www.hg80229.com Here is a copy of Intelligent Internet Mapping in the future. Please check it out

Here is a copy of Intelligent Internet Mapping in the future. Please check it out

Public number: Hangxing certification Source: Time: 2020-02-21 02:00:32

Millennium Tongzhou vitality north stream

Conquest of alien cute wife boss

[Write in front]

Five, six, seven, eight, ninety little air vents came again.

In this era where some people go to the mall and sing k, some people brush their presence in mobile games, and investors invest in researching how to jump jump, and entrepreneurs dig deeper how to let pigs out of the market ... As a VC with a "deep" degree We, again, chose to study an extremely indifferent "bottom thinking" topic:

"Smart Internet"

Yes, don't doubt it, it's so cold ...

In the past two months, the Jiuhe investment team has been launched, and AI entrepreneurs, mentors, professors, investors from various fields ... and even the grandpa of Xiupai (yes! Grandpa now earns a lot ) Have done in-depth exchanges, and we have also almost searched all the Chinese foreign languages for thirty days ago and three days ago, just for the one you are about to see:

[Wohe's first external research report]

I was walking on the road wearing skate shoes that day. When I heard that an uncle at a fruit stand was talking to another uncle with a big fork, and he did n’t have to collect fake coins to find change, I knew that this smart The Internet may not chase anymore ... After all, even my 90-year-old grandfather is playing chess on the iPad, and even people like me with Lin Zhiling's navigation are no longer foolish. Everyone who sees the black mirror and the western world thinks You may be a bionic person and you do n’t need to care about the meaning of life ...

Looking at the graph that people predicted now 100 years ago, it is still a little scary.

People imagine 100 years ago now ... this is a WeChat video? Still live? !!

-What are the reasons behind the various Internet and messy air outlets that subvert our lives?

-Can the popular mining data and deadlock algorithm produce the next Jack Ma?

-Will the rate of return of the next generation industry exceed the house price? !!

If you are interested in the above, maybe you can find the answer you want in the following indifferent report ... At the same time, we also welcome to join our Jiuhe AI investment research team in the background to join our discussion.

Here is the full report:

"Entering the era of intelligent Internet-from connection to data, from traffic economy to efficiency economy"


Rest assured, Xiaobian is serious

Guide to this issue

In 1990, Tim Berners-Lee started the World Wide Web (WWW) program at the European Quantum Physics Laboratory. Since the birth of the World Wide Web, from the PC to the smartphone terminal, the "connection" of the Internet has gradually created a business society with freely flowing information, and has quietly carried on the next round of technological changes.

等风口概念依然此起彼伏,不少创业者与同行疲于贴上趋势的标签。 In the past year known as the "Capital Winter", Alpha Go ’s victory over Li Shishi pushed the rapid and automated automation and intelligent transformation to the apex of public attention: “Artificial Intelligence”, “Internet of Things” Concepts such as "Internet +", "Industry 4.0" and so on are still one after another, and many entrepreneurs and their peers are tired of labeling trends.

Jiuhe has never been lazy to think from the bottom, but the nature and commonality of this round of changes in social production methods. Taking the opportunity of the 9th Anniversary, we will refine our thinking into the concept of "smart Internet" and discuss with you:

(1) What is the intelligent Internet?

The intelligent Internet is an evolutionary form in which the "connection" of the Internet has developed to a certain mature stage. It is based on the interconnection, and uses the "data" generated by the connection as the core. Through the data processing, it outputs "intelligent" value, which guides the efficiency of people and machines Decision-making and resource utilization.

智能互联网的五个核心特征 1.1 Five Core Features of Smart Internet

(1) The intelligent Internet is the evolutionary form of the Internet. Its primary manifestation is the development of the core value driving force of the Internet from "connection" to "data". "Connection" is the foundation of intelligent Internet, and "data" is the core of intelligent Internet to create value.

Evolution from the traditional Internet to the intelligent Internet

In the traditional Internet era, the characteristics of interconnection have gradually overcome information asymmetry, eliminated time and space barriers, and achieved value creation and business transformation with "connection" as the core. During this period, people were the absolute active players in connection actions, and goods existed only as a sign of human ownership: the connection between people and things in e-commerce, "things" are goods that sell ownership; "things" and "things" in the sharing economy The exchange is also the ownership of the exchange of items and the right to use.

