Internet Trends 2017 Report

Internet Trends 2017 Report

Internet Trends 2017 Report from Kleiner Perkins

Internet Trends 2017 Report

Mary Meeker May 31, 2017 kpcb.com/InternetTrends INTERNET TRENDS 2017 – CODE CONFERENCE

KP INTERNET TRENDS 2017 | PAGE 2KP INTERNET TRENDS 2017 | PAGE 2 Internet Trends 2017 1) Global Internet Trends = Solid…Slowing Smartphone Growth 4-9 2) Online Advertising (+ Commerce) = Increasingly Measurable + Actionable 10-79 3) Interactive Games = Motherlode of Tech Product Innovation + Modern Learning 80-150 4) Media = Distribution Disruption @ Torrid Pace 151-177 5) The Cloud = Accelerating Change Across Enterprises 178-192 6) China Internet = Golden Age of Entertainment + Transportation 193-231 (Provided by Hillhouse Capital) 7) India Internet = Competition Continues to Intensify…Consumers Winning 232-287 8) Healthcare @ Digital Inflection Point 288-319 9) Global Public / Private Internet Companies… 320-333 10) Some Macro Thoughts… 334-351 11) Closing Thoughts… 352-353

KP INTERNET TRENDS 2017 | PAGE 3 Thanks... Kleiner Perkins Partners Alexander Krey & Ansel Parikh - who were fearless and sometimes sleepless - helped steer the ideas / presentation we hope you find useful / learn from / improve on. Key contributors to specific content include: Noah Knauf & Nina Lu (Healthcare), Bing Gordon (Interactive Games), Alex Tran & Anjney Midha (India), Daegwon Chae (Ads + Commerce) and Alex Kurland & Lucas Swisher (Enterprise). In addition, Mood Rowghani, Eric Feng, Daniel Axelsen, Dino Becirovic and Shabih Rizvi were more than on call with help. Hillhouse Capital Especially Liang Wu…his / their contribution of the China sector of Internet Trends provides an especially thoughtful overview of the largest market of Internet users in the world… Participants in Evolution of Internet Connectivity From creators to consumers who keep us on our toes 24x7...and the people who directly help us prepare this presentation... Kara & Walt For continuing to do what you do so well...

KP INTERNET TRENDS 2017 | PAGE 4 GLOBAL INTERNET TRENDS = SOLID USER GROWTH… SLOWING SMARTPHONE GROWTH

KP INTERNET TRENDS 2017 | PAGE 5 Global Internet Trends = Solid User Growth…Slowing Smartphone Growth 1) Global Internet Users = 3.4B…Flat Growth +10% vs. 10% Y/Y… +8% vs. 8% Y/Y (ex. India) 2) Global Smartphone Shipments = Slowing +3% vs. +10% Y/Y 3) Global Smartphone Installed Base = Slowing +12% vs. +25% Y/Y 4) USA Internet Usage (Engagement) = Solid +4% Y/Y

KP INTERNET TRENDS 2017 | PAGE 6 Source: United Nations / International Telecommunications Union, US Census Bureau. Internet user data is as of mid-year. Internet user data for: USA from Pew Research, China from CNNIC, Iran from Islamic Republic News Agency / InternetWorldStats / KPCB estimates, India from KPCB estimates based on IAMAI data, Indonesia from APJII. Global Internet Users = 3.4B @ 46% Penetration... +10% Y/Y vs. +10%...+8% Y/Y vs. +8% (Ex-India) 0% 5% 10% 15% 20% 25% 30% 35% 0 500 1,000 1,500 2,000 2,500 3,000 3,500 4,000 2009 2010 2011 2012 2013 2014 2015 2016 Y/Y%Growth GlobalInternetUsers(MM) Global Internet Users (MM) Y/Y Growth (%) Global Internet Users (MM), 2009 – 2016

KP INTERNET TRENDS 2017 | PAGE 7 Global Smartphone Unit Shipments = Continue to Slow... @ +3% Y/Y vs. +10% (2015) / +28% (2014) Source: Morgan Stanley Research (5/17) Smartphone Unit Shipments by Operating System (MM), Global, 2009 – 2016 0% 20% 40% 60% 80% 100% 0 300 600 900 1,200 1,500 2009 2010 2011 2012 2013 2014 2015 2016 Y/YGrowth(%) GlobalSmartphoneUnitShipments(MM) Android iOS Other Y/Y Growth

KP INTERNET TRENDS 2017 | PAGE 8 Source: Morgan Stanley Research (5/17) Note: Owing to use of different source, prior period data may have slight adjustments vs prior reports. Smartphone installed base based on preceding 8 quarters of smartphone shipments. Global Smartphone Installed Base = 2.8B… +12% Y/Y vs. +25% (2015) / +37% (2014) 0% 10% 20% 30% 40% 50% 60% 70% 0 500 1,000 1,500 2,000 2,500 3,000 3,500 2009 2010 2011 2012 2013 2014 2015 2016 Y/Y%Growth GlobalSmartphoneInstalledBase(MM) Global Smartphone Installed Base (MM) Y/Y Growth (%) Global Smartphone Installed Base (MM), 2009 – 2016

KP INTERNET TRENDS 2017 | PAGE 9 Internet Usage (Engagement) = Solid Growth…+4% Y/Y… Mobile >3 Hours / Day per User vs. <1 Five Years Ago, USA Source: eMarketer 9/14 (2008-2010), eMarketer 4/15 (2011-2013), eMarketer 4/17 (2014-2016). Note: Other connected devices include OTT and game consoles. Mobile includes smartphone and tablet. Usage includes both home and work. Ages 18+; time spent with each medium includes all time spent with that medium, regardless of multitasking. Time Spent per Adult User per Day with Digital Media, USA, 2008 – 2016 0.2 0.3 0.4 0.3 0.3 0.3 0.3 0.4 0.4 2.2 2.3 2.4 2.6 2.5 2.3 2.2 2.2 2.2 0.3 0.3 0.4 0.8 1.6 2.3 2.6 2.8 3.1 2.7 3.0 3.2 3.7 4.3 4.9 5.1 5.4 5.6 0 1 2 3 4 5 6 2008 2009 2010 2011 2012 2013 2014 2015 2016 HoursperDay Other Connected Devices Desktop / Laptop Mobile

KP INTERNET TRENDS 2017 | PAGE 10 ONLINE ADVERTISING (+ COMMERCE) = INCREASINGLY MEASURABLE + ACTIONABLE

KP INTERNET TRENDS 2017 | PAGE 11 Ad Growth = Driven by Mobile

KP INTERNET TRENDS 2017 | PAGE 12 Online Advertising = Growth Accelerating, +22% vs. +20% Y/Y... Mobile $ > Desktop (2016) on Higher Growth, USA Source: IAB / PWC Internet Advertising Report (2016) USA Internet Advertising ($B), 2009 – 2016 $23 $26 $32 $37 $43 $50 $60 $73 0% 5% 10% 15% 20% 25% 30% 35% 40% $0 $10 $20 $30 $40 $50 $60 $70 $80 2009 2010 2011 2012 2013 2014 2015 2016 %Y/YGrowth USAInternetAdvertising($B) Desktop Advertising Mobile Advertising Y/Y Growth

KP INTERNET TRENDS 2017 | PAGE 13 Advertising $ = Shift to Usage (Mobile) Continues % of Time Spent in Media vs. % of Advertising Spending, USA, 2016 4% 9% 38% 20% 28% 12% 9% 38% 20% 21% 0% 10% 20% 30% 40% 50% Print Radio TV Internet Mobile %ofTotalMediaConsumptionTime orAdvertisingSpending Time Spent Ad Spend Total Internet Ad = $73B Of Which Mobile Ad = $37B ~$16B Opportunity in USA Source: Internet and Mobile advertising spend based on IAB and PwC data for full year 2016. Print, Radio, and TV advertising spend based on Magna Global estimates for full year 2016. Print includes newspaper and magazine. Internet (IAB) includes desktop + laptop + other connected devices. ~$16B opportunity calculated assuming Mobile (IAB) ad spend share equal its respective time spent share. Time spent share data based on eMarketer (4/17). Arrows denote Y/Y shift in percent share. Excludes out-of-home, video game, and cinema advertising.

KP INTERNET TRENDS 2017 | PAGE 14 Advertising $ = Internet > TV Within 6 Months, Global Internet vs. TV Ad Spend ($B), Global, 1995-2017E $0 $50 $100 $150 $200 $250 GlobalAdvertisingSpend($B) Global Internet Ad Spend Global TV Ad Spend Source: Zenith Advertising Expenditure Forecasts (3/17)

KP INTERNET TRENDS 2017 | PAGE 15 $ $5 $10 $15 $20 $25 $30 $35 $40 $45 $50 USAAdvertisingRevenue($B) $0 $5,000 10,000 15,000 20,000 25,000 30,000 35,000 40,000 45,000 50,000 Google + Facebook = 85% (& Rising) Share of Internet Advertising Growth, USA Advertising Revenue ($B) and Growth Rates (%) of Google vs. Facebook vs. Other, USA, 2015 – 2016 2015 2016 2015 2016 2015 2016 Google Facebook Others +20% Y/Y +62% Y/Y +9% Y/Y Source: IAB / PWC Advertising Report (2016), Facebook, Morgan Stanley Research Note: Facebook revenue includes Canada. Google USA ad revenue per Morgan Stanley estimates as company only discloses total ad revenue and total USA revenue. “Others” includes all other USA internet (mobile + desktop) advertising revenue ex-Google / Facebook.

KP INTERNET TRENDS 2017 | PAGE 16 Ad Measurability = Can Be Triple-Edged… When Things Are Measured = People Don’t Always Like What They See… Users Don’t Always Like Data Collected

KP INTERNET TRENDS 2017 | PAGE 17 56% 21% 15% 0% 10% 20% 30% 40% 50% 60% 70% Engagement Conversion & Revenue Amplification & Brand Awareness %ofRespondentsAdvertisers = Like Measurable Engagement Metrics But… Some Find Measuring ROI Challenging (as with Offline) Social Advertisers Metrics Used to Measure Success, 6/16 Social Media Marketing Top Challenges, 6/16 61% 38% 34% 0% 10% 20% 30% 40% 50% 60% 70% Measuring ROI Securing Budget & Resources Tying Social Campaigns to Business Goals %ofRespondents Source: SimplyMeasured State of Social Marketing Annual Report (6/16) Note: Based on a survey of social media advertisers, n=350.

KP INTERNET TRENDS 2017 | PAGE 18 Ad Blocking = Growth Continues…Especially in Developing Markets… Users Increasingly Opt Out of Stuff They Don’t Want 0 100 200 300 400 2009 2010 2011 2012 2013 2014 2015 2016 GlobalAdblockingUsers(MM) Desktop Adblocking Software Users Mobile Adblocking Browser Users Adblocking Users on Web (Mobile + Desktop), Global, 4/09 – 12/16 Adblocking Penetration (Mobile + Desktop), Selected Countries, 12/16 Country Desktop Mobile China 1% 13% India 1% 28% USA 18% 1% Brazil 6% 1% Japan 3% -- Russia 6% 3% Germany 28% 1% Indonesia 8% 58% UK 16% 1% France 11% 1% Canada 24% -- Source: PageFair 2015, 2017 reports. These two data sets have not been de-duplicated. The number of desktop adblockers after 1/16 are estimates based on the observed trend in desktop adblocking and provided by PageFair. Note that mobile adblocking refers to web / browser-based adblocking and not in-app adblocking. Desktop adblocking estimates are for global monthly active users of desktop adblocking software between 4/09 – 12/16, as calculated in the PageFair’s 2015 and 2017 reports. Mobile adblocking estimates are for global monthly active users of mobile browsers that block ads by default between 9/14 – 12/16, including the number of Digicel subscribers in the Caribbean (added 10/15), as calculated in the PageFair & Priori Data 2016 and PageFair 2017 Adblocking Report.

KP INTERNET TRENDS 2017 | PAGE 19 Leading Platform Ad Offerings = Rapidly Improving with Back-End Data + Front-End Measurement Tools + Targeted Delivery of Ads Users Increasingly Want

KP INTERNET TRENDS 2017 | PAGE 20 Source: Facebook, Google, Snap Leading Online Ad Platforms = Providing More Ways to Target + Measure Ads Facebook (Delivery Insights) Google (AdWords) Snap (Snap Ads)

KP INTERNET TRENDS 2017 | PAGE 21 Product Listing Ads (Google) = Driving Clicks to Product Pages Google Product Listing Ads (PLAs) Share of Retail Paid Clicks on Google, USA, 2014-2016 29% 52% 0% 10% 20% 30% 40% 50% 60% PLAShareofRetailPaidClicksonGoogle(%) Google PLA on Mobile Web, 12/16 Source: Merkle Digital Marketing Report (Q1:14-Q1:17), Right image: Search Engine Land

KP INTERNET TRENDS 2017 | PAGE 22 Targeted Pins (Pinterest) = Driving Product Discovery + Purchase Pinterest Browsing Turning into Buying, 4/17 33% 12% 44% 24% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 50% Browse Buy %ofRespondents 4/15 4/17 Which of these services is a great place to browse for things you might want to buy? Which of these services is a great place to buy things online? Shop the Look Inspired Purchases, 2/17 Source: Pinterest Note: Based on an internal survey of global internet users, n=12K. Other answers to the questions include Facebook, Instagram, Twitter, Snap, YouTube, and Google with each respondent only allowed to choose one option.

KP INTERNET TRENDS 2017 | PAGE 23 Contextual Ads (Facebook) = Driving Direct Purchases Facebook Messenger Conversational Transactions, 9/16 26% 7% 10% 74% 93% 90% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Clicked on an Ad Didn’t Click on Ad Not Sure %ofRespondents Made a Purchase Did Not Make a Purchase Facebook Users 26% that Click Ads Make Purchase, USA, 3/17 In past 30 days, have you clicked an ad on Facebook? In past 30 days, have you purchased a product you saw on Facebook? Source: Survata (4/17), Messenger Image: Facebook Blog (9/16) Note: Based on survey of USA internet users, n=1,500 (3/17).

