RESEARCH REPORT NOTE: This document is a free excerpt of a larger research report. If you are interested in purchasing the full report, please contact Tractica at sales@tractica.com. EXECUTIVE SUMMARY Artificial Intelligence for Enterprise Applications Deep Learning, Predictive Computing, Image Recognition, Speech Recognition, and Other AI Technologies for Enterprise Markets: Global Market Analysis and Forecasts BRUCE DALEY Principal Analyst CLINT WHEELOCK Managing Director
SECTION 1 EXECUTIVE SUMMARY 1.1 ARTIFICIAL INTELLIGENCE HAS COME OF AGE After 60 years of false starts, the integration of artificial intelligence (AI) with probability and statistics has led to a marriage of machine learning, control theory, and neuroscience that is yielding practical benefits. This shared theoretical foundation, combined with the exponential growth of processing power and the unprecedented growth in the amount of data available to analyze, has made AI systems attractive for businesses to adopt. AI technologies are already being used in a number of different industries including online advertising services, automotive, agriculture, consumer finance, data storage, education, investment, healthcare, law, manufacturing, media, medical diagnostics, oil and gas, philanthropies, and retail. Systems modeled on the human brain such as deep learning are being applied to tasks as varied as question-answering medical diagnostic systems, credit scoring, program trading, fraud detection, product recommendations, image classification, speech recognition, language translation, and self-driving vehicles. Thanks to some notable successes, the only limit to the problems AI is being asked to solve seems to be the human imagination. Not all of these AI trials will prove successful, but executives in almost every industry should consider the potential impact AI will have on their businesses, business models, and bottom lines. This can be a challenge because AI is a loose term that covers a number of different technologies. Cognitive computing, machine learning, deep learning, predictive application programming interfaces (APIs), natural language processing, image recognition, and speech recognition are all under its umbrella. Which technologies are considered AI at any given point in time is fairly fluid. As new models of computing come out of academia, such as neuromorphic computing, they tend to be called AI. Over time, as AI technologies mature and become proven, they cease to be considered intelligent anymore, with autopilot systems in aviation being one notable example. During the time period considered in this study, it is highly likely that as some AI technologies become mainstream, they will no longer be considered AI at all and new ones will take their place. Although AI systems are relatively small in size compared with other parts of computing, AI technologies are the sharp point of the information technology spear. Because they are both compute and storage intensive, they require a disproportionate share of hardware resources and will drive significant sales of servers, networks, and storage. In addition, as AI applications become increasingly integral to critical business processes, the professional services required to support, implement, and maintain them will also increase disproportionately. Widespread adoption of AI will create strong commercial and financial incentives for the companies making the technology, many of which are now quite small, to create extensive business ecosystems around them. 1.2 MARKET DRIVERS Although AI technology has advanced significantly in the past few years, what is most advancing the need for AI technology is an explosion in the volume of available data. In 2013, IBM estimated that 90% of the data in the world had been created in the two preceding years. As large as this amount is, by some estimates, it will grow another 50 times before the year 2020. This volume of data creates the need for AI in two different ways. 1
The first need is to perform calculations and make decisions on large amounts of simple data, which human beings can do easily but do not have enough time to complete. The second is to identify complex patterns in very large data sets that no individual human mind could easily comprehend. AI also has the significant advantage over natural intelligence of being more consistent. Computers do fail, make mistakes, and need to be taken down for maintenance, but as a rule they are able to perform routine tasks more untiringly than human beings who need to sleep, get bored, and whose cognitive awareness waxes and wanes in regular cycles. 