (9/17) Note: n = 1,538 USA consumers in cooperation with SSI. 206

Internet Companies = Making Consumer Privacy Tools More Accessible (2018) Facebook Google 2008 2018 2008 2018 Source: Facebook, Google 207

Data Sharing = Varying Views 208

EU / Asia / Americas = Rising Regulatory Focus on Data Collection + Sharing… General Data Protection Regulation Enacted = 5/25/18 Act on Protection of Personal Information Privacy Act of 1974 Enacted = 5/30/17 Enacted = 12/31/74 Personal Information Protection Act Enacted = 09/30/11 Data Privacy Laws Enacted in Past 10 Years Developing (2018) Source: Wikimedia, USA Congress, EU, Japan Government, South Korea Government, Argentina Government. Note: Argentina proposed a 2017 draft amendment to the Personal Data Protection Act that would strengthen current regulation 209 and align with most GDPR requirements. Japan enacted an amendment to its Act on Protection of Personal Information that went into effect on 5/30/17. All EU countries grouped due to passage of EU-wide GDPR laws.

...China = Encouraging Data Collection [Xi Jingping] called for building high-speed, mobile, ubiquitous & safe information infrastructure, integrating government & social data resources, & improving the collection of fundamental information... [Xi stated] The Internet, ‘Big Data,’ Artificial Intelligence, & ‘The Real Economy’ should be interconnected. - Xinhua State News Agency, 12/9/17 Ministry of Industry & Information China to Further Promote Government Training to Build ‘Big Data’ Datacenter Information Sharing & Disclosure Xinhua State Press Agency, 5/07/17 Xinhua State Press Agency, 12/7/17 China Launches ‘Big Earth’ Big Data Project To Boost Science Data Sharing Xinhua State Press Agency, 2/13/18 Source: Xinhua (PRC’s official Press Agency). . 210

Cybersecurity = Threats Increasingly Sophisticated...Targeting Data Observed Malware Volume 15x Adversaries are taking malware to al unprecedented levels of ob l G sophistication & impact... 1:16), 10x Q ( Weaponizing cloud services & other o t ed technology used for legitimate purposes... ex nd I ( And for some adversaries, 5x the prize isn’t ransom, but olume V e obliteration of systems & data. ar w al M - Cisco 2018 Annual Cybersecurity Report, 2/18 0 Q1 Q2 Q3 Q4 Q1 Q2 Q3 Q4 2016 2017 Source: Cisco 2018 Annual Cybersecurity Report. Note: Data collected by Cisco endpoint security equipment / software. While Malware volume increased ~11x from Q1:16 to Q4:17, traffic events process by the same equipment only increased ~3x. 211

Global Internet Leadership = USA & China 212

Economic Leadership... 213

Relative Global GDP (Current $) = USA + China + India Gaining…Other Leaders Falling Global GDP Contribution (Current $) 40% 40% 30% GDP 26% l 25% loba 22% 20% G of % 15% 10% 6% 3% 4% 7% 3% 0% 1960 1970 1980 1990 2000 2010 2017 USA Europe China India Latin America Source: World Bank (GDP in current $). Other countries account for ~30% of global GDP. 214

Cross-Border Trade = Increasingly Important to Global Economy Trade as % of Global GDP 40% 30% are 20% Sh 10% 0% 2016 1960 1965 1970 1975 1980 1985 1990 1995 2000 2005 2010 2015 Source: World Bank. Note: ‘World Trade’ refers to the average of Imports & Exports (to account for goods in-transit between years) for all nations. 215

Internet Leadership = A Lot’s Happened Over 5-10 Years… 216

Today’s Top 20 Worldwide Internet Leaders 5 Years Ago* = USA @ 9…China @ 2… Public / Private Internet Companies, Ranked by Market Valuation (5/29/18) Rank Market Value ($B) 2018 Company Region 5/29/13 1) Apple USA $418 2) Amazon USA 121 3) Microsoft USA 291 4) Google / Alphabet USA 288 5) Facebook USA 56 6) Alibaba China -- 7) Tencent China 71 8) Netflix USA 13 9) Ant Financial China -- 10) eBay + PayPal** USA 71 11) Booking Holdings USA 41 12) Salesforce.com USA 25 13) Baidu China 34 14) Xiaomi China -- 15) Uber USA -- 16) Didi Chuxing China -- 17) JD.com China -- 18) Airbnb USA -- 19) Meituan-Dianping China -- 20) Toutiao China -- Total $1,429 Source: CapIQ, CB Insights, The Wall Street Journal, media reports. *Only includes public companies in 2013. **eBay + PayPal combined for comparison purposes though PayPal spun-off of eBay on 7/20/15. 217

…Today’s Top 20 Worldwide Internet Leaders Today = USA @ 11…China @ 9 Public / Private Internet Companies, Ranked by Market Valuation (5/29/18) Rank Market Value ($B) 2018 Company Region 5/29/13 5/29/18 1) Apple USA $418 $924 2) Amazon USA 121 783 3) Microsoft USA 291 753 4) Google / Alphabet USA 288 739 5) Facebook USA 56 538 6) Alibaba China -- 509 7) Tencent China 71 483 8) Netflix USA 13 152 9) Ant Financial China -- 150 10) eBay + PayPal* USA 71 133 11) Booking Holdings USA 41 100 12) Salesforce.com USA 25 94 13) Baidu China 34 84 14) Xiaomi China -- 75 15) Uber USA -- 72 16) Didi Chuxing China -- 56 17) JD.com China -- 52 18) Airbnb USA -- 31 19) Meituan-Dianping China -- 30 20) Toutiao China -- 30 Total $1,429 $5,788 Source: CapIQ, CB Insights, Wall Street Journal, media reports. *eBay + PayPal combined for comparison purposes though PayPal spun-off of eBay on 7/20/15. Market value data as of 5/29/18. The Wall Street Journey, Recode, TechCrunch, Reuters, and the Information articles detail the latest 218 valuations for Ant Financial (4/18), Xiaomi (5/18), Uber (2/18), Didi Chuxing (12/17), Airbnb (3/17), Meituan-Dianping (10/17), and Toutiao (12/17).