KP INTERNET TRENDS 2017 | PAGE 24 Goal Based Bidding Ads (Snap) = Driving User Action Snap / Gatorade Ad Campaign Users Swipe Through Ad to Web Game, 8/16 Users Spend Average of 196 Seconds Playing Game Source: Snap Case Study: Gatorade (8/16)

KP INTERNET TRENDS 2017 | PAGE 25 Geo-Targeted Local Ads (Google) = Driving Foot Traffic to Stores Google Location-Tagged Ads 99% Accuracy Tracking Visits to 200MM Stores Globally, 9/16 5B Cumulative Tracked Store Visits, Up 5x Y/Y*, 5/17 Source: Google Adwords Blog (5/16, 9/16, 5/17), Image: Google Adwords Blog (9/16) * 5B (5/17) vs. 1B cumulative tracked (5/16).

KP INTERNET TRENDS 2017 | PAGE 26 Incentive-Based + Skippable Video Ads = Driving Positive Interactions Incentive-Based + Skippable Video Ads More Likely to be Viewed Positively, 5/16 How would you characterize your attitude towards the following formats of online video advertising? 19% 20% 21% 26% 45% 46% 51% 52% 68% 81% 80% 79% 74% 55% 54% 49% 48% 32% Mobile App Pop-Up Pre-Roll In-Banner Auto-Play Social Auto-Play In-Banner Click-to-Play Skippable Mobile Pop-up Skippable Pre-Roll Social Click-to-Play Mobile App Reward % of RespondentsPositive Negative Source: MillwardBrown AdReaction Video Creative in a Digital World (5/16) Note: Survey of people from Argentina, Australia, Brazil, France, Germany, Mexico, UK, and USA who watched 20 ads (at least 100 per ad) and answered positive or negative, n=10.739. The survey included TV, YouTube skippable pre-roll, Facebook auto-play, Facebook click-to-play, and mobile video ad formats.

KP INTERNET TRENDS 2017 | PAGE 27 In-App Ads + Dynamic Creative (Vungle) = Driving Higher In-App Install Performance Dynamic Tab Ad Video + Images 0.00% 0.10% 0.20% 0.30% 0.40% 0.50% 0.60% 0 100 200 300 400 500 600 1/16 2/16 3/16 4/16 5/16 6/16 7/16 8/16 9/16 10/16 11/16 12/16 1/17 2/17 3/17 4/17 ConversionRate(%) DynamicCreativeVariations(K) Dynamic Creative Variations Conversion Rate Vungle Dynamic Creative Ads Improving Conversion Rates, 5/17 Source: Vungle (5/17) Note: “Dynamic creative” is any creative ad that changes automatically based on information about the user (behavior, location, or context). A dynamic tab ad includes multiple interactive promotional modules alongside a video ad.

KP INTERNET TRENDS 2017 | PAGE 28 In-Ride / In-Hand Recommendations (Uber + Foursquare) = Location + Route + Destination + Time of Day (+ an Offer) Uber / Foursquare Partnership In-App Recommendations for Nearby Businesses, 4/17 Source: Uber (4/17)

KP INTERNET TRENDS 2017 | PAGE 29 Hyperlocal Targeting (Nextdoor…xAd) = From Home (Neighborhood) to Work (Commute) xAd Tracking Where / When Purchases Likely to be Made Nextdoor Neighbors Drive Word of Mouth +8% Engagement Lift for Ring Source: Nextdoor, xAd

KP INTERNET TRENDS 2017 | PAGE 30 Advertising Inefficiency = Increasingly Exposed by Data… Right ‘Ad’ @ Right Place / Time

KP INTERNET TRENDS 2017 | PAGE 31 Right Ad @ Right Place / Time (Driven by Algorithms)… User-Typed Input (Words)… Linked to Relevant Ad = Google AdWords (Launched 2000) Source: Historyofinformation.com, Google

KP INTERNET TRENDS 2017 | PAGE 32 …Right Ad @ Right Place / Time… Based on User-Typed Input (Words) = Big Business for Google Google = $679B Market Capitalization +30x vs. IPO $0 $200 $400 $600 $800 $1,000 $1,200 8/04 2/05 8/05 2/06 8/06 2/07 8/07 2/08 8/08 2/09 8/09 2/10 8/10 2/11 8/11 2/12 8/12 2/13 8/13 2/14 8/14 2/15 8/15 2/16 8/16 2/17 SharePrice($) Source: Yahoo Finance Note: Priced as of 5/26/17 market close. Google IPO’ed @ $85 / share on 8/19/04.

KP INTERNET TRENDS 2017 | PAGE 33 Right Ad @ Right Place / Time (Driven by Algorithms)… User-Uploaded Input (Real-Time Images)… Linked to Relevant Ad = SnapAds (Launched 2014) Source: Image: Adweek (10/14)

KP INTERNET TRENDS 2017 | PAGE 34 …Right Ad @ Right Place / Time… Based on User-Uploaded Input (Images) = Big Business for Snap Snap = $25B Market Capitalization 0 20 40 60 80 100 120 140 160 180 $0 $20 $40 $60 $80 $100 $120 $140 $160 $180 Q1:15 Q2:15 Q3:15 Q4:15 Q1:16 Q2:16 Q3:16 Q4:16 Q1:17 DAU(MM) NetRevenue($MM) Net Revenue DAU Source: Snap Filings Note: Priced as of 5/26/17 market close. Snap IPO’ed @ $17 / share on 3/2/17.

KP INTERNET TRENDS 2017 | PAGE 35 A lot of the future of search is going to be about pictures instead of keywords. - Ben Silbermann, Pinterest Founder / CEO, 4/17 Source: CNBC interview (4/3/17)

KP INTERNET TRENDS 2017 | PAGE 36 Ads Evolving Rapidly = Often Organic + Data @ Core

KP INTERNET TRENDS 2017 | PAGE 37 Emerging Retailers + Crafty Big Brands = Finding Ways to Make Collaborative Ad Creation (Social + UGC) Work for Them…

KP INTERNET TRENDS 2017 | PAGE 38 Brands + Consumers = Re-Distribution Driving Engagement… Ben & Jerry’s / UGC on Instagram, 5/17 Effective UGC can generate 6.9x higher engagement than brand generated content on Facebook, per Mavrck, 2/17 Source: Mavrck Facebook UGC Benchmark Report (2/17), Image: benandjerrys Instagram featuring mistress_spice (4/17) Note: Study based on 536,238 micro-influencer brand activations completed via Mavrck Platform from 1/1/16-12/13/16.

KP INTERNET TRENDS 2017 | PAGE 39 …Brands + Consumers = Brands Sourcing Content from Fans… Brands = Leveraging UGC on Instagram 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% Amazon Video Emirates Turkish Airlines Cathay Pacific Starbucks Netflix Sephora Wayfair BMW Red Bull Qatar Airways % of Instagram Content Regrammed Source: SimplyMeasured (11/16) Note: Data collected from each company’s Instagram page from 7/16-10/16. Posts were manually tagged for regrams based on mentions on ‘regram’ in the post or the camera emojis.

KP INTERNET TRENDS 2017 | PAGE 40 …Brands + Influencers = Re-Distribution Driving Engagement Influencers = Can Impact Followers Source: Stance

KP INTERNET TRENDS 2017 | PAGE 41 …Emerging Retailers + Crafty Big Brands = Finding Ways to Make Images (+ Video) + Data + Algorithms + Voice Work for Them

KP INTERNET TRENDS 2017 | PAGE 42 Image-Based Platform Front-Ends = Tap + Augment Can Replace Typing… ‘Front-End’ User-Generated Real-Time Geolocated Images Source: Left Image: Snap, Right Image: Instagram blog (3/17)

KP INTERNET TRENDS 2017 | PAGE 43 …Image-Based Platform Front-Ends = Taking Pictures Can Replace Typing… ‘Front-End’ Google Lens Will Provide Greater Context to Images Source: Google I/O (5/17)

KP INTERNET TRENDS 2017 | PAGE 44 …Image-Based Platform Back-Ends = Algorithms Infer User Context from Images… ‘Back-End’ Algorithms Infer Images / Project AR Objects into Scenes Source: Images: CB Insights, Seene Patents (acquired by Snap in 6/16) and Looksery Patents (acquired by Snap in 9/15)

KP INTERNET TRENDS 2017 | PAGE 45 …Image Recognition Back-Ends = Can Provide Contextual Relevance for Advertisers Snap Image Recognition Potential Ad Targeting Tool Google Visual Positioning Service Tracking Path to Purchase…In-Store Source: Left Image : Snap Patent (7/16), Right Image: Google I/O (5/17)

KP INTERNET TRENDS 2017 | PAGE 46 Voice-Based Mobile Platform Front-Ends = Voice Can Replace Typing… Google Assistant Nearly 70% of Requests are Natural / Conversational Language, 5/17 20% of Mobile Queries Made via Voice, 5/16 Source: Google I/O (5/16), Image: Macrumors (2/17)

KP INTERNET TRENDS 2017 | PAGE 47 …Voice-Based In-Home Platform Front-Ends = Voice Can Replace Typing… 0 2,000 4,000 6,000 8,000 10,000 12,000 14,000 NumberofSkills Amazon Echo Device Installed Base, USA Amazon Echo Skills Broadening Use Cases Amazon Echo Evolution, 11/14 – 5/17 Echo = Shopping + Media Echo Look = Shopping + Recommendations Echo Show = Video + Voice Calls 0 2 4 6 8 10 12 EchoInstalledBase(MM) Source: Image: Amazon, Consumer Intelligence Research Partners LLC, Geekwire, Technology Review, Wired, Fast Company

KP INTERNET TRENDS 2017 | PAGE 48 …Voice-Based Platform Back-Ends = Voice Recognition Accuracy Continues to Improve Google Machine Learning Achieving Higher Word Accuracy, 2013-2017 95% 95% 70% 80% 90% 100% 2013 2014 2015 2016 2017 WordAccuracyRate(%) Google Threshold for Human Accuracy Source: Google (5/17) Note: Data as of 5/17/17 and refers to recognition accuracy for English language. Word error rate is evaluated using real world search data which is extremely diverse and more error prone than typical human dialogue.

KP INTERNET TRENDS 2017 | PAGE 49 Ads = Becoming Targeted Storefronts

KP INTERNET TRENDS 2017 | PAGE 50 Ads / Content / Products / Transactions = Lines Blurring. Fast… The Content = The Store Emails Curated Storefronts Facebook Feed Browsable Storefronts Source: Left Image: Facebook, Right Image: Stitch Fix

KP INTERNET TRENDS 2017 | PAGE 51 …Ads / Content / Products / Transactions = Lines Blurring. Fast. The Ad = The Transaction Instagram Feed Tap to Book, 4/17 Snap eCommerce Ad Swipe Up to Buy, 5/16 Source: Left Image: Instagram, Right Image: Snap

KP INTERNET TRENDS 2017 | PAGE 52 Product Quality + Customer Support + Transparency Bars Rising = Owing to Social Media

KP INTERNET TRENDS 2017 | PAGE 53 60% 53% 29% 21% 0% 10% 20% 30% 40% 50% 60% 70% Easier Access to Online Support Channels Faster Agent Reponse Times Consistent Customer Experience Across Channels Faster Access to Live Support %ofResponses Social Media = Can Provide Opportunity to Improve Customer Service… If you could choose two things for organizations to improve in customer service, what would they be? (Select two), 8/16 Source: Ovum Get It Right: Deliver the Omni-Channel Support Customers Want (8/16) Note: Survey of consumers ages 18-80 in Australia, Europe, New Zealand, and USA, n=400.

KP INTERNET TRENDS 2017 | PAGE 54 …Social Media = Can Drive Accountability… 82% of Customers Stopped Doing Business with a Company After Bad Experience vs. 76% in 2014, 8/16 Source: Image: Allbirds, Ovum Get It Right: Deliver the Omni-Channel Support Customers Want (8/16) Note: Survey based on consumers ages 18-80 in Australia, Europe, New Zealand, and USA; n=400.

KP INTERNET TRENDS 2017 | PAGE 55 Intercom Simple + Engaging UI …Real-Time Online Customer Conversations = Rising Rapidly… Intercom Conversations Started, Global, 12/13-12/16 0 100 200 300 400 ConversationsStarted(MM) Source: Intercom Note: Conversations include messages initiated by businesses & consumers.