1.3 MARKET BARRIERS Many of the biggest names in science and technology have expressed reservations about AI. For example, Microsoft founder Bill Gates stated he was in the camp that is concerned about artificial intelligence, and predicted that AI could become strong enough to pose a risk to society. The intellectual resistance to AI is likely to grow into political resistance in the coming years and some pundits are already advocating the establishment of governmental departments to provide oversight for AI systems, although this is currently a low priority for most governments. Complicating the public debate is the question of how AI differs from types of automation that have gone before. If a machine can think like a person, does that make it a person? Many of the most important questions in AI are not technical at all but legal, political, and ultimately theological. A technical barrier to AI adoption is the absolute requirement most AI systems have for statistically valid, clean, accurate data. As with any information system, bad data will result in bad assumptions and predictions. Although some data, such as pictures of cats, are freely abundant on the internet, other data is much rarer, which could prove a barrier to adoption for AI systems that need to make decisions based on reliable data. Being software, AI applications are also subject to all the same limitations as other software systems including bugs, hacking, and poor design. 1.4 TECHNOLOGIES Among the many different approaches to AI, the following are considered in this report: Cognitive computing, which simulates human thought processes by using a computerized model to acquire knowledge and then make decisions. Instead of being programmable in the traditional sense, cognitive systems are designed to make their own inferences from data. Machine learning is a type of AI that involves using computerized mathematical algorithms that can learn from data and can depart from strictly following rule-based, pre-programmed logic. Machine learning algorithms build a probabilistic model and then use it to make assumptions and predictions about similar data sets. Deep learning is a form of machine learning that uses the model of human neural nets to make its predictions about new data sets. Tractica believes that it is currently the most promising of all AI technologies and is advancing other branches of the science including cognitive computing, image recognition, and speech recognition. Predictive application programming interfaces (APIs). An API formalizes access to software modules into standardized inputs and outputs, thus freeing the 2
programmer from having to understand the details of the operations of the module. A predictive API takes input in the form of data, and uses some form of AI to produce a predictive output. Natural language processing enables computers to understand human language as it is spoken and written and to produce human like speech and writing. Machine translation of one human language into another language is also a form of natural language processing. Image recognition attempts to identify pictures of objects that people can see. It can also include attempts to use the same technology to identify patterns in data, such as seismographic readings, that humans beings cannot see. It is seems likely that, sometime during the period covered by this report, pattern recognition will evolve to become its own category of AI. Speech recognition is the process of capturing spoken words and then converting them into binary data that can be used for natural language processing and other applications such as voice recognition. Voice recognition attempts to identify the person who is doing the speaking rather than focusing on what they are saying. 1.5 MARKET FORECAST Tractica forecasts that annual revenue for enterprise applications of AI will increase from $202.5 million worldwide in 2015 to $11.1 billion in 2024, representing a compound annual growth rate (CAGR) of 56.1%. Given both the data- and compute-intensive nature of the technology, AI systems will drive hardware sales at an even higher rate than the missioncritical systems have to date, especially graphics processing unit (GPU) chips. AI systems will also require a higher level of services. Tractica also forecasts significantly different levels of growth in certain industries as well as in different regions. Chart 1.1 Artificial Intelligence Revenue by Region, World Markets: 2015-2024 ($ Millions) $12,000 $10,000 $8,000 $6,000 $4,000 North America Western Europe Eastern Europe Asia Pacific Latin America Middle East Africa $2,000 $- 2015 2016 2017 2018 2019 2020 2021 2022 2023 2024 (Source: Tractica) 3