Smartphones = China @ #1 Worldwide OEM... @ 40% vs. 0% Share Ten Years Ago...USA @ 15% vs. 3% Worldwide New Smartphone Shipments by OEM Headquarters 1.5B 60% s 40% nt e 1.0B 40% r ha S hipme S M E ide O na orldw 0.5B 20% Chi W 0% 0 0% 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 China USA Korea Other % from China OEMs Source: Katy Huberty @ Morgan Stanley (3/18), IDC. Note: OEM = Original Equipment Manufacturer. 219

Internet Globally = USA Platforms = Lead User Numbers… Active Users By Platform 2.5B 2.2B 2.0B 2.0B obal Gl orm,1.5B tf la P by 1.0B 1.0B rs Use 0.7B e tiv 0.5B c A 0 Facebook Google Tencent Alibaba (All Platforms) (Android) (WeChat) (E-Commerce) Source: Facebook (4/18), Google (5/17), Tencent (3/18), Alibaba (5/18). Note: Facebook = MAUs, Google = MAUs, Tencent WeChat = MAUs, Alibaba = Mobile MAUs. 220

…Internet by Country = China Platforms = Lead User Numbers…in China Active Users By Platform 2.5B 2.2B 2.0B 2.0B obal Gl orm,1.5B tf la P by 1.0B 1.0B rs 0.7B Use e tiv 0.5B c A 0 Facebook Google Tencent Alibaba (All Platforms) (Android) (WeChat) (E-Commerce) China Asia (ex-China) Europe North America Rest Of World Source: Hillhouse Capital. Facebook (4/18), Google (5/17), Newzoo (Google Android USA estimate, 1/18), Tencent (3/18), Alibaba (5/18). Note: Facebook = MAUs, Google = MAUs (Newzoo Global Mobile Market Report estimates that there are 125MM active Android smartphones in the USA in 2017), Tencent WeChat = monthly active accounts 221 vs. users as many Chinese users have multiple accounts (ex. 688MM users sent red envelopes during the 2018 Chinese New Year), Alibaba = Annual active consumers. Estimated WeChat ex-China MAU <5% of total per Hillhouse. Estimate Alibaba ex-China annual active consumers (Lazada + Aliexpress) = 80MM annual active customers per Hillhouse.

China Feature + Data-Rich Internet Platforms = Largest # of Users in One Country Tencent Alibaba WeChat + WeChat Pay TaoBao + Alipay Photos…Friends…Games… Searches…News…Brands… Apps…Finances…Bills… Feedback…Finances...Bills… Source: Tencent, Alibaba 222

China Internet Users = More Willing to Share Data for Benefits vs. Other Countries per GfK Would you share personal data (financial, driving records, etc.) for benefits (e.g., lower cost, personalization, etc.)? China China 38% Mexico 30% Russia 29% Italy 28% Global Global 27% Brazil 26% USA 25% USA South Korea 20% Australia 17% UK 16% Spain 16% France 15% Canada 14% Germany 12% Netherlands 12% Japan 8% 0% 10% 20% 30% 40% % of Global Respondents Very Willing to Share (6 or 7 on 7 Point Scale) Source: GfK Survey (1/17). Note: n = 22K of internet users ages 15+. A scale of 1-7 were used to identify the level of agreement with the following statement: “I am willing to share my personal data (health, financial, driving records, energy use, etc.) in exchange for benefits or 223 rewards like lower costs or personalized service” – using a scale where “1” means “don’t agree at all” and “7” means “agree completely.” .

China Digital Data Volume @ Significant Scale & Growing Fast = Providing Fuel for Rapid Artificial Intelligence Advancements 224

Artificial Intelligence = USA & China… 225

Artificial Intelligence Competition = Increasingly Complex Tasks…China Momentum Strong 1985 1995 2005 2018 th 6 World Computer Large Scale Visual Chess Championship Recognition Challenge 2010 1) Deep Thought (USA) 1) NEC-UIUC* (USA + Japan) 2) Bebe (USA) 2) XRCE (France / EU) 3) Cray Blitz (USA) 3) University of Tokyo (Japan) th China = No Entrants China = 11 Place RoboCup-99 Soccer Stanford Question Answering Simulation League Dataset (Ongoing) 1) Google + Carnegie Mellon (USA) 1) CMUnited-99 (USA) 2) Microsoft* + NUDT (USA + China) 2) MagmaFreiburg (Germany) 3) YUANFUDAO (China) 3) Essex Wizards (UK) 4) HIT + iFLYTEK (China) China = No Entrants 5) Alibaba (China) Source: International Computer Games Association, RoboCup, Image-Net, Stanford. Note: Stanford Question Answering Dataset is a set of 100,000+ human-generated questions covering 500+ Wikipedia articles. Scores ranked by Exact Match Accuracy, which refers to the share of questions correctly parsed / answered. Highest Score 226 included for teams with multiple results (i.e. Google + Carnegie Melon) *National Affiliation refers to main campus of sponsoring Group / Company / University. Microsoft submitting team based in Beijing (lead by Feng-Hslung Hsu who was lead developer of ‘Deep Thought’ while @ Carnegie Mellon). NEC team based in USA.

Natural Science & Engineering Higher Education = China Graduation Rates Rising Rapidly per National Science Foundation Annual Natural Science & Engineering Degrees (Agricultural Sciences / Biological Sciences / Computer Sciences / Earth, Atmospheric & Ocean Sciences / Mathematics / Engineering) First University Doctoral (Bachelor’s Equivalent) es 1.5MM 60K omies mi on no c E co / E s / ie r es nt ri t n 1.0MM 40K ou d Cou C e t c le e S ected , el s S e 0.5MM 20K gre rees, De eg ity D s r e orate t Univ t 0 oc 0 s D 2000 2002 2004 2006 2008 2010 2012 2014 2000 2002 2004 2006 2008 2010 2012 2014 Fir China USA EU –Top 8 Japan Source: USA National Science Foundation analysis of National Bureau of Statistics (China), Government of Japan, UNESCO, OECD, National Center for Education Statistics, IPEDS, & National Center for Science / Engineering data. Note: Data for the majority of the countries were collected under same OECD, EU, and UIS guidelines & field groupings in the ISCED-F are similar to fields used in China, a major degree producer. Natural sciences include agricultural sciences; biological sciences; computer sciences; earth, atmospheric, and ocean sciences; & mathematics. EU-Top 8 for doctoral degrees includes UK / Germany / France / Spain / Italy / Portugal / Romania / Sweden. EU-Top 8 for first university degrees includes UK / Germany / France / Poland / Italy / Spain / Romania / The Netherlands. The # of S&E doctorates awarded rose from about 8K in 2000 to more than 34K in 2014. Despite the growth in the quantity of doctorate recipients, some question the quality of the doctoral programs in China (Cyranoski et al. 2011). The rate of growth in doctoral degrees in 227 S&E and in all fields has considerably slowed starting in 2010, after an announcement by the Chinese Ministry of Education indicating that China would begin to limit admissions to doctoral programs & focus on quality of graduate education (Mooney 2007). Also in China, first university degrees increased greatly in all fields, with a larger increase in non-S&E than in S&E fields. China experienced an increase of almost 1.2MM degrees and up more than 400% from 2000 to 2014. China has traditionally awarded a large proportion of its first university degrees in engineering, but the percentage declined from 43% in 2000 to 33% in 2014.