KP INTERNET TRENDS 2017 | PAGE 56 Customers = Increasingly Expect to Understand How Things Work SoFi ‘How It Works’ Most Viewed Content, After Home Page SoFi Member Dashboard Send Questions Directly to CEO Source: SoFi

KP INTERNET TRENDS 2017 | PAGE 57 Retailers Emerging With Especially Effective Strategies…

KP INTERNET TRENDS 2017 | PAGE 58 Chewy.com = Pet Treats / Food / Supplies… Strong User Community + Great Target Market Engaged Community + High Customer Satisfaction Strong Revenue Growth $0 $100 $200 $300 $400 $500 $600 $700 $800 $900 $1,000 2015 2016 Revenue($MM) Dynamic Customer Service Source: Chewy.com

KP INTERNET TRENDS 2017 | PAGE 59 Glossier = Skincare & Beauty Products... Content Marketing User Generated Content = Influencers 300% 350% 400% 450% 500% 550% 600% 2015 2016 ActiveCustomer%Y/YGrowth Accelerating Active Customer Growth ‘Into the Gloss’ = Content Marketing Source: Glossier, Top Left Image: Instagram user genius_hotel, Bottom Left Image: Glossier

KP INTERNET TRENDS 2017 | PAGE 60 UNTUCKit = Shirts… Online-Offline Synergies in Marketing + Merchandising Source: UNTUCKit Note: Online session defined as website visit. 0 1 2 3 4 5 6 7 8 2015 2016 Sessions(MM) Online Sessions Up >2.5x Y/Y Offline Engagement Direct Touchpoints in Physical World Online Storefront Digital Merchandising Insights In-Store Interactions Intimacy + Active Dialogue Digital-Physical Feedback Loop Deliberate Branding + Clear Messaging @ Core

KP INTERNET TRENDS 2017 | PAGE 61 Allbirds = Shoes… Innovative Product + Simple Choice (Less Selection = More) Source: Allbirds * 3/16-6/16, H1:17 data from 1/17-5/17. 0 1 2 3 4 5 6 H1:16* H2:16 H1:17 eCommerceSessions(MM) Growing eCommerce SessionsTwo Comfortable, High Quality Styles Tongue base – double row stitch implemented Tongue lace loop reworked Tongue reinforcement layer added New internal toe reinforcement added 3x adjustments to U-throat opening Insole geometry modified New process/ material logo tabs developed Outsole durometer reduced Outsole redesigned Alternative insole cover material developed New vamp lining wool textile introduced Product Changes Based on Customer Input

KP INTERNET TRENDS 2017 | PAGE 62 Trendyol = Apparel... Private Label + Local Sourcing for Local Consumers (Middle East) High Purchase Re-Engagement Items Purchased per Shopper Continue to Rise 0 2 4 6 8 10 12 2014 2015 2016 2017 YTD AverageUnitsperActiveShopper Private Label + Local Sourcing Low Prices + Short Lead Times ~1K Suppliers 50km from Trendyol HQ Fast Replenishment (7-10 days) Private Label @ 38% of Revenue Other Fashion Brands Source: Trendyol Note: Average units per active shopper calculated over the course of shopper lifetime.

KP INTERNET TRENDS 2017 | PAGE 63 MM.LaFleur = Women’s Professional Wardrobe… Relationship-Driven Experience (Online & Offline) High Growth + Retention $0 $5 $10 $15 $20 $25 $30 $35 2015 2016 Revenue($MM) Returning Customers New Customers Wardrobe Survey Algorithmic Optimization Bento Box Curated Impressions Online Shopping Ongoing Customer Engagement In-Store Stylist Appointments Human Touch + Active Dialogue Source: MM.LaFleur

KP INTERNET TRENDS 2017 | PAGE 64 eCommerce A-Ha’s

KP INTERNET TRENDS 2017 | PAGE 65 0% 2% 4% 6% 8% 10% 12% 0 2 4 6 8 10 12 2010 2011 2012 2013 2014 2015 2016 Y/YGrowth(%) ParcelVolume*(B) If It Seems Like Package / Parcel Growth is Accelerating… It’s Because It Is, +9% Y/Y… Parcel Volume*, USA, 2010-2016 Source: USPS, Fedex, UPS Filings *Combines USPS's Domestic Shipping and Package Services volumes, Fedex's calendar year Domestic Package volumes, and UPS's Domestic Package volumes.

KP INTERNET TRENDS 2017 | PAGE 66 …Apartment Building Lobbies Becoming Warehouses… Doormen Becoming Foremen… Landlords Expanding Package Rooms to Accommodate Rising Online Order Delivery Source: Image: NYTimes Photographer Tony Cenicola

KP INTERNET TRENDS 2017 | PAGE 67 …Unwrapping Boxes… Becoming Entertainment… Unboxing YouTube Top 5 Channels = 33MM+ Subscribers, 5/17 Source: YouTube: Ryan’s Toy Review, Fun Toys Collector Disney Toys Review, Disney Car Toys, Toys AndMe, Blu Toys Club Surprise, Images: CKN Toys

KP INTERNET TRENDS 2017 | PAGE 68 …Eating Out is… Increasingly Eating In… Top 10 DoorDash San Francisco Bay Area Restaurants Delivery as % of Revenue = 7% vs. 2% (2015) Revenue Growth = +45% Y/Y vs. 10% (2015) Eating Out Eating In Source: DoorDash, Left image: Pexels, Right image: DoorDash

KP INTERNET TRENDS 2017 | PAGE 69 …Grocery Shopping… Getting Personal / Fast / Easy… Instacart = Personalized Grocery Recommendations 8x More Likely to Buy When Prompted with ‘Buy It Again’ Option 85% of In-Store Replacements... Chosen Based on Algorithmic Recommendations Source: Instacart

KP INTERNET TRENDS 2017 | PAGE 70 …Lowe’s Doing Augmented Reality… Helping Consumers Find Products In-Store… Lowe’s / Google Partnership Guides Customers to In-Store Items via Augmented Reality on Mobiles, 3/17 Source: Google, Lowe’s

KP INTERNET TRENDS 2017 | PAGE 71 …Stitch Fix Launching Another Private-Label Clothing Brand &… It’s Computer-Generated (1% of Products for Now)... Product Attributes + Customer Feedback + Data Science / Testing New Style, 5/17 Silhouette Sleeve Lace Feature Hem Type Print Cassie Crochet Detail Top Source: Stitch Fix, Left Image: Stitch Fix Algorithms Tour, Right Image: Stitch Fix

KP INTERNET TRENDS 2017 | PAGE 72 …Retail Store Closings May Break 20 Year Record While... Amazon Opens Retail Stores… Retail Unit Closings, USA, 1995-2017 YTD 0 1,000 2,000 3,000 4,000 5,000 6,000 7,000 8,000 1995 1996 1997 1998 1999 2000 2001 2002 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017E UnitClosings Average Amazon Looks to Expand its Physical Footprint Source: Credit Suisse, Amazon Note: 2017 is YTD as of 4/6/17. 2017 estimate per Credit Suisse.

KP INTERNET TRENDS 2017 | PAGE 73 …Digitally Native Brands = Go Offline… I don't think retail is dead. Mediocre retail experiences are dead. - Neil Blumenthal, Co-CEO @ Warby Parker, 1/17 Bonobos Guide Shops Try On In-Store…Ship to Home Warby Parker Schedule Eye Exams…Buy Glasses Source: WSJ interview 1/23/17, Left image: Pinterest, Right image: Fashion Trends Daily

KP INTERNET TRENDS 2017 | PAGE 74 …World’s Largest Offline Retailer (Wal-Mart)… Getting Aggressive Online… Wal-Mart eCommerce Revenue Y/Y Growth, Global 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% %Y/YGrowth 90% of Americans Live Within 10 Miles of a Wal-Mart FQ1:18 eCommerce Revenue Growth @ 63% Y/Y vs. 29% FQ4:17, USA Modcloth.com, 3/17 Moosejaw, 2/17 JD.com (Increased to 12%), 2/17 Shoebuy, 1/17 Recent Acquisitions & Investments Organic + Inorganic Growth Source: Wal-Mart Note: Fiscal year ends January. Wal-Mart stopped disclosing global eCommerce revenue growth after FQ4:17 and began disclosing USA eCommerce revenue growth. Acquired Jet.com, 8/16

KP INTERNET TRENDS 2017 | PAGE 75 …Amazon Becoming a Leading Private-Label Supplier of… Baby Wipes + Batteries, USA… Amazon Basics Market Share, 8/16 USA 0% 5% 10% 15% 20% 25% 30% 35% Online BabyWipesMarketShare(%) 0% 5% 10% 15% 20% 25% 30% 35% OnlineBatteryMarketShare(%) Source: Images: Amazon, 1010 Data Note: Data collected from 9/15-8/16

KP INTERNET TRENDS 2017 | PAGE 76 …eCommerce Growth = +15% Y/Y… Accelerating, Again, USA… Online Retail Sales vs. Y/Y Growth, USA 2010-2016 0% 2% 4% 6% 8% 10% 12% 14% 16% 18% 20% $0 $50 $100 $150 $200 $250 $300 $350 $400 $450 2010 2011 2012 2013 2014 2015 2016 %Y/YGrowth USAOnlineRetailSales($B) Source: St. Louis Federal Reserve FRED Database

KP INTERNET TRENDS 2017 | PAGE 77 …And Now We Have a New Kind of Store = A Subscription Store Amazon Subscription Store = Central Hub for Monthly Services, 4/17 Entertainment Education Professional Services Cloud Storage News Source: Amazon

KP INTERNET TRENDS 2017 | PAGE 78 More / Faster Than Ever = Great Products Find Customers… Customers Find Great Products… Process + Data Collection + Intermediaries = Changing @ Torrid Pace…

KP INTERNET TRENDS 2017 | PAGE 79 1) Ad Growth = Driven by Mobile 2) Ad Measurability = Can Be Triple-Edged 3) Ads Evolving Rapidly = Often Organic + Data @ Core 4) Ads = Becoming Targeted Storefronts 5) eCommerce Growth = Accelerating, Again 6) eCommerce A-Ha’s… Online Advertising (+ Commerce) = Increasingly Measurable + Actionable

KP INTERNET TRENDS 2017 | PAGE 80 INTERACTIVE GAMES = MOTHERLODE OF… TECH PRODUCT INNOVATION / EVOLUTION + MODERN LEARNING WITH THANKS TO BING GORDON FOR INSIGHT + INSPIRATION

KP INTERNET TRENDS 2017 | PAGE 81 Global Interactive Gaming = Mainstream / Evolving Rapidly / Still Early Days… 2.6B Gamers* vs. 100MM in 1995 Source: Unity Q1:17 estimate (5/17), Electronic Arts 2016 estimate (12/16), Electronic Arts 1995 estimate (5/17) *Unity estimates reflect the total number of users seen playing mobile games (at least once every three months) powered by both proprietary and leading 3rd party game engines. This number assumes all PC or Console gamers also play at least 1 mobile game.

KP INTERNET TRENDS 2017 | PAGE 82 Gaming Evolution = Individual Play  Global Collaborative Play (1967-2017)… 45 Years Solo – Living Room… …Many – Arena (Thousands)… Online (Millions) Moore’s Law (Processing) Zuckerberg’s Law* (Sharing) 1 Player = Arcade 2 Players = Consoles 2+ Players = Consoles + LAN Millions of Players = Online Network Millions of Players + Spectators = eSports Source: Images: National Museum of American History (Brown Box), Wikipedia Creative Commons (Pac-Man, Atari 2600, SG-1000, SNES, N64, PS1, Xbox, PS2), Flickr user Sham Hardy (World of Warcraft), Flickr user coneybeare (Words with Friends), ESL (ESL Logo), Twitch (Twitch Logo), Major League Gaming (MLG Logo), Wikimedia Creative Commons (Pong), Flickr user BagoGames (eSports Stadium) Note: In 1967 TV Game Unit #7, also known as the “Brown Box” was launched as a prototype and is considered the father of video game consoles per the National Museum of American History. *Zuckerberg’s Law describes the exponential growth of online social networks as per Saul Hansell in NY Times, 11/6/08.

KP INTERNET TRENDS 2017 | PAGE 83 Source: Images: Wikimedia Creative Commons (Pong, Asteroids, Space Invaders, Pac-Man), Flickr user BagoGames (Mario Bros), Mobygames (John Madden Football), Electronic Arts (FIFA), Pokémon (Pokémon Red and Blue versions), World of Warcraft (Warcraft), Supercell (Clash of Clans), Minecraft (Minecraft Logo), Riot Games (League of Legends), King (Candy Crush Saga), Activision Blizzard (Overwatch), Pokémon Go (Pokémon Go) …Gen X + Millennials = Gamified Since Birth 1970 2000 20101980 1990 Gen X Millennials

KP INTERNET TRENDS 2017 | PAGE 84 Gaming = Large + Broad + Growing Business… Revenue @ $100B, +9% Y/Y Interactive Gaming Revenue Estimates per Newzoo, Global, 2016 $3 $3 $4 $17 $25 $47 Middle East & Africa Eastern Europe Latin America Western Europe North America Asia Pacific Revenue ($B) Source: Newzoo Global Games Market Report (2016) Note: Excludes console / gaming PC hardware revenue.

KP INTERNET TRENDS 2017 | PAGE 85 Gamers = All Ages… 35 Year-Old Average, USA Gamer Demographics vs. Average Age, USA, 2003-2016 0 5 10 15 20 25 30 35 40 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% 2003 2004 2005 2006 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 AverageAge %ofGamers Under 18 18-50 50+ Average Age Source: Entertainment Software Association (ESA) Essential Facts About the Computer and Video Game Industry 2003-2016 Note: Based on a survey of 4,000 U.S. households.

KP INTERNET TRENDS 2017 | PAGE 86 Female Gamers = Players Since Early Days But Genres Vary… 2000 (Year) Marked Rise of Casual Female Gamer % of Female Players by Game Genre, Global, 1/17 2% 6% 7% 16% 18% 36% 42% 69% 69% Sports Racing First-Person Shooter MMOs (Sci-Fi) Action Adventure MMOs (Fantasy) Casual Puzzle Family / Farm Sim Match 3 Match 3 Pioneered by Diamond Mine / Bejeweled, 2000 Family / Farm Sim Pioneered by Sims, 2000 Source: Quantic Foundry, Top Right Image: Popcap, Bottom Left Image: MoMA Note: Each genre analyzed contained between 3-5 game titles. The median sample size for each game title was 1,184. And the median sample size for each genre was 4,657.