SECTION 9 TABLE OF CONTENTS SECTION 1... 1 Executive Summary... 1 1.1 Artificial Intelligence Has Come of Age... 1 1.2 Market Drivers... 1 1.3 Market Barriers... 2 1.4 Technologies... 2 1.5 Market Forecast... 3 SECTION 2... 4 Market Issues... 4 2.1 Is Artificial Intelligence a Threat to Humanity?... 4 2.1.1 Summoning the Demon... 4 2.1.2 The Ethical Dilemma... 4 2.1.3 Can a Computer Have a Soul?... 5 2.2 Why Are Cars that Drive Themselves Smart, but Planes that Fly Themselves Are Mindless?... 5 2.2.1 Smart Cars... 5 2.2.2 Mindless Planes... 6 2.2.3 Familiarity Breeds Contempt... 6 2.3 Key Industries... 6 2.3.1 Ad Service Technology... 6 2.3.2 Automotive... 7 2.3.3 Agriculture... 7 2.3.4 Consumer Finance... 8 2.3.5 Data Storage... 8 2.3.6 Education... 8 2.3.7 Investment... 8 2.3.8 Healthcare... 9 2.3.9 Legal... 9 2.3.10 Manufacturing... 9 2.3.11 Media... 9 2.3.12 Medical Diagnostics... 10 2.3.13 Oil and Gas... 10 2.3.14 Philanthropies... 10 2.3.15 Retail... 10 2.4 Market Barriers... 11 2.4.1 Societal Backlash... 11 2.4.2 Over-hyped and Over-sold... 11 2.4.3 Maintenance and Recovery... 11 2.4.4 The Resistance of Some Problems to Yield to Any Solution... 11 SECTION 3... 13 Technology Issues... 13 3.1 Old Bottle, New Wine... 13 3.1.1 Larger Data Sets... 13 3.1.2 Better Algorithms... 13 3.1.3 Cloud and Parallel Computing... 13 3.2 Cognitive Computing... 14 3.3 Deep Learning... 15 3.4 Machine Learning... 16 90
3.5 Predictive APIs... 17 3.6 Natural Language Processing... 18 3.7 Image Recognition... 18 3.8 Speech Recognition... 19 3.9 A Word About Other Senses... 20 3.10 Technologies Not Considered in This Report... 20 3.10.1 Avionics... 20 3.10.2 Robotics and Motion Control... 21 3.10.3 Governmental Applications... 21 3.11 Future Directions... 21 3.11.1 The Next Wave... 21 3.11.2 The Larger Impact... 22 SECTION 4... 23 Key Industry Players... 23 4.1 IBM... 23 4.2 Microsoft... 24 4.3 Facebook... 24 4.4 Google... 25 4.5 Baidu... 25 4.6 Rocket Fuel... 26 4.7 Dstillery... 27 4.8 The Climate Corporation... 27 4.9 Prism Skylabs... 28 4.10 Continental Corporation... 28 4.11 Tesla Motors... 29 4.12 Mobileye... 29 4.13 Lending Club... 30 4.14 Kabbage... 30 4.15 SwiftKey... 31 4.16 Coursera... 31 4.17 Knewton... 32 4.18 Bloomberg... 33 4.19 FinGenius... 33 4.20 Zephyr Technology... 34 4.21 Apple... 34 4.22 Intel... 35 4.23 NEC... 35 4.24 Qualcomm... 36 4.25 AlchemyAPI... 36 4.26 DeepMind... 37 4.27 Salesforce... 37 4.28 DataMinr... 38 4.29 Bina Technologies... 38 4.30 Cognitive Scale... 39 4.31 Declara... 39 SECTION 5... 44 Market Forecasts... 44 5.1 Forecast Methodology... 44 5.2 Global AI Market Forecasts... 44 5.3 Global Economic Growth Rates... 45 5.4 AI Revenue by Industry... 46 5.5 AI Revenue by Technology... 47 5.6 AI-Driven Revenue Streams... 48 91
5.6.1 AI-Driven Services... 48 5.6.1.1 AI-Driven Installation Services... 49 5.6.1.2 AI-Driven Training... 50 5.6.1.3 AI-Driven Customization Services... 51 5.6.1.4 AI-Driven Application Integration Services... 52 5.6.1.5 AI-Driven Maintenance and Support Services... 53 5.6.2 AI-Driven Hardware Revenue... 54 5.6.2.1 AI-Driven Cloud Revenue... 55 5.6.2.2 AI-Driven GPU Chip Revenue... 56 5.6.2.3 AI-Driven Compute Revenue... 57 5.6.2.4 AI-Driven Network Product Revenue... 58 5.6.2.5 AI-Driven Storage Device Revenue... 59 5.7 AI Revenues by Industry Segment... 60 5.7.1 AI in the Ad Service Industry... 60 5.7.2 AI in the Automotive Industry... 61 5.7.3 AI in the Agriculture Industry... 62 5.7.4 AI in the Consumer Finance Industry... 63 5.7.5 AI in the Data Storage Industry... 64 5.7.6 AI in the Education Industry... 65 5.7.7 AI in the Investment Industry... 66 5.7.8 AI in the Healthcare Industry... 67 5.7.9 AI in the Legal Industry... 68 5.7.10 AI in the Manufacturing Industry... 69 5.7.11 AI in the Media Industry... 70 5.7.12 AI in the Medical Diagnostics Industry... 71 5.7.13 AI in the Oil and Gas Industry... 72 5.7.14 AI in the Philanthropies Industry... 73 5.7.15 AI in the Retail Industry... 74 5.8 AI Revenue Forecasts by Technology... 75 5.8.1 Cognitive Computing... 76 5.8.2 Machine Learning... 77 5.8.3 Deep Learning... 78 5.8.4 Predictive APIs... 79 5.8.5 Natural Language Processing... 80 5.8.6 Image Recognition... 81 5.8.7 Speech Recognition... 82 5.8.8 Other AI Technologies... 83 SECTION 6... 84 Conclusion and Recommendations... 84 6.1 Key Findings... 84 6.2 Recommendations for Potential Customers... 84 6.3 Recommendations for Technology Vendors... 84 6.4 Conclusion... 85 SECTION 7... 86 Company Directory... 86 SECTION 8... 89 Acronym and Abbreviation List... 89 SECTION 9... 90 Table of Contents... 90 SECTION 10... 94 Table of Charts and Figures... 94 92
SECTION 11... 95 Scope of Study... 95 Sources and Methodology... 95 Notes... 96 93