Artificial Intelligence Focus = China Government Highly Focused on Developing AI Artificial Intelligence - Next Generation Development Plan Goals 1) Build Open & Coordinated AI Innovation Systems 2) Foster a Highly Efficient Smart Economy 3) Construct Safe / Convenient Intelligent Society 4) Strengthen Military-Civilian Integration in AI 5) Build Safe & Efficient Information Infrastructure 6) Plan Next Generation AI Science & Technology Projects Source: New America Translation of China State Council documents (7/20/17). 228

Artificial Intelligence = USA Ahead… China = Focused + Organized + Gaining I’m assuming that [USA’s] lead [in Artificial Intelligence] will continue over the next five years, & that China will catch up extremely quickly. In five years we’ll kind of be at the same level, possibly. It’s hard to see how China would have passed us in that period, although their rate of improvement is so impressively good. - Eric Schmidt, Chairman, US Defense Innovation Advisory Board, Keynote Address at Artificial Intelligence & Global Security Summit, 11/13/17 229

ECONOMIC GROWTH DRIVERS = EVOLVE OVER TIME… 230

Century Economic Growth Drivers th Pre-18 Cultivation & Extraction th 19-20 Manufacturing & Industry st 21 … Compute Power & Human Potential 231

Lifelong Learning = Crucial in Evolving Work Environment & Tools Getting Better + More Accessible 232

Lifelong Learning = 33MM Learners +30% (Coursera)… Top Courses, 2017 Learners 40MM Machine Learning Stanford Neural Networks & Deeper LearningDeeplearning.ai obal Gl Learning How to Learn: Powerful Mental Tools to , 30MM Help You Master Tough Subjects UC San Diego rs Introduction to Mathematical ThinkingStanford rne a Bitcoin & Cryptocurrency Technologies Princeton Le20MM d Programming for Everybody University of Michigan tere Algorithms, Part I Princeton s 10MM English for Career Development University of Pennsylvania Regi Neural Networks / Machine LearningUniversity of Toronto Financial Markets Yale 0 2014 2015 2016 2017 Learners by Geography 30% 28% 20% 11% 5% 0% 20% 40% 60% 80% 100% North America Asia Europe South America Africa Source: Coursera. Note: Course popularity based on average daily enrollments. Graph shows learners as of 5/18. 233

…Lifelong Learning = Educational Content Usage Ramping Fast (YouTube)… Selected Education Channel Subscribers 1B 9MM Daily Learning Video Views +6MM +7 +6 70% 6MM +6 Viewers Use Platform to Help Solve ers Work / School / Hobby Problems crib +2 Subs 3MM +38% Growth Y/Y (2017) Job Search Video Views 0 (e.g., Resume-Writing Guides) Asap Crash TED- Smarter Khan SCIENCE Course Ed Every Academy Day 2013 2018 Source: YouTube (5/18). 234

…Lifelong Learning = Employee Re-Training Engagement High (AT&T)… ‘Workforce 2020’ / ‘Future Ready’ Programs $1B Allocated for web-based employee training. Partners = Coursera / Udacity / Universities. 2.9MM Emerging tech courses completed by employees. Most popular courses = Cyber Security / Machine Learning / Data-Driven Decision Making / Virtual Collaboration. 194K Employees (77% of workforce) actively engaged in re-training. 61% Share of promotions received by re-trained employees (2016-Q1:18) Source: AT&T (4/18). 235

…Lifelong Learning = >50% of Freelancers Updated Skills Within Past 6 Months When Did You Last Participate in Skill-Related Training? 75% 70% 55% ts en 50% 45% nd po Res 30% USA 25% of % 0% Feelancers Non-Freelancers Within Past 6 Months >6 Months Ago / Never Source: Edelman Research / Upwork ‘Freelancing In America: 2017.’ Note: Survey conducted July-August 2017 on 2,173 Freelance Employees who have received 236 payment for supplemental temporary, or project-oriented work in the past 12 months.

CHINA INTERNET = ROBUST ENTERTAINMENT + RETAIL INNOVATION *Disclaimer – The information provided in the following slides is for informational and illustrative purposes only. No representation or warranty, express or implied, is given and no responsibility or liability is accepted by any person with respect to the accuracy, reliability, correctness or completeness of this Information or its contents or any oral or written communication in connection with it. Hillhouse Capital may hold equity stakes in companies mentioned in this section. A business relationship, arrangement, or contract by or among any of the businesses described herein may not 237 exist at all and should not be implied or assumed from the information provided. The information provided herein by Hillhouse Capital does not constitute an offer to sell or a solicitation of an offer to buy, and may not be relied upon in connection with the purchase or sale of, any security or interest offered, sponsored, or managed by Hillhouse Capital or its affiliates.

China Macro Trends = Strong 238

China Consumer Confidence = Near 4 Year High… Manufacturing Index = Expanding China Consumer Confidence Index + Manufacturing Purchasing Managers' Index (PMI) 140 54 a n i h C130 53 ex, d n I120 52 a n i ce h C en , 110 51 I d fi n PM o C 100 50 mer su 90 49 n o C 80 48 2014 2015 2016 2017 2018 China Consumer Confidence Index (LHS) China Manufacturing PMI (RHS) Source: China National Bureau of Statistics (CNNIC), Morgan Stanley Research. Note: The Purchasing Managers Index is Measured by China National Bureau of Statistics Based on New Orders, Inventory Levels, Production, Supplier Deliveries & 239 the Employment Environment. Score of 50+ Indicates an Expanding Manufacturing Sector. Consumer Confidence is a Measure of Consumers’ Sentiment About the Current / Future State of the Domestic Economy, Indexed to 100.

China GDP Growth = Increasingly Driven by Domestic Consumption… @ 62% vs. 35% of GDP Growth (2003） China Domestic Consumption Contribution to GDP Growth 80% a n i h C , 62% th w60% o Gr P GD 40% to n o 35% ti u b i tr 20% n o C % 0% 2003 2006 2009 2012 2015 2018E Source: China National Bureau of Statistics, Morgan Stanley Research. Note: Domestic Consumption Includes Household and Government Consumption. Other Drivers of GDP 240 Growth Include Investments (Gross Capital Formation) and Net Export of Goods and Services.