KP INTERNET TRENDS 2017 | PAGE 87 Gaming Tools = Pervasive Online… Can Optimize Learning + Engagement… Foundational for Internet Services

KP INTERNET TRENDS 2017 | PAGE 88 Gaming Tools = Can Optimize Learning + Engagement… Foundational for Internet Services Repetition Dynamic Difficulty Adjustment Solving Puzzles Planning Workflows Completing Projects Leveling Up Competing Exploring / Discovering Following Rules Collaborating – Social Connection / Leadership Observing Interacting With / Analyzing Data Self Optimizing Creative Story Telling

KP INTERNET TRENDS 2017 | PAGE 89 Source: Center quote from Len Schlesinger: “Failure doesn’t mean the game is over, it means try again with experience,” Global Leadership Summit (8/11/11), Images: Playpacmanonline.net Repetition = Learn from Losing… Trial & Error Gaming Lifecycle Try again, with experience Respawn Fail Play Test Tactics

KP INTERNET TRENDS 2017 | PAGE 90 …Dynamic Difficulty Adjustment = Ultimate Trial & Error Experience… Engaging Learning Process Machine-Learning Fine-Tunes Gaming Mechanics Source: Image: Games for Learning Institute

KP INTERNET TRENDS 2017 | PAGE 91 …Solving Puzzles = Pattern Recognition + Critical Thinking… Defined Rules + Strategy (Short-Form) Minesweeper Unstructured Puzzles (Long-Form) L.A. Noire Detective Cases Source: Left image: Game Set Watch, Right image: L.A. Noire (Rockstar Games)

KP INTERNET TRENDS 2017 | PAGE 92 …Planning Workflows = Manage Time + Resource Efficiency… Time Management Legend of Zelda: Majora’s Mask Quest Progress Resets Periodically Resource Management Starcraft II ‘Require More Minerals’ Source: Left image: Zelda Informer, Right image: Activision Blizzard Battle.net

KP INTERNET TRENDS 2017 | PAGE 93 …Completing Projects = Track Finish Line from Start… Focus on End Goal Pokémon ‘Gotta catch ‘em all!’ Track Experience Skyrim Source: Left Images: Logos Wikia, Bulbapedia, Right Images: Skyrim, YouTube user HighlandMarker, Portforward.com, metagamebook, Stack Exchange

KP INTERNET TRENDS 2017 | PAGE 94 …Leveling Up = On-Going Progress Measurement… Leveling Up Candy Crush Saga Gain Experience Completing Puzzles Quantified Mastery Max Level in World of Warcraft Source: Left Images: Apptipper, King, Right Image: Blizzardwatch

KP INTERNET TRENDS 2017 | PAGE 95 …Competing = Play Against Self + Others Sharpens Skills… Competing Against Yourself Time Trials in Mario Kart 64 Competing Against Others Scoring Goals Online in Rocket League Source: Left Image: YouTube user Drew Weatherton, Right Image: GameSpot

KP INTERNET TRENDS 2017 | PAGE 96 …Exploring / Discovering = Open Closed Doors…Hack to Improvement… Discovering Easter Eggs Silent Hill 2 + Tony Hawk’s Pro Skater 2 Discovering Glitches Secret Level in Super Mario Bros Source: Left Images: Nintendo, Right Images: Digital Trends, Games Radar

KP INTERNET TRENDS 2017 | PAGE 97 A game is a system in which players engage in an artificial conflict, defined by rules, that results in a quantifiable outcome. - Salen & Zimmerman, Rules of Play: Game Design Fundamentals, 9/03 …Following Rules = Structured Play… Players = Free to Break Rules… …But = Consequences Source: Salen & Zimmerman, Rules of Play: Game Design Fundamentals, Left image: YouTube user Ross Campbell, Right image: YouTube user x Pepper

KP INTERNET TRENDS 2017 | PAGE 98 …Collaborating – Social Connection / Leadership = Learn From / Work With Others… Blizzard = Millions Playing Together Online, Global Key Multiplayer Franchises = World of Warcraft + Diablo + Starcraft + Overwatch 21 29 28 26 26 33 42 41 41 0 5 10 15 20 25 30 35 40 45 Q1:15 Q2:15 Q3:15 Q4:15 Q1:16 Q2:16 Q3:16 Q4:16 Q1:17 MAU(MM) Source: Activision, Morgan Stanley Note: Graph emphasizes Blizzard over Activision and King users due to the multiplayer nature of most Blizzard franchises.

KP INTERNET TRENDS 2017 | PAGE 99 …Observing = Learn From Watching Others Perform… Live Streamed Gameplay Live Streamed Player Reactions Live Chat Interaction with Player Subscribe to Streamer 1.2 2.4 3.2 4.0 4.9 0.0 0.5 1.0 1.5 2.0 2.5 0 1 2 3 4 5 6 2012 2013 2014 2015 2016 UniqueMonthlyStreamers(MM) HoursStreamed(B) Hours Streamed Unique Monthly Streamers Twitch Hours Streamed vs. Unique Monthly Streamers Source: Left image: Twitch Streamer: cherrysamora, Twitch Annual Reports 2013-2016 Twitch Streaming 10MM DAU, 2/17

KP INTERNET TRENDS 2017 | PAGE 100 …Interacting With / Analyzing Data = Many Games Have Strong Math Underpinnings… Fantasy Sports Fans Engaged in Analytics, USA, 1988-2016 Live Stats Feed Into Video Games + Fantasy Sports 0% 5% 10% 15% 20% 25% 0 10 20 30 40 50 60 70 1988 1991 1994 2003 2004 2005 2006 2007 2008 2009 2010 2011 2014 2015 2016 %ofUSAPopulationOver18 FantasySportsPlayers*(MM) USA Fantasy Sports Players % of USA Population John Madden Football Released Source: FSTA, Left image: Flickr user We Are Social, U.S. Census Bureau *Fantasy Sports Players are defined as U.S. individuals aged 18+ having played fantasy sports in the past year. Based on survey of USA individuals aged 18+, n=1,000.

KP INTERNET TRENDS 2017 | PAGE 101 …Self-Optimizing = Driven by Math (Statistics / Metrics / Rankings)… In-Game Player Analytics / Dashboards Increasingly Found in Enterprise / Consumer Products / Services Madden 2017 Player Stats Looker Business Intelligence Dashboard Source: Top left image: YouTube user Brian Mazique, Bottom left image: Uproxx, Right Image: Looker

KP INTERNET TRENDS 2017 | PAGE 102 …Creative Story Telling = Can Be Master of a Universe Laying Building Blocks of a Virtual World Minecraft Choosing Gameplay Experience Mass Effect 3 Source: Left image: Gamepedia blog, Right image: Kotaku Minecraft

KP INTERNET TRENDS 2017 | PAGE 103 Gaming Tools = Can Optimize Learning + Engagement… Foundational for Internet Services Reputation / Rankings Digital Recognition Interactive Storytelling Interactive Learning Upgrades + Downloadable Content Secondary Markets Messaging Live Camera Angles Graphics Computation

KP INTERNET TRENDS 2017 | PAGE 104 Reputation / Rankings = Deep Roots in Gaming… Airbnb Superhost Program Recognizes Top Performing Hosts Early Gaming (1978) Mainstream Internet (Now) Space Invaders First Arcade Game to Record High Scores Source: Left image: Codexdex, Right image: Airbnb, Probnb

KP INTERNET TRENDS 2017 | PAGE 105 …Digital Recognition = Deep Roots in Gaming… Activision 2600 Games Physical Badges for In-Game Achievements Facebook Give Digital Badges to Others Early Gaming (1980) Mainstream Internet (Now) Source: Left image: Atari Age, Right image: Facebook

KP INTERNET TRENDS 2017 | PAGE 106 …Interactive Storytelling = Deep Roots in Gaming… Atari First Role Playing Game Netflix + Amazon / Twitch Experimenting with Interactive Shows Early Gaming (1980) Mainstream Internet (Now) Source: Left image: mprd.se, Right images: Netflix, Amazon

KP INTERNET TRENDS 2017 | PAGE 107 Duolingo Leveling Up in Languages Lemonade Stand Teaching Economics 101 …Interactive Learning = Deep Roots in Gaming… Early Gaming (1979) Mainstream Internet (Now) Source: Left Image: Archive.org, Right Image: Duolingo,

KP INTERNET TRENDS 2017 | PAGE 108 Tesla Over-the-Air Software Updates Sega Downloadable Content via Cable …Upgrades + Downloadable Content = Deep Roots in Gaming… Early Gaming (1993) Mainstream Internet (Now) Source: Left image: Gamecrate, Right image: Tesla

KP INTERNET TRENDS 2017 | PAGE 109 …Secondary Markets = Deep Roots in Gaming… Runescape Secondary Markets for Items / Currency Apple iMessage 3rd Parties Offer Sticker Packs Early Gaming (2001) Mainstream Internet (Now) Source: Left image: RPGStash, Right image: Macstories

KP INTERNET TRENDS 2017 | PAGE 110 …Messaging = Deep Roots in Gaming… 768MM DAU 12/16 1999 5MM DAU 1/17 2009 2013 9MM DAU 5/17 Early Gaming Mainstream Internet (Now) Source: WeChat 2016 Year End Report (12/16), Ali Rayl Interview (Head of Global Customer Experience at Slack) (1/17), Venture Beat (5/17), Top left image: Pingwest, Top right images: Tencent, Middle images: SiteProNews, Bottom left image: ifeng, Bottom right image: Corsair

KP INTERNET TRENDS 2017 | PAGE 111 …Live Camera Angles = Deep Roots in Gaming… Madden Football Unique Game Perspectives Cable TV Cameras Unique Angles of Live Games Early Gaming (1996) Mainstream Media (Now) Source: Left image: Electronic Arts, Right image: Giants NFL

KP INTERNET TRENDS 2017 | PAGE 112 …Graphics Computation = Deep Roots in Gaming NVIDIA Launches GeForce 256 GPU Early Gaming (1999) Mainstream Internet (Now) Many Companies GPUs Used for Artificial Intelligence Source: Left Images: NVIDIA, VGA Museum, Right images: Google Deepmind, Amazon, IBM

KP INTERNET TRENDS 2017 | PAGE 113 In Era of Perceived Disengagement = ‘Engagement’ + Measurement Rising…

KP INTERNET TRENDS 2017 | PAGE 114 Video Gaming = Most Engaging Form of Social Media Daily Minutes Spent per User Across Select Digital Media Platforms 51 50 35 30 21 0 10 20 30 40 50 60 Video Gaming (Consoles, 9/16) Facebook Ecosystem (4/16) King (Mobile Games, 5/17) Snapchat (5/17) Instagram (10/14) AverageDailyMinutesSpentbyActiveUsers Source: Global Web Index (9/16), Facebook Q1:16 Earnings Call (4/16) & Q3:14 Earnings Call (10/14), Activision Q1:17 Earnings Call (5/17), Snapchat Q1:17 Earnings Call (5/17) Note: Video Gaming (Consoles): Global survey, n=17,990, of console users aged 16-64 asking “Roughly how many hours do you spend playing on game consoles during a typical day.” Includes Xbox One, Nintendo Wii U, PS4, Xbox 360, PS3, Nintendo Wii King: Average time spent per DAU. King used to illustrate mobile gaming time spent given the global nature of the platform and large base of daily active users (peaked at 158MM as of Q1:15, 128MM in Q4:15 was last disclosure). Snapchat: Average of the 25-30 minutes of daily usage found in the S-1 filing.

KP INTERNET TRENDS 2017 | PAGE 115 Mobile Daily Gaming Session Duration = +33% (3/17 vs. 7/15), Global, per Unity Games Mobile Average Daily Gaming Session Duration on Unity Games, Global, 7/15 – 3/17 Source: Unity

KP INTERNET TRENDS 2017 | PAGE 116 When I play a video game, it’s the only time I put away the phone and forget it exists. Video games command your attention in a way that nothing else can or will. - Gary Whitta, Screenwriter, Rogue One: A Star Wars Story, 5/17 Source: GamesBeat Summit 2017: How games, sci-fi, and tech create real-world magic (5/12/17)

KP INTERNET TRENDS 2017 | PAGE 117 Perhaps Interactive Gaming Evolution / Growth / Usage… Has Been Helping Prepare Society for Ongoing Rise of Human-Computer Interaction?

KP INTERNET TRENDS 2017 | PAGE 118 Gaming Tools = Improving Human Performance… Virtual + Augmented Reality / Simulations / Real-Time Analytics

KP INTERNET TRENDS 2017 | PAGE 119 Immersive Gaming Tools = Improving Athlete Performance…

KP INTERNET TRENDS 2017 | PAGE 120 Video + Virtual Reality = Mental Reps Can Improve Performance STRIVR Labs + Stanford Football Utilize Video + Virtual Reality to Repeatedly Run Plays / Scenarios Source: STRIVR Labs, Inc.

KP INTERNET TRENDS 2017 | PAGE 121 Video + Machine Learning = Visuals + Deep Analytics Can Improve Performance Second Spectrum 150K+ Tracked Events per Game,* 5/17 0 2 4 6 8 10 0 5 10 15 20 25 Q1:16 Q2:16 Q3:16 Q4:16 Q1:17 VideosperTeam(K) Teams Teams Video Sessions per Team Video Analytics of Key Plays Teams vs. Video Sessions per Team Source: Second Spectrum Note: Video session is defined as every time a user at a team watches a play using the Second Spectrum system. *Events are data surrounding key in-game actions such as pick-and-roll defenses, off-ball screens, shot probability or rebound probability. This allows players to query specific tactical actions during a game to gain better insight into how individuals / the team played.