In the era of the intelligent Internet, the change in the social production mode affected by "connection" has reached a certain stage of maturity. The "data" accumulated on this basis has played a core role, bringing about a new round of efficiency change and social production mode change. The connection between things and things is not a share of imprinted ownership, but also contains the automation value of "Internet of Everything." As the core driving force of "data" value creation, the machines with data value in the intelligent Internet have also been given meaning as independent data bodies, parallel to those who have absolute active control in the traditional Internet era.

(2) The carriers of the traditional Internet are relatively uniform and the migration of carriers has a greater impact on the original business landscape VS.

Intelligent Internet carriers are scattered and will appear increasingly discrete.

In the traditional Internet with "connection" as the core , "people" is the absolute active party of the "connection" action. The user interacts with information, goods and other individuals through a certain carrier. A unified carrier is more likely to form a unified interaction rule and The use scene, therefore, also corresponds to the convergence effect of using the carrier.

From "PC Internet" to "Mobile Internet", the carrier has a relatively uniform monopoly in a certain period of time. The consumption data of the carrier (eg. Smartphones, PC computer shipments) is an important data to judge the industry maturity curve, and the carrier ’s Changes bring new user shuffles and business changes.

"Data" as the core value of the driving force of the intelligent Internet, both humans and machines exist in parallel as independent data bodies, emphasizing the intelligent value of data output rather than behavioral interaction based on a unified carrier. Therefore, the smart Internet carrier is scattered and will penetrate with it. To the various scenes of production and life and showing more and more discrete characteristics:

  • The collection of data can come from a variety of channels such as the Internet of Things, web crawlers, and telecom operator data;

  • ,也可能是拥有核心算法SaaS 软件; Applications that generate intelligent value through data may be smartphone applications such as Jiuhe ’s outstanding investment project Caiyun Weather , or it may have SaaS software with core algorithms;

  • Users can choose Siri on the mobile phone to give instructions, or use the smart home / home robot to assume the role of virtual assistant;

  • Enterprises can embed algorithm software in production machines or use mobile control


(3) The core value driving force of the intelligent Internet-"data", includes both horizontal data acting on industry and industry, as well as vertical data in a specific direction; its essence is beyond the range that can be processed by the human brain, Data that require algorithms to play intelligent value.

Big data requires new processing models in order to have stronger decision-making, insight and process optimization capabilities to adapt to massive, high growth rates and diverse information assets.

——Gartner's definition of "big data"

In recent years, many opinions have also put forward the term "small data":

"Small data" refers to a data set with a specific attribute tendency, such as:

Data set per person / family (eg. All behavior data of someone at each moment)

A data set that points to a specific issue (eg. The temperature of the water tank at each moment of the wind turbine)

The "big data" and "small data" here are not limited to the "big" and "small" literally. "Big" itself is a relative concept; more precisely, it is a different way of applying data from horizontal to vertical:

The "big data" methodology is to extract common features or common problems from massive multi-dimensional data;

"Small data" is the accumulation of vertical / time data to create intelligent value.

But its essence is all big data that is beyond the range that human brain can process and requires algorithms to exert intelligent value.

As the American historian of science and technology Melvin Kranzberg's famous six laws of science and technology, the view of the third law: Technology comes in packages, big and small.

认为,智能互联网有其复杂综合的运作机制,而作为智能互联网核心特征的「数据」,同样拥有多元的应用方式和应用场景,而非简单称谓的“大数据”。 Jiuhe believes that the intelligent Internet has its complex and comprehensive operating mechanism, and the "data", which is the core feature of the intelligent Internet, also has multiple application methods and application scenarios, rather than simply "big data".

(4) The main manifestations of the intelligent Internet value chain are: data output decisions, decision-making to increase efficiency, and efficiency affects resource utilization and time costs. The "smart value" of the intelligent Internet includes the intelligent value of individual life and the intelligent value of social productivity.

(5) Any form that "generates value through connection and data" belongs to the category of intelligent Internet.