China Internet Usage = Accelerating 241

China Mobile Internet Users = 753MM…+8% vs. 12% Y/Y China Mobile Internet Users vs. Y/Y Growth 800MM 80% a n i h C 600MM 60% s, ser Y U Y/ et , n h t 400MM 40% w ter o n I Gr e l i b o 200MM 20% M 0 0% 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017 China Mobile Internet Users Y/Y Growth Source: China Internet Network Information Center (CNNIC). Note: Mobile Internet User Data is as of Year-End. 242

China Mobile Internet (Data) Usage = Accelerating…+162% vs. +124% Y/Y China Cellular Internet Data Usage & Growth Y/Y 30EB 180% te) y Exab = 20EB 120% (EB a n Y i h Y/ C th e, w o sag 10EB 60% Gr U ata D e l i b o 0 0% M 2012 2013 2014 2015 2016 2017 China Mobile Data Consumption Y/Y Growth Source: China Ministry of Industry and Information Technology. Note: Cellular Internet Refers to 3G/4G Network data. 243

China Online Entertainment = Long + Short-Form Video & Team-Based Multiplayer Mobile Games Growing Quickly 244

China Mobile Media / Entertainment Time Spent = +22% Y/Y…Mobile Video Growing Fastest China Mobile Media / Entertainment Daily Time Spent March 2016 March 2018 2.0B Hours 3.2B Hours, +22% Y/Y Audio News 4% Game 11% 13% Reading 3% Social Networking Video Video 60% 13% 22% Game Social 13% Networking 47% Reading News Audio 3% 7% 4% Source: QuestMobile (3/18). 245

China Short-Form Video = Usage Growing Rapidly… China Daily Mobile Media Time Spent 500MM s 400MM r Hou Media300MM e Mobil200MM y il Da 100MM 0 2016 2017 2018 Long Form Video Short Form Video Live & Game Streaming Long-Form Video Short-Form Video Source: QuestMobile (3/18). Note: Short-Form Videos are typically <5 Minute in Length and include companies such as Kuaishou, Douyin, Xigua. Long-Form 246 Videos include companies such as iQiyi, Tencent Video, Youku, Bilibili.

…China Short-Form Video Leaders = 100MM+ DAU… Massive Growth + High Engagement (50 Minute Daily Average) Douyin (Tik-Tok) Kuaishou AI-Augmented Mobile Video Creation De-Centralized / Personalized / / Personalized Feed Location-Based Mobile Video Discovery DAU = 95MM +78x Y/Y DAU = 104MM +2x Y/Y Daily Time Spent = 52 Minutes Daily Time Spent = 52 Minutes DAU / MAU Ratio = 57% DAU / MAU Ratio = 46% Source: QuestMobile (4/18). 247

China Online Long-Form Video Content Budgets = Exceeded TV Networks (2017)… China TV Networks* vs. Online Video Platform** Content Budget $8B et g d $6B u B t ten n o $4B C al u n n A $2B $0 2009 2010 2011 2012 2013 2014 2015 2016 2017 TV Networks* Online Video Platforms** Source: Public disclosures, Goldman Sachs, Bank of America, Hillhouse estimates. *Includes estimates from CCTV, provincial satellite TV channels 248 and major local TV networks. **Includes iQiyi + Tencent Video + Youku.

…China Online Long-Form Video Original / Exclusive Content = Driving Industry-Wide Paying Subscriber Growth Original / Exclusive Content iQiyi Paying Subscribers 75MM s er b i 50MM scr b Su g n i Pay 25MM 0 2015 2016 2017 2018 Source: Subscriber data per iQiyi (3/18). Tencent Video and Youku are not standalone publicly listed companies hence do not provide regular disclosure on paying 249 subscribers. Tencent Video last announced more than 62MM subscribers in 2/18.

China Team-Based Multiplayer Mobile Games = Lead Game Time Spent in China Honor of Kings China Mobile Games Daily Hours 80MM+ China DAU 320MM a n i h C t, 240MM en Sp s r u o 160MM H e l PUBG Mobile i b 50MM+ China DAU o M y 80MM l ai D 0 2016 2017 2018 MOBA / FPS / Survival Casual Other*** Source: Questmobile (3/18). *MOBA is Multiplayer Online Battle Arena; **FPS is First Person Shooting, FPS / Survival games include Tencent’s PUBG Mobile and NetEase’s Rules of 250 Survival. ***Other genre includes RPG, action, racing, strategy, card battle, and other games.

Global Interactive Game Revenue = China #1 Market in World* > USA (2017) Interactive Game Software Revenue $30 ) ($B e u en ev R $20 e) ar w d ar H . $10 (excl Game eo d $0 i V 2012 2013 2014 2015 2016 2017 China USA EMEA Asia (ex. China) Source: Newzoo. *Excludes console / gaming PC hardware revenue. 251

China Retail Innovation = Spreading from Online to Offline 252

Worldwide E-Commerce Share Gains Continue… China @ 20% = Highest Penetration Rate + Fastest Growing E-Commerce % of Retail Sales 25% 20% 15% e Shar 10% 5% 0% 2003 2005 2007 2009 2011 2013 2015 2017 Korea UK China USA Germany Japan France Brazil Source: Euromonitor. Note data excludes certain consumer-to-consumer (C2C) transactions. 253

China E-Commerce = Strong Growth +28% Y/Y… Mobile = 73% of GMV ChinaB2CE-Commerce GrossMerchandiseValue $600B 300% a n i h C , Y V $400B 200% Y/ GM th w ce o Gr e l mmer i o b C o - $200B 100% M E 2C B $0 0% 2013 2014 2015 2016 2017 Desktop Mobile Mobile Growth Y/Y Source: iResearch. Note: Assumes constant USD / RMB rate = 6.9. 254

Hema Stores = Re-Imagining Grocery Retail Experience… High Quality + Convenience + Digital… Digital Grocery Store Restaurant Real-Time E-Commerce SKU Selection = Cook To Order Chefs / Ceiling-Conveyor System / Based on Customer Data.. Eat-in-Shop In-Store Fulfillment / Alipay Membership To Pay 30-Minute Delivery Source: Hillhouse. Note: As of 4/18, there are 37 Hema stores in China. 255

…Hema Stores= Material Portion of Orders Online… Driving Higher Sales Productivity vs. Offline Peers Daily Retail Transactions per Store, 11/17 12,000 e r Sto er 8,000 p s n o sacti an r 4,000 T y l ai D 0 Carrefour China Yonghui Hema Offline Online Source: Bernstein Research. Note: Hema data points in chart came from stores in Shanghai and Hangzhou in 11/17. In Q1:18, more than 50% of Hema store orders were placed online for home delivery. 256

Belle = Re-Imagining Offline Retail Experience with Online Analytics RFID in Shoes / Traffic Heat Map 3D Foot Scan Floor Mat Optimize Layout Conversion Analysis Personalization 138 fittings / 37 sales 27% conversion 168 fittings / 5 sales 3% conversion Source: Belle, Hillhouse Capital. 257

China Online Payments / Advertising / On-Demand Transportation = Growing Rapidly 258

China Mobile Payment Volume = +209% vs. +116% Y/Y Led by Alipay + WeChat Pay China Mobile Payment Volume China Mobile Payment Share* $16T Others a 8% n i h C $12T me, u l o V t $8T WeChat men Pay AliPay 38% 54% Pay e l i b o $4T M $0 2012 2013 2014 2015 2016 2017 Source: Analysys (Q1:18, 3/18). *Excludes certain P2P and 259 transfer payments. Assumes constant USD / RMB rate = 6.9.