KP INTERNET TRENDS 2017 | PAGE 122 Audio + Guided Meditation = Mental Focus Can Improve Performance CJ McCollum, NBA Shooting Guard Uses Headspace to Maintain Focus, 6/16 There’s a lot of stress in my job…and a 10 minute Headspace meditation helps you take care of all of those things and more. - CJ McCollum, 4/17 Headspace Run Streak Reinforce Habits Source: Headspace

KP INTERNET TRENDS 2017 | PAGE 123 Physically Interactive Media (PIM) = Real-Time Activity / Analytics Can Boost Intensity / Focus for Athletes Peloton 100K+ Bike Subscribers (95% Retention After 1 Year)…400K+ Home Riders 1MM+ Home Workouts Streamed in 3/17 2 Workouts per Week per Subscriber Source: Peloton

KP INTERNET TRENDS 2017 | PAGE 124 I could go ten hours at a stretch [playing soccer video games] and I’d often spot solutions in the games that I parlayed into real life. – Zlatan Ibrahimovic, I Am Zlatan: My Story On and Off the Field, 6/14 Video Games = Simulations Can Improve Athlete Strategy + Performance… From FIFA Online… …To the Real Game Source: I Am Zlatan: My Story On and Off the Field 2014, Left Image: New York Times (10/16), Right Image: The Sun

KP INTERNET TRENDS 2017 | PAGE 125 …Video Games = Stats Can Assist Athletes + Coaches… Video Game Player Stats Real-Time Feedback Offline, 9/16 Hoffenheim Scout Discovers Roberto Firmino… Using Football Manager Video Game, 11/16 Players + Coaches View Digital Stats as Important Performance Measure Source: Left Image: Twitter user Michy Batshuay, Right Image: Hardware Zone

KP INTERNET TRENDS 2017 | PAGE 126 Game Year Teams Madden Actual Winner Score Winner Score Super Bowl LI 2017 Patriots vs. Falcons Patriots 27-24 Patriots 34-28 Super Bowl L 2016 Broncos vs. Panthers Panthers 24-20 Broncos 24-10 Super Bowl XLIX 2015 Patriots vs. Seahawks Patriots 25-24 Patriots 28-24 Super Bowl XLVIII 2014 Broncos vs. Seahawks Broncos 31-28 Seahawks 43-8 Super Bowl XLVII 2013 49ers vs. Ravens Ravens 27-24 Ravens 34-31 Super Bowl XLVI 2012 Patriots vs. Giants Giants 27-24 Giants 21-17 Super Bowl XLV 2011 Steelers vs. Packers Steelers 24-20 Packers 31-25 Super Bowl XLIV 2010 Saints vs. Colts Saints 35-31 Saints 31-17 Super Bowl XLIII 2009 Steelers vs. Cardinals Steelers 28-24 Steelers 27-23 Super Bowl XLII 2008 Patriots vs. Giants Patriots 38-30 Giants 17-14 Super Bowl XLI 2007 Colts vs. Bears Colts 38-27 Colts 29-17 Super Bowl XL 2006 Steelers vs. Seahawks Steelers 24-19 Steelers 21-10 Super Bowl XXIX 2005 Patriots vs. Eagles Patriots 47-31 Patriots 24-21 Super Bowl XXXVIII 2004 Patriots vs. Panthers Patriots 23-20 Patriots 32-29 …Video Games = Stats Can Be Predictive… Madden Super Bowl Winner Prediction Accuracy @ 71% (14 Years) Madden Football Super Bowl Predictions vs. Actual Results, 2004-2017 Source: Electronic Arts, ESPN, USA Today, Forbes

KP INTERNET TRENDS 2017 | PAGE 127 …Immersive Gaming Tools = Improving Performance Across Disciplines

KP INTERNET TRENDS 2017 | PAGE 128 Gamification = Influencing Multiple Consumer Services… Education Duolingo Personal Finance Acorns Energy Conservation Nest Personal Health Mango Health Exercise myfitnesspal Food Starbucks Dating Bumble Advertising Snapchat Source: Top Row Images: Duolingo, Mango, Acorns, Nest, Bottom Row: iPhone in Canada (Starbucks), Consumer fitness news (Myfitnesspal), 5why.com (Bumble), Snapchat

KP INTERNET TRENDS 2017 | PAGE 129 …Gamification = Influencing Multiple Businesses… Military TrainingHealthcare Research Foldit Pilot Training Boeing Healthcare Training Simulated Surgery Neuroscience PTSD Therapy Work Productivity Betterworks Source: Top Left Image: Fold.it, Top Middle Image: US Army Sgt 1st Class Caleb Barrieau, Top Right Image: Betterworks, Bottom Left Image: Boeing, Bottom Center Image: Simulated Surgical Systems, Bottom Right Image: Archpaper.com

KP INTERNET TRENDS 2017 | PAGE 130 …Gamification = Influencing Complex Virtual Worlds + Real-World Simulations Improbable in Real World Simulate Cities + Power / Web Networks Improbable in Gaming Simulate Vast Virtual Worlds Source: Worlds Adrift: Bossa Studios, Improbable

KP INTERNET TRENDS 2017 | PAGE 131 As Rapid Data Growth Continues = Gaming Tools / Interfaces / Processors Will Continue to Organize + Drive Usefulness

KP INTERNET TRENDS 2017 | PAGE 132 Data Volume Growth Continues @ Rapid Clip… % Structured / Tagged (~10%) Rising Fast… 0 20 40 60 80 100 120 140 160 180 Zetabtytes(ZB) 2005: 0.1 ZB 2010: 2 ZB, 9% 2015: 12 ZB, 9% % Structured / Tagged Information Created Worldwide = Expected to Continue Accelerating 2020: 47 ZB, 16% 2025E: 163 ZB, 36% Source: IDC DataAge 2025 Study, sponsored by Seagate (3/17) Note: 1 petabyte = 1MM gigabytes, 1 zeta byte = 1MM petabytes

KP INTERNET TRENDS 2017 | PAGE 133 …GPU Processing Power Ramp Continues NVIDIA Transistors, 1998-2016 Source: NVIDIA Note: 1 GFLOP = 1B FLOPS, or “floating point operations per second.” 0 1,000 2,000 3,000 4,000 5,000 6,000 1998 2000 2002 2004 2006 2008 2010 2012 2014 2016 Transistors(MM) Transistors 2002 = 5 GFLOPS Battlefield 1942 2007 = 350 GFLOPS Unreal Tournament 3 2016 = 10K GFLOPS Paragon

KP INTERNET TRENDS 2017 | PAGE 134 Gaming Platforms = Evolving @ High Speed

KP INTERNET TRENDS 2017 | PAGE 135 New Gaming Development Tools / Platforms = Evolving to Continue to Build Virtual Worlds… Development Platforms In-Game Sandboxes Developers Players Have Virtual Experiences Build / Share / Explore Creations Build Virtual Worlds / Share Ideas Construct Virtual Worlds with New Dimensions Build / Share Creations Distribute Content Discover / Buy / Share Content Explore Virtual Worlds VR / AR Platforms Gaming Marketplaces Source: Unreal, Unity, HTC Vive, Oculus (Facebook) Microsoft, Minecraft, Roblox, Tencent, Steam (Valve), Sony

KP INTERNET TRENDS 2017 | PAGE 136 …New Gaming Development Tools / Platforms = Supporting Rapid Growth 0 1 2 3 4 5 6 7 RegisteredDevelopers(MM) 0 10 20 30 40 50 60 May-16 Jun-16 Dec-16 Mar-17 MAU(MM) 0 5 10 15 PeakConcurrent Users(MM) Unity = Registered Developers Roblox = Monthly Active Users Steam = Peak Concurrent Users* Source: Unity, Roblox, Steam (Valve), Forbes, Venturebeat, Bloomberg *Taken on the last available day of each month using waybackmachine.org.

KP INTERNET TRENDS 2017 | PAGE 137 eSports = Expanding Gaming Ecosystem via Fans / Spectators

KP INTERNET TRENDS 2017 | PAGE 138 eSports = 45 Year Evolution to Global Stage 1997 Red Annihilation Quake Tournament = Early eSports Competition 2000 Electronic Sports League + Korea eSports Assn. = Emerge as First eSports Leagues 2006 Justin.tv Founded = Precursor to Twitch.tv 2016 League of Legends 2016 World Championship = 43MM viewers 2009 League of Legends Released = Becomes One of Most Played Strategy Games (100MM MAU, 9/16) 2012 OnGameNet Begins Broadcasting League of Legends = First Major Korean Tournament on TV Evolution of Global eSports 1972 Stanford University AI Lab = First Ever Gaming Tournament (Spacewars) 1980 Atari Space Invader Competition = Early National Gaming Tournament Source: Spacewars image: Kokatu, Atari image: Ausretrogamer.com, Red Annihilation image: timetoast, Fortune (7/15), ESL logo: ESL, MLG, KeSPA Logo: Team Liquid, MLG Logo: MLG, Halo 2 Logo: Pinterest, Justin.tv Logo: startupgenome.com, Twitch Logo: Twitch, Riot Games (10/08), League of Legends Logo: League of Legends, Forbes (9/16), na.lolesports, OGN Logo: Twitch, eSports.inquirer.net, eSportsTV logo: mb.cision.com, League of Legends World Championship

KP INTERNET TRENDS 2017 | PAGE 139 eSports = People Watch What They Play… League of Legends Expands from Home to Staples Center, LA (Worlds 2016 Finals = ~20K in Stadium + 43MM Online) Source: Top left image: Mel Melcon Los Angeles Times, Bottom left image: Dexerto, Top right image: Red Bull, Bottom right image: YouTube

KP INTERNET TRENDS 2017 | PAGE 140 …eSports Trending vs. Traditional Sports = Very Strong with Younger Generations Millennials = 27% ‘Significant Preference’ for eSports vs. 27% for Traditional Sports Non-Millennials = 45% for Traditional Sports vs. 13% for eSports Which do you prefer, your favorite traditional sport or favorite eSport? 21% 27% 13% 13% 13% 13% 19% 18% 20% 13% 15% 11% 34% 27% 45% 0% 10% 20% 30% 40% 50% 60% 70% 80% 90% 100% All Responses All Millennials All Non-Millennials %ofRespondentsthatindicatedat leastaslightinterestineSports Significantly prefer favorite eSport Slightly prefer favorite eSport No preference Slightly prefer favorite traditional sport Significantly prefer favorite traditional sport Source: L.E.K. Sports Survey, Digital Engagement Part One: Sports and the “Millennial Problem” (2/17)

KP INTERNET TRENDS 2017 | PAGE 141 eSports Monthly Viewers @ 161MM… +40% Y/Y & Accelerating eSports Monthly Viewers, Global, 2012-2016 58 74 90 115 161 28% 22% 28% 40% 0% 5% 10% 15% 20% 25% 30% 35% 40% 45% 0 20 40 60 80 100 120 140 160 180 2012 2013 2014 2015 2016 Y/YGrowth(%) eSportsViewers(MM) eSports Monthly Viewers Y/Y Growth Source: Newzoo Global eSports Market Report (2/17), Newzoo press release (1/16), Newzoo Casual Connect Europe Presentation (2/15) *eSports Enthusiasts watch eSports once a month and/or participate in tournaments.

KP INTERNET TRENDS 2017 | PAGE 142 eSports League of Legends Championship Viewers @ 43MM… +19% Y/Y League of Legends World Championship Global Viewership Largest eSports Viewer Base 2 8 32 27 36 43 0 1 9 11 14 15 0 5 10 15 20 25 30 35 40 45 50 2011 2012 2013 2014 2015 2016 Viewers(MM) Total Unique Viewers Peak Concurrent Viewers Total Unique Viewers vs. Peak Concurrent Viewers Source: Engadget, Polygon, The Verge, eSports Marketing, LoLeSports

KP INTERNET TRENDS 2017 | PAGE 143 eSports Monthly Viewers = 79% <35 Years Old…29% Female Monthly eSports Viewers by Age / Gender, Global, 2016 27% 53% 18% 2% 0 20 40 60 80 100 120 10-20 21-35 36-50 51-65 GlobaleSportsEnthusiasts(MM) Male Female % of Total Monthly eSports Viewers Source: Newzoo 2017 Global eSports Market Report (2/17)

KP INTERNET TRENDS 2017 | PAGE 144 eSports (Like Sports) = Money Follows Viewers + Winners… Fan In-Game Purchases Boost Prize Pools Prize Pool for The International (DOTA 2), 2011-2017 $1.6 $1.6 $2.9 $10.9 $18.4 $20.8 $10.7 $0 $5 $10 $15 $20 $25 2011 2012 2013 2014 2015 2016 2017 YTD PrizePool($MM) Base Prize Pool Compendium* Source: ESPN, eSports Earnings, DOTA 2 (5/24/17) Note: * The International Compendium represents 25% of in-game purchases during a promotional period leading up to the event. Players can buy virtual items, levels, and other in-game content. As the total prize pool reaches different milestones all players who participated gain access to more exclusive content. 2017 YTD as of 5/25/17.

KP INTERNET TRENDS 2017 | PAGE 145 Facebook = Expands eSports Relationships with ESL Streaming Deal, 5/17 NBA + Up with Take Two = 2K eSports League, 2/17 German Soccer Club, FC Schalke 04 = Acquires eSports Team, Elements, 5/16 Partnerships + Investments = Helping Bring eSports into Mainstream Philadelphia 76ers = Acquire eSports Teams, Dignitas & Apex, 9/16 Miami Heat = Invests in eSports Team, Misfits, 1/17 Expanding Connections with Sports / Media Platforms Italian Soccer Club AS Roma = Partners with eSports Team, Fnatic, 2/17 Riot Games + BAMTech = $300MM 6yr LoL Streaming Rights, 12/16 Source: ESPN, Engadget, Yahoo Sports, Live Production, Dot eSports, Forbes

KP INTERNET TRENDS 2017 | PAGE 146 Gaming Experience => Technology Leadership + Innovation?