SECTION 10 TABLE OF CHARTS AND FIGURES Chart 1.1 AI Revenue by Region, World Markets: 2015-2024... 3 Chart 5.1 AI Revenue by Region, World Markets: 2015-2024... 44 Chart 5.2 Estimated Economic Growth Rates by Region, World Markets: 2015-2024... 45 Chart 5.3 AI Revenue by Industry, World Markets: 2015... 46 Chart 5.4 AI Revenue by Technology, World Markets: 2015... 47 Chart 5.5 AI-Driven Services Revenue by Service Category, World Markets: 2015-2014... 48 Chart 5.6 Installation Services Revenue Driven by AI by Region, World Markets: 2015-2024... 49 Chart 5.7 AI-Driven Training Revenue by Region, World Markets: 2015-2024... 50 Chart 5.8 AI-Driven Customization Services Revenue by Region, World Markets: 2015-2024... 51 Chart 5.9 AI-Driven Application Integration Services Revenue by Region, World Markets: Chart 5.10 2015-2024... 52 AI-Driven Maintenance and Support Services Revenue by Region, World Markets: 2015-2024... 53 Chart 5.11 AI-Driven Hardware Revenue by Product Category, World Markets: 2015-2014... 54 Chart 5.12 AI-Driven Cloud Revenue by Region, World Markets: 2015-2024... 55 Chart 5.13 AI-Driven GPU Chip Revenue by Region, World Markets: 2015-2024... 56 Chart 5.14 AI-Driven Compute Revenue by Region, World Markets: 2015-2024... 57 Chart 5.15 AI-Driven Network Product Revenue by Region, World Markets: 2015-2024... 58 Chart 5.16 AI-Driven Storage Device Revenue by Region, World Markets: 2015-2024... 59 Chart 5.17 AI Revenue in the Ad Service Industry by Region, World Markets: 2015-2024... 60 Chart 5.18 AI Revenue in the Automotive Industry by Region, World Markets: 2015-2024... 61 Chart 5.19 AI Revenue in the Agriculture Industry by Region, World Markets: 2015-2024... 62 Chart 5.20 AI Revenue in the Consumer Finance Industry by Region, World Markets: 2015-2024... 63 Chart 5.21 AI Revenue in the Data Storage Industry by Region, World Markets: 2015-2024... 64 Chart 5.22 AI in the Education Industry by Region, World Markets: 2015-2024... 65 Chart 5.23 AI Revenue in the Investment Industry by Region, World Markets: 2015-2024... 66 Chart 5.24 AI Revenue in the Healthcare Industry by Region, World Markets: 2015-2024... 67 Chart 5.25 AI Revenue in the Legal Industry by Region, World Markets: 2015-2024... 68 Chart 5.26 AI Revenue in the Manufacturing Industry by Region, World Markets: 2015-2024... 69 Chart 5.27 AI Revenue in the Media Industry by Region, World Markets: 2015-2024... 70 Chart 5.28 AI Revenue in the Medical Diagnostics Industry by Region, World Markets: 2015-2024... 71 Chart 5.29 AI Revenue in the Oil and Gas Industry by Region, World Markets: 2015-2024... 72 Chart 5.30 AI Revenue in the Philanthropies Industry by Region, World Markets: 2015-2024... 73 Chart 5.31 AI Revenue in the Retail Industry by Region, World Markets: 2015-2024... 74 Chart 5.32 Cognitive Computing Revenue by Region, World Markets: 2015-2024... 76 Chart 5.33 Machine Learning Revenue by Region, World Markets: 2015-2024... 77 Chart 5.34 Deep Learning Revenue by Region, World Markets: 2015-2024... 78 Chart 5.35 Predictive API Revenue by Region, World Markets: 2015-2024... 79 Chart 5.36 Natural Language Processing Revenue by Region, World Markets: 2015-2024... 80 Chart 5.37 Image Recognition Revenue by Region, World Markets: 2015-2024... 81 Chart 5.38 Speech Recognition Revenue by Region, World Markets: 2015-2024... 82 Chart 5.39 Other AI Revenue by Region, World Markets: 2015-2024... 83 Chart 11.1 Tractica Research Methodology... 96 Table 4.1 Additional Industry Participants... 40 94
SECTION 11 SCOPE OF STUDY This report examines the practical application of artificial intelligence within commercial enterprises. The technologies covered include cognitive computing, deep learning, machine learning, predictive APIs, natural language processing, image recognition, and speech recognition. Other AI technologies that seem promising, but have not yet achieved full commercialization, such as algorithms that learn and adapt, spatial and contextual awareness, reasoning automation, semantic understanding, common sense, and emotional intelligence are covered in the other category. One of the challenges of creating forecasts in such a highly innovate field is that a technology not yet invented is likely to play a significant role during the time period (2015 to 2024) covered by this study. The report does not consider the use of AI in governmental, military, or intelligence applications. Within that scope, the report considers a number of different ways AI is being used across 16 different industry groups and identifies those applications best suited for commercial use. The report also considers the impact such technologies will have on the demand for services and hardware to support AI based applications. The accompanying databook forecasts revenues for AI applications across geographic region, technology, and industry during the period from 2015 through 2024. SOURCES AND METHODOLOGY Tractica is an independent market research firm that provides industry participants and stakeholders