China Online Advertising Revenue = +29%vs. 29% Y/Y ChinaOnlineAdvertising Revenue $50B 50% $40B 40% a n i h Y C$30B 30% Y/ , g th n w si o ti $20B 20% er Gr v d A e n$10B 10% i l On $0 0% 2012 2013 2014 2015 2016 2017 Online Advertising Y/Y Growth Source: iResearch. Note: Assumes constant USD / RMB rate = 6.9. . 260

China On-Demand Transportation (Cars + Bikes) = +96%... 68% Global Share & Rising On-Demand Transportation Trip Volume by Region 8B al ROW b o Gl 6B SE Asia s, p i r T India eted l 4B EMEA mp o C y N. America l ter 2B ar China Bike China Bike Qu China Car China Car 0 Q1:13 Q1:14 Q1:15 Q1:16 Q1:17 Q1:18 Source: Hillhouse Capital estimates. Note: Includes on-demand taxi, private for-hire vehicles, 261 as well as on-demand for-hire motorbike and bike trips booked through smartphone apps.

ENTERPRISE SOFTWARE = USABILITY / USAGE IMPROVING 262

Consumer-Like Apps = Changed Enterprise Computing 263

Dropbox (2007) = Pioneered… Consumer-Grade Product With Enterprise Appeal… Dropbox synchronizes files across your / your team's computers…files are securely backed up to Amazon S3. It takes concepts that are proven winners from the dev community & puts them in a package that my little sister can figure out… Competing products force the user to constantly think & do things… With Dropbox, you hit "Save," as you normally would & everything just works. - Drew Houston, Founder, Y Combinator Application, Summer 2007 264

...Dropbox = Pioneered... Consumerization of Enterprise Software Business Model Inflection Points Users 500MM 4% e ar 2008 = Consumer / Individual h 2.2% S 1.7% s r 250MM 2% Free Premium Features for Referral Launch… e ing 8 Months to 1MM Users • Us ay P 0 0% 2013 = Enterprise / Team 2015 2016 2017 Dropbox for Business Launch… Paying Users Other Registered Users 30% = Dropbox Business Share of Paid Users (2018) Paying Share 2015 = Revenue / Sales Efficiency Revenue & Gross Margin Free-to-Pay User Conversion Launch… 90% = Revenue From Self-Serve Channels (2018)… $1.5B 80% >40% = New Teams with Former Individual Paid User (2018) 67% in e $1.0B arg 2018 = Platform M 33% enu 40% ss Integrated Product Suite Launch… $0.5B Rev ro 3= Major Product Launches Since 2017* G $0 0% 2015 2016 2017 Revenue Gross Margin Source: Ilya Fushman @ Kleiner Perkins. Dropbox, Techcrunch, JMP Securities estimates of Dropbox public releases of registered users. *Major products = Paper, Showcase, & Smart Sync. 265

Slack (2013) = Pioneered… Enterprise-Grade Product With Consumer Look & Feel... When you want something really bad, you will put up with a lot of flaws. But if you do not yet know you want something, your tolerance will be much lower. That’s why it is especially important for us to build a beautiful, elegant and considerate piece of software. Every bit of grace, refinement, & thoughtfulness on our part will pull people along. Every petty irritation will stop them & give the impression that it is not worth it. - Stewart Butterfield, Slack Founder / CEO (2013) 266

...Slack = Pioneered… Consumerization of Enterprise Software Business Model Slack Inflection Points Slack Daily Active Users 8MM 40% 2013 = Small Teams Consumer-Like Onboarding Launch… 128K Users 6 Months Post-Launch (2014) s 6MM U 34% A 2015 = Platform D rd 3 -Party App Directory Launch… rs of se % >1.5K Apps in Slack App Directory (2018) U as >200K Developers on Slack Platform (2018) e 4MM 30% rs ctiv se A U 2015 = Revenue / Sales Efficiency ly ai Free-to-Pay User Conversion Launch… 26% ing D y >400% = 2015 Y/Y Paid Subscription Growth Pa 2MM 2017 = Enterprise / Large Teams Enterprise Features Plan Launch… >70K = Paid Teams (2018)… 0 20% >500K = Organizations Using Slack (2018) 2014 2015 2016 2017 >150 = Large Enterprises Using Slack Grid (2018) Paying Users Other Active Users Paying Share Source: Slack. 267

Enterprise Software Success Formula Build Amazing Consumer-Grade Product Leverage Virality Across Individual Users To Grow Personal + Professional Adoption @ Low Cost Harvest Individual Users for Enterprise Go-to-Market With Dedicated Product + Inside / Outbound Sales Build Enterprise-Grade Platform + Ecosystem Net = Low Cost Product-Driven Customer Acquisition + Strong / Sticky Business Model - Ilya Fushman @ Kleiner Perkins Source: Ilya Fushman @ Kleiner Perkins. 268

Messaging Threads = Transforming Collaboration... Distributing + Increasing Productivity 269

Messaging Threads = Increasingly Foundational for Consumers + Enterprises Consumer Services… …Enterprise Services Snapchat Square Cash Dropbox Slack Social Payments File Management Communication Intercom Strava Customer Interactions Workouts Source: Snapchat, Square, Strava, Dropbox, Slack, Intercom. 270

Google Set Out to… ‘Organize the World’s Information & Make It Universally Accessible & Useful’ Now Apps... Organize Business Information & Make It Accessible & Useful Within Enterprises 271

Enterprise Messaging Threads = Organizing Information + Teams… Providing Context + History... 272

Slack = Communication Threads... Organizing Information by Channel Topic… Slack Benefits Slack Daily Active Users • 32% Decline in Email Usage 8MM • 24% Reduction in Employee Onboarding rs6MM Time Use e tiv • 23%Faster Time to Market For c A 4MM Development Teams ly Dai 2MM • 23% Decline in Meetings • 10% Rise in Employee Satisfaction 0 2013 2014 2015 2016 2017 Source: Slack (5/18), IDC “The Business Value of Slack” research report (2017). 273