KP INTERNET TRENDS 2017 | PAGE 147 Ten Years Ago (2007) = A Stanford Professor Said… If you want to see what business leadership may look like in three to five years, look at what’s happening in online games. - Byron Reeves, Professor of Communication, Stanford University, 6/07 Source: Byron Reeves

KP INTERNET TRENDS 2017 | PAGE 148 …~Ten Years Later = Entrepreneurs Often Fans of Gaming Experience I like video games. In fact, that’s what got me into software engineering when I was a kid. I wanted to make money so I could buy a better computer so I could play better video games. - Elon Musk, CEO Tesla & SpaceX, 10/16 As a child I played a lot of Avalon Hill board games. And each board game is actually a complex set of rules and circumstances… So it was actually in fact my childhood gaming — for being able to build a model of what a game was — that was essentially the fundamental thing that informs my strategic sense. - Reid Hoffman, Co-Founder of LinkedIn, 8/15 I do think this dynamic around kids growing up, building games, and playing games, is an important one because I think this is how a lot of kids get into programming. I definitely wouldn't have gotten into programming if I hadn't played games. - Mark Zuckerberg, CEO Facebook, 5/15 Source: Elon Musk: Forbes Interview (10/1/16), Reid Hoffman: Interview on the Tim Ferris Show (8/31/15), Mark Zuckerberg: Facebook Q&A Session (5/14/15)

KP INTERNET TRENDS 2017 | PAGE 149 Perhaps Interactive Gaming Evolution / Growth / Usage With Related Data Collection / Analytics / Real-Time Simulations + Engagement… Has Been Helping Prepare Society for On-Going Rise of Human-Computer Interaction?

KP INTERNET TRENDS 2017 | PAGE 150 1) Global Gaming = Mainstream / Evolving Rapidly / Still Early Days 2) Gaming Tools = Pervasive Online 3) Gaming Tools = Improving Human Performance 4) Gaming Platforms = Evolving @ High Speed 5) eSports = Expanding Gaming Ecosystem via Fans / Spectators 6) Gaming Experience => Technology Leadership + Innovation? Interactive Games = Motherlode of… Tech Product Innovation + Modern Learning

KP INTERNET TRENDS 2017 | PAGE 151 MEDIA = DISTRIBUTION DISRUPTION @ TORRID PACE

KP INTERNET TRENDS 2017 | PAGE 152 Digital Leaders = Transforming Media With Better User Experiences + Lower Prices…Data + Scale

KP INTERNET TRENDS 2017 | PAGE 153 Music = Why Streaming? Access / Choice / Discovery / Personalization / Mobile / Fewer Ads Source: Goldman Sachs Research, BPI Note: BPI Survey as of 12/15, n=1,000 (UK only). Questions: “Why did you decide to pay for a music streaming subscription?” and “Thinking about music streaming, to you, how important are the following?” 9% 19% 21% 22% 24% 27% 29% 42% Viewed Ad Offline Listening Bundled Recs from Friends / Family Listening Choice Mobile Access Get Rid of Ads Free Trial Convert Reasons for Paying for Music Streaming, 12/15 Importance of Streaming Product Features, 12/15 0% 20% 40% 60% 80% 100% Share Playlists Build Playlists Simultaneous Music Videos Curation / Recs Keep Up with Hits Support Artists Multi-Device Listening New Music Discovery Size of Catalog Very Fairly Not much Not at all

KP INTERNET TRENDS 2017 | PAGE 154 Video = Why Cord-Cutting? Lower Price + Convenience Source: TiVo Q4 2016 Video Trends Report Note: Survey includes 18+ year olds in USA and Canada, n=3,079. Other categories omitted include “Not My Choice,” “Share SVOD Login,” “Moved In With New Roommate,” “Other.” 11% 13% 19% 27% 48% 80% Bulk of Viewing is Streaming Service Original Content Dropped Cable Upon Moving / Relocating Like to Binge Entire Seasons via Streaming Use Antenna to Get Basic Channels Use an Internet Streaming Service Too Expensive Reasons for Cutting Pay-TV Service, Q4:16

KP INTERNET TRENDS 2017 | PAGE 155 Digital Evolution of Music + Video = Ramping Rapidly…

KP INTERNET TRENDS 2017 | PAGE 156 $0 $5 $10 $15 $20 1973 1976 1979 1982 1985 1988 1991 1994 1997 2000 2003 2006 2009 2012 2015 RecordedMusicRevenues($B) Subscription & Streaming Download & Synchronization Physical Source: Recording Industry Association of America Note: “Subscription and Sttreaming” includes paid subscriptions (full and limited tier), SoundExchange, ad-supported streaming and other digital. “Download & Synchronization” includes download single / album, kiosk, download music video, ringtone / ringback, and synchronization. “Physical” includes LP / EP, vinyl single, 8-track, cassette (full / single), CD / CD single, music video, DVD audio and SACD. Recorded Music = Revenue +11% After 16 Years of -4% Annual Average Growth… Subscription & Streaming @ 52% of Revenue vs. 0% Thirteen Years Ago, USA Recorded Music Revenues by Format ($B), USA, 1973-2016

KP INTERNET TRENDS 2017 | PAGE 157 Source: Spotify * Subscribers as of 3/2017, when Spotify announced they had reached the 50 million subscriber mark. Spotify = Catalyst for Internet-Driven Evolution of Music Industry… 0  50MM Paid Subscribers / 126MM MAUs in <9 Years… €0 €500 €1,000 €1,500 €2,000 €2,500 €3,000 0 10 20 30 40 50 60 2008 2009 2010 2011 2012 2013 2014 2015 2016* SpotifyGlobalRevenue(€MM) SpotifyPaidSubscribers(MM) Spotify Paid Subscribers Spotify Global Revenue Spotify Subscribers (MM) & Revenue (€MM), 2008 – 2016*, Global Q4:16 Monthly ARPU = €5.80 ($6.10)

KP INTERNET TRENDS 2017 | PAGE 158 Source: Spotify, IFPI 2017 Global Music Report * Subscribers as of 3/2017, when Spotify announced they had reached the 50 million subscriber mark. …Spotify = 20% of Global Music Industry Revenue vs. 0% in 2008 0% 5% 10% 15% 20% 25% 30% 0 10 20 30 40 50 60 2008 2009 2010 2011 2012 2013 2014 2015 2016* Spotifyas%ofGlobalMusicRevenue SpotifyPaidSubscribers(MM) Spotify Paid Subscribers (MM) Share of Global Music Revenue (%) Spotify Subscribers (MM), 2008 – 2016*, Global Q4:16 Monthly ARPU = €5.80 ($6.10)

KP INTERNET TRENDS 2017 | PAGE 159 Source: Spotify Spotify = Users Listen to 41 Artists per Week, +40% (vs. 1/14) Owing to… Recommendation Engine (Data + Algorithms) 0 10 20 30 40 50 1/14 1/15 1/16 1/17 UniqueArtistsListenedtoperWeek Time Spotify Unique Artists Listened to Per Week, Average, Global, 1/14-1/17

KP INTERNET TRENDS 2017 | PAGE 160 Network TV* Minutes Delivered = 2011 Top 5 Networks -10% Average… Netflix +669% Over 5 Years, USA… Source: Matthew Ball – REDEF Original 3/14/16, Nielsen, Sandvine, Netflix, SNL Kagan, BTIG Note: Inclusive of Broadcast + Basic Cable + Premium Cable, C7 Live + VOD + DVR. Does not account for multiple viewers (i.e. unique minutes delivered) or TV everywhere (though note that even if every TV Everywhere stream started in 12/15 was completed and 1 hour long, consumption would have increased national TV time by only 1.9%). 0 100 200 300 400 NBC Universal Disney 21st Century Fox Netflix CBS Time Warner Viacom Discovery A&E ESPN MonthlyMinutesDelivered(B) 2010-2011 (through Feb) 2015-2016 (through Feb) Monthly Minutes Delivered By Network Group, USA, 2010/11-2015/16 669% 11% 14% 30%21% 35% 16% 33%1% 1%

KP INTERNET TRENDS 2017 | PAGE 161 $0 $600 $1,200 $1,800 $2,400 $3,000 0 20 40 60 80 100 NetflixQuarterlyRevenue($MM) NetflixPaidSubscribers(MM) Netflix Paid Subscribers (MM) Netflix Quarterly Revenue ($MM) 9/11 DVD / Streaming Split Source: Netflix Note: Netflix subscription DVD service launched 9/1998. Data before Q3 2001 represents all subscribers because paid subscribers not broken out. Netflix split streaming subs from DVD subs in Q3 2011; graph shows only streaming subs thereafter. ARPU shown ex-DVD. Netflix = Catalyst for Internet-Driven Evolution of Video Industry… 95MM Streaming Subscribers in 10 Years… Netflix Subscribers (MM) & Quarterly Revenue ($MM), 2/99 – 3/17, Global Q1:17 Streaming ARPU per Month = $9.14 9/99 DVD Subscription Launch 1/07 Streaming Launch

KP INTERNET TRENDS 2017 | PAGE 162 0% 10% 20% 30% 40% 50% 0 20 40 60 80 100 2007 2008 2009 2010 2011 2012 2013 2014 2015 2016 2017* ShareofUSAHomeEntertainmentRevenue(%) NetflixPaidSubscribers(MM) Netflix Paid Subcribers (MM) Share of USA Home Entertainment Revenue (%) Source: Netflix Note: Share represented by Netflix domestic streaming revenue over total home entertainment revenue in USA. Domestic streaming not broken out as individual segment until 2012. Netflix split streaming subs from DVD subs in Q3 2011; graph shows only streaming subs thereafter. * Q1:17 represents Netflix annualized domestic streaming revenue figure. ARPU shown ex-DVD …Netflix Streaming = From 0% to >30% of Home Entertainment Revenue in 10 Years, USA Netflix Subscribers, 2009 – 2017*, Global Q1:17 Streaming ARPU per Month = $9.14 9/11 DVD / Streaming Split 1/07 Streaming Launch

KP INTERNET TRENDS 2017 | PAGE 163 Google Pioneered Search / Find / Obtain (SFO) for Content + Products… Netflix + Spotify Pioneered Search / Find / Serve Up (SFS) for Media Source: Netflix; Spotify; Michelle Ufford, “Data-Driven at Netflix”, talk given at PASS 10/31/16; Gomez-Uribe and Hunt (both Netflix), “The Netflix Recommender System: Algorithms, Business Value, And Innovation”, ACM Transactions on Management Information Systems 6.4, 12/15 Note: Netflix estimated cost savings due to improved engagement and reduction of monthly churn, driving lower need for subscriber acquisition cost in the future. 98MM Different Netflixes... $1B cost savings / year from recommendations (12/15) 126MM Different Spotifys… ~5B Discover Weekly streams in <1 year post-launch (5/16) From Give to Get…With Data + Algorithms

KP INTERNET TRENDS 2017 | PAGE 164 Digital Evolution of Music + Video = Multiple Approaches…

KP INTERNET TRENDS 2017 | PAGE 165 Facebook / Instagram / Snap = Mobile Video Traffic Share Gainers Over 4 Years… Source: Sandvine Global Internet Phenomena Report (2H 2012 and 2016) 0% 5% 10% 15% 20% 25% 30% 35% 40% YouTube Facebook Instagram Snap Netflix iTunes Google Cloud HTTP / SSL - Other Other 2012 2016 Share of Downstream Video Traffic (%), North America, 2H 2016

KP INTERNET TRENDS 2017 | PAGE 166 …Netflix / YouTube = Fixed-Access Video Traffic Share Leaders 0% 5% 10% 15% 20% 25% 30% 35% 40% Netflix YouTube Amazon Video iTunes Hulu Xbox One FacebookBitTorrent HTTP / SSL - Other Other 2012 2016 Share of Downstream Video Traffic (%), North America, 2H 2016 Source: Sandvine Global Internet Phenomena Report (2H 2012 and 2016)

KP INTERNET TRENDS 2017 | PAGE 167 Facebook Platform MAUs, Global, Months Since Launch Facebook (Facebook / WhatsApp / Messenger / Instagram) = Video Ramping Across Platform Source: Facebook, Instagram, Whatsapp, Financial Times, TechCrunch 0 500 1,000 1,500 2,000 2,500 0 12 24 36 48 60 72 84 96 108 120 132 144 156 MAU(MM) Months Since Founding Instagram Facebook WhatsApp Facebook Messenger

KP INTERNET TRENDS 2017 | PAGE 168 Snap = Ramping Original Short-Form Content Snap ‘Original Shows’ Source: Snap Second Chance 8MM+ Views for 1st Episode, 5/17 Phone Swap 10MM+ Views for 1st Episode, 5/17

KP INTERNET TRENDS 2017 | PAGE 169 Generational Media Usage = Chasm Increasing… Shifts to Internet-Enabled Media Continue

KP INTERNET TRENDS 2017 | PAGE 170 7:16 7:12 7:06 4:14 2:44 2:17 Q4:16 Q4:15 Q4:14 Analog Digital Mobile Device Time per Day = +2x Over 2 Years… Daily Time Spent by Media (Not De-Duped), USA, Q4:14-Q4:16 Source: Nielsen Total Audience Report Q4:16 Note: “Analog” includes Live / DVR / Time-shifted TV, DVR / time-shifted TV, AM / FM radio, DVD / Blu-ray, and game consoles. “Digital” includes Multimedia devices (viewing on Apple TV, Roku, Chromecast, smartphone, computer etc. connected to TV), internet on PC, video on PC, app / web on smartphone / tablet, and video on smartphone.

KP INTERNET TRENDS 2017 | PAGE 171 …Mobile Device Time per Day = 18-24 Year-Olds @ 49% Digital…65+ Year-Olds @ 13%, USA 9:49 9:09 7:17 5:42 4:35 1:30 4:41 5:19 4:42 4:27 65+ 50-64 35-49 25-34 18-24 Analog Digital Daily Time Spent by Media & Age Bracket (Not De-Duped), USA, Q4:16 Source: Nielsen Total Audience Report Q4:16 Note: “Analog” includes Live / DVR / Time-shifted TV, DVR / time-shifted TV, AM / FM radio, DVD / Blu-ray, and game consoles. “Digital” includes Multimedia devices (viewing on Apple TV, Roku, Chromecast, smartphone, computer etc. connected to TV), internet on PC, video on PC, app / web on smartphone, and video on smartphone.