with an objective, unbiased view of market dynamics and business opportunities within its coverage areas. The firm s industry analysts are dedicated to presenting clear and actionable analysis to support business planning initiatives and go-to-market strategies, utilizing rigorous market research methodologies and without regard for technology hype or special interests, even including Tractica s own client relationships. Within its market analysis, Tractica strives to offer conclusions and recommendations that reflect the most likely path of industry development, even when those views may be contrarian. The basis of Tractica s analysis is primary research collected from a variety of sources, including industry interviews, vendor briefings, product demonstrations, and quantitative and qualitative market research focused on consumer and business end-users. Industry analysts conduct interviews with representative groups of executives, technology practitioners, sales and marketing professionals, industry association personnel, government representatives, investors, consultants, and other industry stakeholders. Analysts are diligent in pursuing interviews with representatives from every part of the value chain in an effort to gain a comprehensive view of current market activity and future plans. Within the firm s surveys and focus groups, respondent samples are carefully selected to ensure that they provide the most accurate possible view of demand dynamics within consumer and business markets, utilizing balanced and representative samples where appropriate and careful screening and qualification criteria in cases where the research topic requires a more targeted group of respondents. Tractica s primary research is supplemented by the review and analysis of all secondary information available on the topic being studied, including company news and financial information, technology specifications, product attributes, government and economic data, industry reports and databases from third-party sources, case studies, and reference customers. As applicable, all secondary research sources are appropriately cited within the firm s publications. All of Tractica s research reports and other publications are carefully reviewed and scrutinized by the firm s 95
senior management team, in an effort to ensure that research methodology is sound, all information provided is accurate, analyst assumptions are carefully documented, and conclusions are well-supported by facts. Tractica is highly responsive to feedback from industry participants and, in the event errors in the firm s research are identified and verified, such errors are corrected promptly. Chart 11.1 Tractica Research Methodology MARKET RESEARCH SUPPLY SIDE DEMAND SIDE PRIMARY RESEARCH Industry Interviews Vendor Briefings Product Evaluations End-User Surveys End-User Focus Groups SECONDARY RESEARCH Company News & Financials Technology & Product Specs Government & Economic Data Case Studies Reference Customers MARKET ANALYSIS QUALITATIVE ANALYSIS Company Analysis Business Models Competitive Landscape Technology Assessment Applications & Use Cases QUANTITATIVE ANALYSIS Market Sizing Market Segmentation Market Forecasts Market Share Analysis Scenario Analysis (Source: Tractica) NOTES CAGR refers to compound annual growth rate, using the formula: CAGR = (End Year Value Start Year Value) (1/steps) 1. CAGRs presented in the tables are for the entire timeframe in the title. Where data for fewer years are given, the CAGR is for the range presented. Where relevant, CAGRs for shorter timeframes may be given as well. Figures are based on the best estimates available at the time of calculation. Annual revenues, shipments, and sales are based on end-of-year figures unless otherwise noted. All values are expressed in year 2015 U.S. dollars unless otherwise noted. Percentages may not add up to 100 due to rounding. 96
Published 2Q 2015 2015 Tractica LLC 1111 Pearl Street, Suite 201 Boulder, CO 80302 USA Tel: +1.303.248.3000 Email: info@tractica.com www.tractica.com This publication is provided by Tractica LLC ( Tractica ). This publication may be used only as expressly permitted by license from Tractica and may not otherwise be reproduced, recorded, photocopied, distributed, displayed, modified, extracted, accessed or used without the express written permission of Tractica. Notwithstanding the foregoing, Tractica makes no claim to any Government data and other data obtained from public sources found in this publication (whether or not the owners of such data are noted in this publication). If you do not have a license from Tractica covering this publication, please refrain from accessing or using this publication. Please contact Tractica to obtain a license to this publication. 97