...Dropbox = File Management Threads... Organizing Data by File + Version Dropbox Benefits Teams % of Paid Users 32% • 6x Rise in Employees on 30% Multi-Department Teams rs Use 28% id • 31%Decline in IT Time Spent a P Supporting Collaboration l a ot26% T of • 3.7K Hours Saved Annually Per % , 24% Organization in Document Management ms a e T 22% • 6% Rise in Sales Team Productivity 20% 2014 2015 2016 2017 2018 Source: Dropbox. Piper Jaffray (4/18, Teams % of paid users). Dropbox + IDC commissioned study for Dropbox on effects of enterprises using Dropbox (Dropbox benefits, 2016). 274

...Zoom = Visual Communication / Meeting Threads... Distributing + Increasing Productivity… Zoom Benefits Annualized Meeting Minutes • 85% Improved Collaboration 40B • 71% Improved Productivity 30B obal • 62% Supported Flexible Work Schedule Gl , • 58% Built Trust Among Remote Workers inutes 20B M d • 58% Reduced Meeting Times ize • 48% Removed Company Silos nnual10B A • 72 Net Promoter Score 0 2015 2016 2017 2018 Source: Survey conducted by Zoom Video Communications of Zoom customers +700 responses (2/18). 275

...Intercom = Customer Transaction Threads... Organizing Customer Dialog Intercom Benefits • 82% Rise in Conversion For Customers Chatting In Intercom • 36% Rise in Conversion For Customers Assisted by ‘Operator’ Chatbot • 13% Rise in Order Value for Customers Chatting in Intercom Source: Intercom. 276

…Enterprise Messaging Threads = Helping Improve Productivity + Collaboration 277

USA INC.* = WHERE YOUR TAX DOLLARS GO * USA, Inc. Full Report: http://www.kleinerperkins.com/blog/2011-usa-inc-full-report 278

USA Income Statement = -19% Average Net Margin Over 30 Years… USA Income Statement F1987 F1992 F1997 F2002 F2007 F2012 F2017 Comments Revenue ($B) $854 $1,091 $1,579 $1,853 $2,568 $2,449 $3,316 +5% Y/Y average over 25 years Y/Y Growth 11% 3% 9% -7% 7% 6% 2% Individual Income Taxes* $393 $476 $737 $858 $1,163 $1,132 $1,587 Largest Driver of Revenue % of Revenue 46% 44% 47% 46% 45% 46% 48% Social Insurance Taxes $303 $414 $539 $701 $870 $845 $1,162 Social Security & Medicare Payroll Tax % of Revenue 36% 38% 34% 38% 34% 35% 35% Corporate Income Taxes* $84 $100 $182 $148 $370 $242 $297 Fluctuates with Economic Conditions % of Revenue 10% 9% 12% 8% 14% 10% 9% Other $74 $101 $120 $146 $165 $229 $270 Estate & Gift Taxes / Duties / Fees / etc. % of Revenue 9% 9% 8% 8% 6% 9% 8% Expense ($B) $1,004 $1,382 $1,601 $2,011 $2,729 $3,537 $3,982 Y/Y Growth 1% 4% 3% 8% 3% -2% 3% Entitlement / Mandatory $421 $648 $810 $1,106 $1,450 $2,030 $2,519 Risen Owing to Rising Healthcare Costs + % of Expense 42% 47% 51% 55% 53% 57% 63% Aging Population Non-Defense Discretionary $162 $231 $275 $385 $494 $616 $610 Education / Law Enforcement / % of Expense 16% 17% 17% 19% 18% 17% 15% Transportation / Government Administration… Defense $283 $303 $272 $349 $548 $671 $590 2007 increase driven by War on Terror % of Expense 28% 22% 17% 17% 20% 19% 15% Net Interest on Public Debt $139 $199 $244 $171 $237 $220 $263 Has Benefitted from Declining Interest % of Expense 14% 14% 15% 9% 9% 6% 7% Rates Since Early 1980s Surplus / Deficit ($B) -$150 -$290 -$22 -$158 -$161 -$1,088 -$666 -19% Average Net Margin, 1987-2017 Net Margin (%) -18% -27% -1% -9% -6% -44% -20% Source: Congressional Budget Office, White House Office of Management and Budget. *Individual & corporate income taxes include capital gains taxes. Note: USA federal fiscal year ends in September. Non-defense 279 discretionary includes federal spending on education, infrastructure, law enforcement, judiciary functions.

…USA Income Statement = Net Loses in 45 of 50 Years USA Annual Profits & Losses $500B s $0 Los / rofit Net P -$500B nnual A USA -$1,000B -$1,500B 1968 1972 1976 1980 1984 1988 1992 1996 2000 2004 2008 2012 2016 Source: Congressional Budget Office, White House Office of Management and Budget. Note: USA federal fiscal year ends in September. 280

Real GDP Growth @ 2.3% (Q1)... 1988-2003 @ 3.0%...2003-2018 @ 2.0% Average Real GDP Growth Y/Y 12% 8% 4% Q1:18 = 2.3% USA 3.0% Average , 0% th 1988-2003 2.0% Average Grow 2003-2018 -4% -8% -12% 1988 1993 1998 2003 2008 2013 2018 Source: Bureau of Economic Analysis (BEA). Note: Real GDP based on chained 2009 dollars. Growth defined as growth over preceding period, seasonally adjusted annual rate. 281

USA Rising Debt Commitments = Non-Trivial Challenge 282

Net Debt / GDP Ratio = Highest Level Since WWII USA Net Debt / GDP Ratio 120% World War II = ~105% 100% DP G 80% 60% Net Debt / World War I = Civil War = USA ~30% 40% ~30% 20% 0% 1790 1815 1840 1865 1890 1915 1940 1965 1990 2015 Historical Net Debt / GDP Source: Congressional Budget Office Long-Term Outlook (3/18). 283

USA Public Debt / GDP Level = th 7 Highest vs. Major Economies Government Debt Government Debt Country % of GDP 2017 ($B) Country % of GDP 2017 ($B) 1) Japan 240% $12,317 11) Egypt 101% $199 2) Greece 180 403 12) Spain 99 1,412 3) Lebanon 152 80 13) France 97 2,730 4) Italy 133 2,798 14) Jordan 96 39 5) Portugal 126 301 15) Bahrain 91 31 6) Singapore 111 362 16) Canada 90 1,482 7) USA 108 20,939 17) UK 89 2,532 8) Jamaica 107 16 18) Mozambique 88 12 9) Cyprus 106 24 19) Ukraine 86 92 10) Belgium 104 561 20) Yemen 83 30 Source: IMF 2017 Estimates Note: Ranking excludes countries with public debt less than $10B in 2015. Public debt includes federal, state and local government debt but excludes unfunded pension liabilities from government 284 defined-benefit pension plans and debt from public enterprises and central banks. FX rates as of 3/28/18.