KP INTERNET TRENDS 2017 | PAGE 172 Traditional Cable Conundrum = Channels + Consumer Prices + Programming Costs Rising… Subscribers Falling

KP INTERNET TRENDS 2017 | PAGE 173 Pay TV Household Growth = -1.3% Average for Last 12 Quarters… While Programming Costs >2x+ since 2006… Source: Nielsen Total Audience / Cross Platform Reports, US Census Bureau, St. Louis Federal Reserve FRED Database Note: Pay TV households represented by Nielsen “Cable Plus” metric, which includes households who receive television via Wired Cable (No Telco), Telco, or Satellite. “Programming Costs” includes total program and production costs for Cable and Other Subscription Programming firms, 2006-2015, as per US Census Services Annual Survey for Employer Firms ($25B in 2015, up from $12B in 2006). Pay TV Households (MM), USA, 2010-2016 (5%) (3%) (1%) 1% 3% 5% 60 70 80 90 100 110 120 Q1:10 Q2:10 Q3:10 Q4:10 Q1:11 Q2:11 Q3:11 Q4:11 Q1:12 Q2:12 Q3:12 Q4:12 Q1:13 Q2:13 Q3:13 Q4:13 Q1:14 Q2:14 Q3:14 Q4:14 Q1:15 Q2:15 Q3:15 Q4:15 Q1:16 Q2:16 Q3:16 Q4:16 %Y/YGrowth TotalTVChannelsReceivedinUSA Pay TV Households (MM) % Y/Y Growth

KP INTERNET TRENDS 2017 | PAGE 174 …# TV Channels Watched <10% of Channels Received… Pay TV ARPU 10-15x > Netflix… Source: Nielsen, Matthew Ball & Tal Shachar, REDEF Original 3/9/16, DirecTV, AT&T, Charter, Dish Network, Comcast Note: TV channel data as of mid-year. DirecTV ARPU calculated by dividing the 2016 Video Entertainment revenue by the average number of Linear Video Connections during 2016. Charter ARPU calculated by dividing 2016 Video revenue by average Video Residential Primary Service Units during 2016. Dish Network ARPU calculated by multiplying the 2016 Pay-TV Average Monthly Revenue per Subscriber by 12. Comcast ARPU calculated by dividing the 2016 Residential Video revenue by the average Video Customers in 2016. Netflix ARPU is based off the Global Streaming revenue and average subscribers in 2016. All estimates are global. Average TV Channels Received vs. Watched per Household, USA, 2008-2016 0% 5% 10% 15% 20% 25% 0 50 100 150 200 250 %Watched TotalTVChannelsReceivedinUSA TV Channels Received % Channels Watched $103 $997 $1,064 $1,131 $1,439 Netflix Comcast Dish Charter DirecTV Annual ARPU, Selected Platforms, 2016 Pay TV Digital Platforms

KP INTERNET TRENDS 2017 | PAGE 175 Digital Subscriptions = Rising Owing to Massive User Experience Improvements… On-Demand / A La Carte Selection + Choice / Personalization / Payment Systems / 2-Way UGC / Mobile...

KP INTERNET TRENDS 2017 | PAGE 176 Media Evolution (1950-2017) = Market of Millions  Market of One x Millions Source: Amanda Lotz, The Television Will Be Revolutionized, New York: NYU Press, 2007, Wikimedia Commons, Google Image Commons, Crunchyroll, Shudder, Fandor, Seeso, Cheddar, Fullscreen Cater to Sub-Genres / Power Users / A La Carte + Subscription Network Era 1950s-1980s Cable Era 1980s-2010s Digital Era Current Broad Genres / Focus on Programming / Limited Bundle Choices Cater to All / High Viewership / No Personalization Digital Distributors Digital Studios

KP INTERNET TRENDS 2017 | PAGE 177 Media = Distribution Disruption @ Torrid Pace 1) Digital Leaders = Transforming Media With Better User Experiences + Lower Prices…Data + Scale 2) Generational Media Usage = Chasm Increasing as Shifts to Internet-Enabled Media Continue 3) Traditional Cable Conundrum = Channels + Consumer Prices + Programming Cost Rising…Subscribers Falling 4) Digital Subscriptions = Rising Owing to Massive User Experience Improvements (On-Demand / Selection + Choice / Personalization / Payment Systems / 2-Way UGC / Mobile...)

KP INTERNET TRENDS 2017 | PAGE 178 THE CLOUD = ACCELERATING CHANGE ACROSS ENTERPRISES ALEX KURLAND @ KLEINER PERKINS

KP INTERNET TRENDS 2017 | PAGE 179 1) Cloud Adoption = Reaching New Heights + Creating New Opportunities 2) Enterprise Software = Customer Expectations  Mirroring Those of Consumer Apps 3) Security = More Applications  More Vulnerabilities The Cloud = Accelerating Change Across Enterprises

KP INTERNET TRENDS 2017 | PAGE 180 Cloud Adoption = Reaching New Heights + Creating New Opportunities

KP INTERNET TRENDS 2017 | PAGE 181 Public + Private Clouds = Approaching Traditional Data Center Spend… +37% to $36B vs. 2014 76% 72% 67% 63% 14% 17% 20% 22% 9% 11% 13% 15% 0% 25% 50% 75% 100% 2013 2014 2015 2016 %ofTotalITInfrastructureSpend IT Infrastructure Spend, Global, 2014-2016 Traditional Data Center Public Cloud Private Cloud Source: IDC Worldwide Quarterly Cloud IT Infrastructure Tracker; Gartner; CloudHealth estimates

KP INTERNET TRENDS 2017 | PAGE 182 Public Cloud Adoption Trends = AWS Maintains Lead…Azure + Google Rising 57% 20% 10% 7% 57% 34% 15% 8% AWS Azure Google Cloud IBM 2016 2017 Public Cloud Adoption, 2016 vs. 2017 % of Respondents Running Applications 57% 34% 15% 8% 17% 21% 17% 9%10% 12% 13% 8% AWS Azure Google Cloud IBM Running Apps Experimenting Planning to Use Public Cloud Adoption, 2017 % of Respondents Running, Experimenting, or Planning to Use Applications Source: Rightscale 2017 State of the Cloud Report Note: Based on survey of IT Professionals, n=1,002.

KP INTERNET TRENDS 2017 | PAGE 183 Cloud Concerns = Shifting from Data Security + Cost Uncertainty  Vendor Lock-In + Compliance / Governance 42% 38% 33% 21% 21% 18% 14% 7% 35% 21% 19% 27% 14% 20% 18% 22% Data Security Uncertainty of Costs and Savings Loss of Control (Upgrades, Timing of Backups) Compliance / Governance Reliability (SLA Requirements) Data Portability and Ownership Software Compatability Lock-In (ability to change vendors) 2012 2015 Share of Respondents Citing Criteria as Top-Three Concern, USA, 2012-2015 Source: Bain Cloud Computing Survey, 2015 (n=347); Morgan Stanley AlphaWise Survey of IT Managers (n=304)

KP INTERNET TRENDS 2017 | PAGE 184 Cloud Evolution / Tools = Paving Way for… Innovation Across Infrastructure Landscape… Containers / Microservices = Simplify software development process / improve consistency between testing & production environments / reduce complexity of managing & updating apps due to modular approach Edge Computing = Pushing compute away from centralized nodes & closer to sources of data… addresses many IT challenges when running data-centric workloads in cloud – reduces latency / can have security + compliance benefits… Elastic Analytical Databases = Likes of Google BigQuery / Snowflake / AWS Redshift Spectrum nearly infinitely scalable / usage based + have minimal maintenance requirements New Methods of Software Delivery = APIs / Browser Extensions…creating new wave of capabilities (+ companies) for both companies and end users Source: Lloyd Tabb, Looker Founder & CTO; Happiest Minds; Azuqua; TheServerSide; Forbes

KP INTERNET TRENDS 2017 | PAGE 185 …New Cloud Companies Emerging… Providing Elegant + Intuitive Experiences for End Users Rubrik Managing data across cloud & on-prem infrastructure, approaching $100MM in annualized bookings Stripe Processing billions of transactions a year across 100K+ businesses in 100+ countries Looker Empowering data analysis for 40K users across every department, each averaging 2 new queries every day CloudHealth Actively managing more than 1.3MM policies globally for hybrid & multi-cloud environments Source: Company-provided & publicly available data; Snipcart

KP INTERNET TRENDS 2017 | PAGE 186 Enterprise Software = Customer Expectations  Mirroring Those of Consumer Apps

KP INTERNET TRENDS 2017 | PAGE 187 Enterprise Software (2000  2017) = Users Expect Products to be as… Well Designed / Easy-to-Use / Reliable as Consumer Apps Perpetual, On-Premise Software  Cloud-Based SaaS Apps  Mobile-First Smart Apps 2000 2017 Delivery Method On-Prem Cloud-based Pricing Perpetual License Subscription UX Generic Personalized Intelligence Constrained Unlimited (AI / ML) Growth Engine Sales Product Purchase Decision Top-Down Bottoms-Up Measure of Engagement & Customer Satisfaction N/A DAUs / MAUs / NPS

KP INTERNET TRENDS 2017 | PAGE 188 Design = Increasingly Core to Enterprise R&D… End-Users Demanding Consumer-Quality Product Experiences Change in Designer : Developer Ratio, Selected Enterprises, 2010-2017 Source: Company data, Figma Note: Ratios for entire orgs, unless noted otherwise. Atlassian historical ratio from 2012; Dropbox data for product org only; IBM historical ratio from 2012, data for product org only; Intercom data for product org only; LinkedIn historical ratio from 2010. 2010 - 2012 2017 1 designer : 1 designer : 25 developers 9 developers 1 designer : 1 designer : 10 developers 6 developers 1 designer : 1 designer : 8 developers 72 developers On Mobile – 1 designer : 3 developers 1 designer : 5 developers N/A 1 designer: 1 designer : 11 developers 8 developers N/A

KP INTERNET TRENDS 2017 | PAGE 189 Security = More Applications  More Vulnerabilities

KP INTERNET TRENDS 2017 | PAGE 190 Cloud-Enabled App Use in Enterprises = Rising Rapidly… Cheaper to Build / Easier to Adopt / Harder to Secure… Avg. # of Cloud Apps Used by Vertical, Global, April 2017 Avg. # of Cloud Services used by Category, Global, April 2017 This has serious security & compliance implications... 94% of all cloud apps used are not “enterprise-ready,” per Netskope 1,206 1,170 1,092 907 893 0 200 400 600 800 1,000 1,200 1,400 Retail, Restaurants, & Hospitality Financial Services, Banking, & Insurance Manufacturing Healthcare & Life Sciences Technology & IT services Avg.CloudServicesUsedperEnterprise Category # Per Enterprise % Not Enterprise Ready Marketing 91 97% HR 90 96% Collaboration 70 87% Finance / Accounting 60 95% CRM / Sales 43 94% Software Development 41 96% Productivity 37 95% Social 30 91% Cloud Storage 27 72% IT Service / Application Management 25 98% Source: Netskope April 2017 Note: 461 cloud apps in April 2017, one year ago = average of 917 from Feb-16 report & 935 from Jun-16 report; “Not enterprise ready” = received a rating of “medium” or below in the Netskope Cloud Confidence Index.

KP INTERNET TRENDS 2017 | PAGE 191 …Network Breaches = Increasingly Caused by Email Spam / Phishing… Spam +350% vs. Q1:15 Monthly Average… 0% 50% 100% 150% 200% 250% 300% 350% 400% 450% Spam Without Malicious Attachments Spam With Malicious Attachments Change in Amount & Type of Spam, Global, 2015-2016 Indexed to Q1:15 Monthly Average Source: AntiPhishing Working Group Phishing Activity Trends Report - Q4 2016; IBM X-Force Threat Intelligence Index 2017 %ChangeinSpam(IndexedtoQ1:15MonthlyAverage)

KP INTERNET TRENDS 2017 | PAGE 192 …Cyber Threats Severity Rising = 10MM+ Identities Exposed in… 15 Breaches in 2016…vs. 11 in 2014 0% 25% 50% 75% 100% 2012 2013 2014 2015 2016 %ofInternetTrafficbySource % Human % Bot 0 2 4 6 8 10 12 14 16 2014 2015 2016 #Breacheswith10MM+IdentitiesExposed Breaches with 10MM+ Identities Exposed, Global, 2014-2016 % of Internet Traffic by Source, Global, 2012-2016 Source: Incapsula 2016 Bot Traffic Report (100k Randomly Selected Domains); 2017 Verizon Data Breach Investigations Report

KP INTERNET TRENDS 2017 | PAGE 193 CHINA INTERNET = GOLDEN AGE OF ENTERTAINMENT + TRANSPORTATION *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 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. 下载中文完整版

KP INTERNET TRENDS 2017 | PAGE 194 China Macro = Positive Trends

KP INTERNET TRENDS 2017 | PAGE 195 48% 49% 50% 51% 52% 53% 90 95 100 105 110 115 1/14 4/14 7/14 10/14 1/15 4/15 7/15 10/15 1/16 4/16 7/16 10/16 1/17 ManufacturingPMIIndex ConsumerConfidenceIndex China Consumer Confidence Index (LHS) China Manufacturing PMI (RHS) China Macro = Confidence Improving Since CH2:16 Source: China National Bureau of Statistics, Bernstein Research Consumer Confidence Index & Manufacturing PMI Index, China, 1/14 – 3/17

KP INTERNET TRENDS 2017 | PAGE 196 China Macro = Service Sector @ 52% GDP Share vs. 23% Thirty-Five Years Ago 0% 10% 20% 30% 40% 50% 60% 1961 1966 1971 1976 1981 1986 1991 1996 2001 2006 2011 2016 %ofGDP Service Sector Output as % of Nominal GDP, China, 1961 – 2016 Source: China National Bureau of Statistics, Morgan Stanley Research Note: Service sector defined as all industries outside of agriculture, forestry, animal husbandry and fishery industries (except support services to agriculture, forestry, animal husbandry and fishery industries), mining (except auxiliary activities of mining), manufacturing (except repairs for metal products, machinery and equipment), production and supply of electricity, steam, gas and water, and construction.

KP INTERNET TRENDS 2017 | PAGE 197 5% 21% 48% 0% 10% 20% 30% 40% 50% 2005 2010 2016 %ofMSCIChinaMarketCap China Macro = Private (Non-SOEs) Enterprises… Increasingly Driving Wealth Creation + Economic Growth + Jobs Source: Morgan Stanley Research, MSCI *SOE = State Owned Enterprise. Private Enterprise (Non-SOE*) % Share of MSCI China Weighted Market Cap

KP INTERNET TRENDS 2017 | PAGE 198 China Macro = Technology Companies Lead Public Market Wealth Creation 0% 20% 40% 60% 80% 100% Telecommunication Services Energy Utilities Industrials Consumer Staples Materials Overall MSCI China Financials Consumer Discretionary Health Care Information Technology 2016 2005 Source: Morgan Stanley Research, MSCI *SOE = State Owned Enterprise. Private Enterprise (Non-SOE) % of MSCI China Market Cap by Sector, 2005 vs. 2016

KP INTERNET TRENDS 2017 | PAGE 199 China Internet Users + Usage = Healthy User Growth… Usage Outpacing Users

KP INTERNET TRENDS 2017 | PAGE 200 Mobile Internet Users & Y/Y Growth, China, 2008 – 2016 Source: CNNIC Note: Internet user data is as of year-end. 0% 10% 20% 30% 40% 50% 60% 70% 0 100 200 300 400 500 600 700 2008 2009 2010 2011 2012 2013 2014 2015 2016 %Y/YGrowth ChinaMobileInternetUsers(MM) China Mobile Internet Users Y/Y Growth China Mobile Internet Users = @ ~700MM, +12% Y/Y vs. 11% in 2015

KP INTERNET TRENDS 2017 | PAGE 201 Source: Hillhouse estimates based on daily media time spent data from ZenithOptimedia and mobile data from QuestMobile 0% 15% 30% 45% 60% 75% 90% 0 500 1,000 1,500 2,000 2,500 3,000 2012 2013 2014 2015 2016 %Y/YGrowth DailyTimeSpent(HoursMM) Mobile Internet Time Spent Mobile Internet Time Spent Y/Y Mobile Internet User Y/Y China Mobile Internet Usage Outpacing User Growth = +30% Y/Y for Usage…+12% for Users Estimated Mobile Internet Daily Time Spent, China, 2012 - 2016

KP INTERNET TRENDS 2017 | PAGE 202 China Entertainment = Online Innovation Driving Robust User + Usage + Monetization Growth

KP INTERNET TRENDS 2017 | PAGE 203 Source: Zenith Optimedia 0% 10% 20% 30% 40% 50% 60% 70% 0 50 100 150 200 250 300 350 2012 2013 2014 2015 2016 Internetas%ofMediaConsumption DailyMediaConsumption(Minutes) Radio TV Magazine Newspaper Desktop Internet Mobile Internet Internet as % of Total Media China Media = Internet @ 55% of Time Spent…Mobile > TV (2016) Average Daily Media Consumption Minutes by Medium, China, 2012 - 2016

KP INTERNET TRENDS 2017 | PAGE 204 0 500 1,000 1,500 2,000 2,500 3,000 3,500 11/14 12/14 1/15 2/15 3/15 4/15 5/15 6/15 7/15 8/15 9/15 10/15 11/15 12/15 1/16 2/16 3/16 4/16 5/16 6/16 7/16 8/16 9/16 10/16 11/16 12/16 1/17 2/17 3/17 4/17 AverageDailyHours(MM) WeChat QQ Tencent Games Tencent Video Tencent News Tencent Music Tencent Other UC Browser Weibo Taobao / Tmall Youku Alibaba Other iQiyi Mobile Baidu Baidu Other Toutiao NetEase All Other TencentBaiduAlibaba China Mobile Internet Daily Hours By App, 11/14 – 4/17 China Entertainment = Key Driver of Mobile Time Spent… eCommerce + Games = Monetize Best Per Time Spent… Source: QuestMobile Note: Only top 100 apps by time spent are categorized by company affiliation. Tencent, Alibaba and Baidu affiliates include strategically invested companies.

KP INTERNET TRENDS 2017 | PAGE 205 China Online Entertainment = Consumers Increasingly Willing to Pay… Led by Games + Livestreaming + Video $0 $10 $20 $30 $40 2011 2012 2013 2014 2015 2016 AnnualRevenue($B) Online Game Online Video Online Literature Livestreaming Digital Music Source: Game industry data per per Newzoo and Hillhouse estimates, excludes console or PC hardware related revenue. Online video data per iResearch (China) and Hillhouse estimates (USA), excludes advertising related revenue. Digital music data (excl. advertising) per iResearch (China) and RIAA (USA). Livestreaming (China) data per Hillhouse estimates. eBook data per Hillhouse estimates (China) and AAP and Hillhouse estimate (USA) $0 $10 $20 $30 $40 2011 2012 2013 2014 2015 2016 AnnualRevenue($B) Online / Console Game Online Video eBook Digital Music Online Entertainment User-Pay Revenue By Vertical, China, 2011-2016 Online Entertainment User-Pay Revenue By Vertical, USA, 2011-2016

KP INTERNET TRENDS 2017 | PAGE 206 $0 $5 $10 $15 $20 $25 $30 2012 2013 2014 2015 2016 2017E VideoGame(excl.Hardware)Revenue($B) China USA EMEA Asia (ex. China) Source: Newzoo * Excluding console / gaming PC hardware revenue. Global Interactive Game Revenue = China #1 Market in World* > USA (2016) Interactive Game Software Revenue by Region, Global, 2012 – 2017E

KP INTERNET TRENDS 2017 | PAGE 207 Tencent – Honor of Kings Mobile Multiplayer Online Battle Arena (MOBA) Leader… 50MM+ DAU, $3B+ Annualized Bookings Driven by Social + Simple UI + Constant Product Improvement NetEase Portfolio of Leading Mobile Massively Multiplayer Online Role Playing Games (MMORPGs)… Driven by Mobile First Mover Advantage + IP + Social Design + Quality Production China Online Gaming = Tencent + NetEase…Mobile MOBA + MMORPG Game Leaders… Source: Tencent, NetEase

KP INTERNET TRENDS 2017 | PAGE 208 $0 $500 $1,000 $1,500 $2,000 $2,500 $3,000 $3,500 $4,000 Revenue($MM) PC Mobile …China Online Gaming = Tencent + NetEase Driving Mobile Innovation + Revenue $0 $500 $1,000 $1,500 $2,000 $2,500 $3,000 $3,500 $4,000 Revenue($MM) PC Mobile Source: Tencent, NetEase, Goldman Sachs Investment Research Note: Assuming 1USD = 6.9RMB. NetEase Online Game Revenue, PC vs. Mobile, Q1:14-Q1:17 Tencent Online Game Revenue, PC vs. Mobile, Q1:14-Q1:17

KP INTERNET TRENDS 2017 | PAGE 209 Diverse Live Content Type Singing / Dancing / Talk Show / Game Play… Local / Social Nearby Livestreams / Chat & Add Friends China Livestreaming = High Consumer Engagement + Willingness to Pay… Interactive / Social / Gamified Like / Chat with Hosts & Audience / Buy Virtual Gifts to Support Performers 20+ Virtual Gift Categories priced from Rmb0.01 Source: Hillhouse research

KP INTERNET TRENDS 2017 | PAGE 210 $0.00 $0.20 $0.40 $0.60 Online Music Radio Online Video TV Online Games Livestreaming Source: Hillhouse estimates based on Newzoo, iResearch, Questmobile, and select company disclosures Note: Revenue data includes subscription, advertising and paid download revenue streams. …China Livestreaming = Compelling Monetization Estimated Revenue per Hour, China, 2016

KP INTERNET TRENDS 2017 | PAGE 211 China On-Demand Transportation = #1 Global Market… Cars + Bikes

KP INTERNET TRENDS 2017 | PAGE 212 China On-Demand Transportation (Cars + Bikes) = Global Leader @... ~67% Global Share (10B+ Annualized Trips, + >2x Y/Y) Source: Hillhouse Capital estimates, include on-demand taxi, private for-hire vehicles, as well as on-demand for- hire motorbike and bike trips booked through smartphone apps On-Demand Transportation Trip Volume by Region, Global, Q1:13 – Q1:17 0 1,000 2,000 3,000 4,000 1Q13 1Q14 1Q15 1Q16 1Q17 QuarterlyCompletedTrips(MM) ROW SE Asia India EMEA N. America China Bike China Car Q1:13 Q1:14 Q1:15 Q1:16 Q1:17

KP INTERNET TRENDS 2017 | PAGE 213 China On-Demand Bike Sharing = Mobile Innovation Driving Significant Usage Ramp In-Bike GPS + Smartphone Bike Sharing Without Stations…Location-Based Virtual Red Envelope Drives Utilization QR Code + Mobile Payment Easy Unlock & Low Friction Payment Ubiquity + Low Cost (¥1/~$0.15 per 30 min) + Convenience Mass Adoption & Bike Utilization Mobike Product Innovation Source: Mobike

KP INTERNET TRENDS 2017 | PAGE 214 0% 20% 40% 60% 80% 100% 120% 0 5 10 15 20 25 7/16 8/16 9/16 10/16 11/16 12/16 1/17 2/17 3/17 %M/MGrowth ChinaBikeSharingMAU(MM) China On-Demand Bike Sharing MAUs M/M Growth Source: TrustData Note: Dip in M/M growth rate in 1/17 was driven by Chinese New Year. China On-Demand Bike Sharing =… @ 20MM+ MAU…100%+ M/M Accelerating Growth China On-Demand Bike Sharing MAU, 7/16 – 3/17

KP INTERNET TRENDS 2017 | PAGE 215 China On-Demand Bike Sharing = High Frequency… 2/3 Users Ride 3+ Times Per Week Commute 50% Leisure / Exercise 29% Shopping 12% Business 3% School 1% Other 5% <1x 9% 1-2x 25% 3-5x 32% 6-10x 18% 10+ 16% Source: Transport Commission of Shenzhen Municipality study on bike sharing, based on operating data from four participating companies and survey data between 10/16 and 3/17, n=16,546 Highlights from Shenzhen Municipality On-Demand Bike Sharing Study, 5/17 On-Demand Bike Trips per Week Purpose of On-Demand Bike Trips

KP INTERNET TRENDS 2017 | PAGE 216 On-Demand Bike Sharing = Positive Environmental Impact + High Customer Satisfaction Source: Transport Commission of Shenzhen Municipality study on bike sharing, based on operating data from four participating companies and survey data between 10/16 and 3/17, n=16,546 * Based on following assumptions – 250k reduction in daily private car trips, avg. trip length of 10km, avg. fuel consumption of 6.9 L/100km, avg. CO2 emission of 2kg/L of fuel. Highlights from Shenzhen Municipality On-Demand Bike Sharing Study, 5/17 11MM Registered Users in Shenzhen, China 530K Available Bikes 2.6MM Daily Trips 5 Trips per Available Bike per Day 50% On-Demand Bike Trips Serving as Last-Mile Connection to Public Transit Trips 10% Bike Trips Replacing Private Car Driving Trips 100K+ Tons Reduction in Annual CO2 Emission* 95% Respondents Support Continued Development of Bike Sharing

KP INTERNET TRENDS 2017 | PAGE 217 China On-Demand Bike Sharing = Complements On-Demand Cars… @ 75% Shorter Trip Distance & 80% Lower Cost per Mile On-Demand Car Share (Didi) On-Demand Bike Share (Mobike / Ofo) Average Trip Distance 8 KM ~5 Miles 2 KM 1.2 Miles Average Trip Cost 20 RMB ~3 USD ~1 RMB ~0.15 USD Cost per Km ~2.50 RMB ~0.50 RMB Cost per Mile ~0.60 USD ~0.12 USD Source: On-demand car share data per Hillhouse estimate. On-demand bike share data per Transport Commission of Shenzhen Municipality study on bike sharing, based on operating data from four participating companies and survey data between 10/16 and 3/17, n=16,5466.92

KP INTERNET TRENDS 2017 | PAGE 218 China Mobile Payment Infrastructure = Enabling Rapid Growth + Monetization of Internet Usage

KP INTERNET TRENDS 2017 | PAGE 219 China Mobile Payment Volume = +2x Y/Y to $5T+ Led by AliPay + WeChat Pay Source: Analysys *Excludes certain P2P and transfer payments. Assume constant FX rate of 1USD = 6.9RMB. $0 $1,000 $2,000 $3,000 $4,000 $5,000 $6,000 2012 2013 2014 2015 2016 ChinaMobilePaymentVolume($B) AliPay 54% WeChat Pay 40% Others 7% China Mobile Payment Volume, 2012 - 2016 China Mobile Payment Market Share*, Q1:17

KP INTERNET TRENDS 2017 | PAGE 220 China Mobile Payments = Convenience vs. Cash & Bank Cards… Small Transactions Growing Especially Fast (<100RMB / $15) Source: PCAC, Bernstein Analysis Size of Mobile Payment Transactions, 2012 - 2016 Reasons for Using Mobile Payments, 2012 - 2016 0% 20% 40% 60% 80% 100% 2014 2015 2016 0% 20% 40% 60% 80% 100% 2014 2015 2016