USA Rising Debt Drivers = Spending on Healthcare Entitlements (Medicare + Medicaid) 285

USA Entitlements = 63% vs. 42% of Government Spending Thirty Years Ago… USA Expenses by Category $1.0T $4.0T 100% 10% 7% 1987 2017 14% Change 15% 80% 8% Debt* ld +$13T (+650%) 16% ie 15% Y Entitlements s ury 60% 6% s +$2.1T (+498%) e a ns re 28% T 63% Non-Defense pe x Y 0 Discretionary E 1 40% 4% +$448B (+277%) US USA Defense +$308B (+109%) 42% 20% 2% Net Interest Cost: +$124B (+89%) 0% 0% 1987 2017 Entitlements / Mandatory Defense 10 Year Treasury Yield Non-Defense Discretionary Net Interest Cost Source: Congressional Budget Office, White House Office of Management and Budget, USA Treasury *Debt reflects net debt (i.e. excludes debt issued by The Treasury and owned by other Government accounts) 286 Note: Yellow line represents yield on 10-year USA Treasury bill from 12/31/86 to 12/31/17.

…USA Entitlements = Medicare + Medicaid Driving Most Spending Growth… USA Entitlements by Category 1987 Entitlements* = 2017 Entitlements* = $349B / 35% of Expenses $2.2T / 56% of Expenses 30% s 24% nt me 20% 20% itle nt E 15% ndatory a 9% M 10% 8% 7% USA 4% 3% 0% Social Security Medicare Medicaid Income Security Medicare Medicaid 1987 2017 Source: Congressional Budget Office, White House Office of Management and Budget. *1987 Income Security programs defined as Food Stamps + SSI + Family Support + Child Nutrition + Earned Income Tax Credit + Other. 2017 Income 287 Security defined as Earned Income Tax Credit + SNAP + SSI + Unemployment + Family Support + Child Nutrition. In 2017, there was an additional ~$200MM in mandatory spending, including Veterans’ pensions & ~$73MM in 1987.

USA Entitlements Growth Over 30 Years = Looking @ Numbers…Closer to Home 2016 $59K = Median USA Household Income $20K = Average Entitlement Payout per Household from Federal Government… Scale = Equivalent to 34% of Household Income 1986 $25K = Median USA Household Income $5K = Average Entitlement Payout per Household from Federal Government… Scale = Equivalent to 19% of Household Income Source: Congressional Budget Office, White House Office of Management and Budget, US Census Bureau 288

IMMIGRATION = IMPORTANT FOR USA TECHNOLOGY JOB CREATION 289

USA = 56% of Most Highly Valued Tech Companies Founded By… st nd 1 or 2 Generation Americans…1.7MM Employees, 2017 Immigrant Founders / Co-Founders of Top 25 USA Valued Public Tech Companies, Ranked by Market Capitalization Mkt Cap LTM Rev Founder / Co-Founder Rank Company ($MM) ($MM) Employees (1st / 2nd Gen Immigrant ) Generation nd 1 Apple $923,554 $239,176 123,000 Steve Jobs 2 –Syria nd 4 Amazon.com 782,608 177,866 566,000 Jeff Bezos 2 –Cuba 3 Microsoft 753,030 95,652 124,000 -- -- st 2 Alphabet / Google 739,122 110,855 80,110 Sergey Brin 1 –Russia st 5 Facebook 537,648 40,653 25,105 Eduardo Saverin 1 –Brazil 6 Intel 257,791 62,761 102,700 --* -- 7 Cisco 202,083 48,096 72,900 -- -- nd Larry Ellison / 2 –Russia / 8 Oracle 188,848 39,472 138,000 nd Bob Miner 2 –Iran 11 Netflix 152,025 11,693 4,850 -- -- st 10 NVIDIA 150,894 9,714 10,299 Jensen Huang 1 –Taiwan nd 9 IBM 129,635 79,139 366,600 Herman Hollerith 2 –Germany 12 Adobe Systems 119,271 7,699 17,973 -- -- 13 Booking.com 100,013 12,681 22,900 -- -- st Cecil Green / 1 –UK / 14 Texas Instruments 108,912 14,961 29,714 nd J. Erik Jonsson 2 –Sweden st Max Levchin / 1 –Ukraine / st Luke Nosek / 1 –Poland / 15 PayPal 95,858 13,094 18,700 st Peter Thiel / 1 –Germany / st Elon Musk*** 1 –South Africa 16 Salesforce.com 94,260 10,480 25,000 -- -- st 17 Qualcomm 86,333 22,360 33,800 Andrew Viterbi 1 –Italy nd 19 Automatic Data Processing 57,237 12,790 58,000 Henry Taub 2 –Poland st 21 VMware 55,282 7,922 20,615 Edouard Bugnion 1 –Switzerland 20 Activision Blizzard 53,772 7,017 9,625 -- -- 18 Applied Materials 52,439 15,463 18,400 -- -- 23 Intuit 50,471 5,434 8,200 -- -- st Francisco D‘Souza / 1 –India** / 22 Cognizant Technology 43,597 14,810 260,000 st Kumar Mahadeva 1 – Sri Lanka st 24 eBay 37,304 9,567 14,100 Pierre Omidyar 1 – France 25 Electronic Arts 34,763 4,845 8,800 -- -- Source: CapIQ as of 4/16/18. “The ‘New American’ Fortune 500” (2011), a report by the Partnership for a New American Economy, as well as “Reason for Reform: Entrepreneurship” (10/16), “American Made, The Impact of Immigrant Founders & Professionals on U.S. Corporations.” *While Andy Grove 290 (from Hungary) is not a co-founder of Intel, he joined as COO on the day it was incorporated. **Francisco D’Souza is a person of Indian origin born in Kenya. ***Max Levchin / Luke Nosek / Peter Thiel’s startup Confinity merged with Elon Musk’s startup X.com to form PayPal in 3/00.

USA = Many Highly Valued Private Tech Companies Founded By… st 1 Generation Immigrants Immigrant Country MarketValue Immigrant Country MarketValue Company Founder / Co-Founder of Origin ($B) Company Founder / Co-Founder of Origin ($B) JetSmarter Sergey Petrossov Russia $2 Uber Garrett Camp Canada $72 Warby Parker Dave Gilboa Sweden 2 SpaceX Elon Musk South Africa 25 Carbon3D Alex Ermoshkin Russia 2 Palantir Peter Thiel Germany 21 Infinidat Moshe Yanai Israel 2 WeWork Adam Neumann Israel 21 Tango Uri Raz, Eric Setton Israel / France 2 John Collison, Stripe Ireland 9 Patrick Collison Louay Eldada, Lebanon / Quanergy 2 Wish Peter Szulczewski, Tianyue Yu China Canada 9 (ContextLogic) Danny Zhang Zoox Tim Kentley-Klay Australia 2 Moderna Noubar Afeyan, Armenia / 8 Eventbrite Renaud Visage France 2 Therapeutics Derrick Rossi Canada Apttus Kirk Krappe UK 2 Baiju Bhatt, India / Robinhood 6 Vlad Tenev Bulgaria Cloudflare Michelle Zatlyn Canada 2 Stewart Butterfield, Proteus Digital Canada / Andrew Thompson UK 2 Slack Serguei Mourachov, 5 Health Russia / UK Cal Henderson Guy Haddleton, New Zealand / Anaplan 1 Tanium David Hindawi Iraq 5 Michael Gould UK Rubrik Bipul Sinha India 1 Credit Karma Kenneth Lin China 4 OfferUp Arean Van Veelen Netherlands 1 Houzz Adi Tatarko, Alon Cohen Israel 4 Actifio Ash Ashutosh India 1 Instacart Apoorva Mehta India 4 Gusto Tomer London Israel 1 Bloom Energy KR Sridhar India 3 Medallia Borge Hald Norway 1 Oscar Health Mario Schlosser Germany 3 Nigel Eccles, Unity David Helgason Iceland 3 FanDuel Tom Griffiths, UK 1 Technologies Lesley Eccles Al Goldstein, Uzbekistan / Avant 2 Daniel Saks, John Sun, Paul Zhang China / China AppDirect Canada 1 Nicolas Desmarais Zenefits Laks Srini India 2 Stepan Pachikov, Azerbaijan / Evernote 1 Phil Libin Russia AppNexus Mike Nolet Holland 2 Udacity Sebastian Thrun Germany 1 ZocDoc Oliver Kharraz Germany 2 Daniel Dines, Marius UiPath* Romania 1 Sprinklr Ragy Thomas India 2 Tirca Compass Ori Allon Israel 2 Zoom Video Eric Yuan China 1 Source for Valuation and Founders Backgrounds: Based on analysis by the Wall Street Journal, CB Insights, Forbes and Business Insider Note: Due to varying definitions of unicorns, may not align with various unicorn lists. As of April 2018 there are 105 US-based, venture-backed 291 unicorns (including rumored valuations). *UiPath is headquartered in New York, NY but was originally founded in Romania.

View Private Tech Companies in Detail

APPENDIX 292

Global Industry Classification System (GICS) (Slides 39 / 41 / 42) GICS is a four-tiered, hierarchical industry classification system. It consists of 11 sectors, 24 industry groups, 68 industries and 157 sub-industries. The GICS methodology is widely accepted as an industry analytical framework for investment research, portfolio management and asset allocation. Companies are classified quantitatively and qualitatively. Each company is assigned a single GICS classification at the sub-industry level according to its principal business activity. MSCI and S&P Global use revenues as a key factor in determining a firm’s principal business activity. Earnings and market, however, are also recognized as important and relevant information for classification purposes. Global industry coverage is comprehensive and precise. The classification system is comprised of over 50,000 trading securities across 125 countries, covering approximately 95% of the world’s equity market capitalization. Company classifications are regularly reviewed and maintained. Specialized teams from two major index providers — MSCI and S&P Global — have defined review procedures, refined over nearly 15 years. Each sector includes the following industries: • Energy= Energy Equipment & Services, Oil, Gas & Consumables Fuels • Materials = Chemicals, Construction Materials, Containers & Packaging, Metals & Mining, Paper & Forest Products • Industrials = Aerospace & Defense, Building Products, Construction & Engineering, Electrical Equipment, Industrial Conglomerates, Machinery, Trading Companies & Distributors, Commercial Services & Suppliers, Professional Services, Air Freight & Logistics, Airlines, Marine, Road & rail, Transportation Infrastructure • Consumer Discretionary = Auto Components, Automobiles, Household Durables, Leisure Products, Textiles, Apparel & Luxury Goods, Hotels, Restaurants & Leisure, Diversified Consumer Services, Media, Distributors, Internet & Direct Marketing Retail, Multiline Retail, Specialty Retail • Consumer Staples =Food & Staples Retailing, Beverages, Food Products, Tobacco, Household Products, Personal Products • Healthcare = Healthcare Equipment & Supplies, Healthcare Providers & Services, Healthcare Technology, Biotechnology, Pharmaceuticals, Life Sciences Tools & Services • Financials = Commercial Banks, Thrifts & Mortgage Finance, Diversified Financial Services, Consumer Finance, Capital Markets, Mortgage Real Estate Investment Trusts (REITs), Insurance • Information Technology = Internet Software & Services, IT Services, Software, Communications Equipment, Computers & Peripherals, Electronic Equipment & Instruments, Semiconductors & Semiconductors Equipment • Telecommunication Services = Diversified Telecommunication Services, Wireless Telecommunication Services • Utilities = Electric Utilities, Gas Utilities, Multi-Utilities, Water Utilities, Independent Power & Renewable Electricity Producers • Real Estate = Equity Real Estate Investment Trusts (REITs), Real Estate Management & Development Source: MSCI, S&P 500 293

Disclaimer This presentation has been compiled for informational purposes only and should not be construed as a solicitation or an offer to buy or sell securities in any entity, or to invest in any Kleiner Perkins (KP) entity or affiliated fund. The presentation relies on data + insights from a wide range of sources, including public +private companies, market research firms + government agencies. We cite specific sources where data are public; the presentation is also informed by non-public information + insights. We disclaim any + all warranties, express or implied, with respect to the presentation. No presentation content should be construed as professional advice of any kind (including legal or investment advice). We publish the Internet Trends report on an annual basis, but on occasion will highlight new insights. We may post updates, revisions, or clarifications on the KP website. KP is a venture capital firm that owns significant equity positions in certain of the companies referenced in this presentation, including those at http://www.kleinerperkins.com/companies. Any trademarks or service marks used in this report are the marks of their respective owners, who are not participating partners or sponsors of the presentation or of KP or its affiliated funds + such owners do not endorse the presentation or any statements made herein. All rights in such marks are reserved by their respective owners. 294