All products / services that include data decision characteristics or serve the data decision process can be considered as a broad category of intelligent Internet.

Artificial intelligence is an important expression of the intelligent value of the intelligent Internet; deep learning and machine learning are the core technical tools of the intelligent Internet; the Internet of Things is the infrastructure of the intelligent Internet data layer.

Jiuhe The concept of "smart internet" was chosen to express the value and awe of the integrity, tolerance and operational order of the intelligent future. The concept does not refer to a single argument. It is to better understand the underlying operating mechanism of technological change, to grasp the core breakthrough points, and to promote the positive development of social productivity in our role.

智能互联网的四个基础层面 1.2 Four Basic Levels of Smart Internet

The intelligent Internet carrier is discrete, but has a relatively constant value production chain, that is, the input and output paradigm from data to intelligent value. The progressive relationship between the four core links in the chain constitutes the four basic aspects of the intelligent Internet: data layer (data source and data collection)-infrastructure (data storage)-technical facilities (data conversion intelligent value)-consumer layer (intelligent value) Conversion of business / social value).

1.2.1 Data layer

Data is the underlying raw material of the intelligent Internet. It needs to go through the two steps of locating the data source and data collection to convert it into tangible data assets. Data sources can be broadly divided into three categories: personal / family data, corporate data, and public / government data.

  • Personal / family data

It mainly includes identity data, behavior data and privacy data.

Identity data: refers to biometric data unique to individuals such as facial features, fingerprints, genes, etc. Identity data is likely to develop into the "Personal ID" in the era of the intelligent Internet, which is also an important part of government citizen file data.

Behavior data: Refers to the collectable behavior trajectories left by individuals / families in life.

On the one hand, it includes behavioral data such as registration, consumption, browsing, and content production left by individuals as users on shopping websites, social media and other goods or information consumption platforms. This part of the data currently has no clear legal boundaries in China. Data also constitutes an important part of the data assets of the enterprise and operator; on the other hand, it is dependent on the development and popularization of smart homes and the Internet of Things, and individual living data accumulated in interaction with smart home devices; vertical accumulation over time And data feeding, personal / home smart devices will be increasingly customized and smart.

Privacy data: At present, there is still a lack of clear legal boundaries and privacy awareness, and it is also a potential blue ocean market. With the development and popularization of smart homes, large-scale commercialized individual / family private clouds may emerge in the future.

  • Enterprise data

It is divided into user data of the enterprise, data assets obtained from public data sets or data vendors, and enterprise data creation needs, as well as private data within the enterprise. Enterprise private data includes asset data such as enterprise equipment and knowledge patents; it also includes data during cash flow and supply chain operations. The processed and structured data set of an enterprise is also an important data asset.

  • Public / Government data

Sources are diverse and have important group values, including citizen / legal data such as identity, politics, law, taxation, city data such as transportation, public facilities, and environmental data. The modest opening of public data plays an important role in promoting the country's breakthrough in data technology and the promotion of smart cities.

Access to data needs to be accomplished through data collection. The improvement of "connected" facilities such as the Internet of Things and sensors is the basic guarantee for rich data sources. In recent years, physical sensors based on the principles of optics, pressure, electromagnetic fields, and sensors based on chemical reactions have become better and better in terms of integration, sensitivity, and cost. According to the US market research company Gartner, by 2020, there will be 2.5 billion devices appear connected to the Internet of Things; IDC predicts that annual data generated globally in 2020 is expected to reach 44 zettabytes. Personal consumption data scattered across application platforms and operators has not yet formed clear data privacy legislation and enforcement boundaries in China, but from another perspective, this status quo also provides China's deep learning with respect to other countries More fertile data soil. For offline data that is difficult to obtain, some companies also purchase data sets through data crowdsourcing or middlemen. According to the U.S. Senate report, the total data intermediary market in the United States in 2012 reached $ 150 billion. China has become the second largest data country after the United States, and it is expected that it will account for 21% of global data by 2020. However, among many enterprises with data source attributes, less than 8% have launched data lease business, and data vendors and dealers still have a large market space.

1.2.2 Infrastructure

Chips and cloud services are the infrastructure for source data storage and computing. The rapid development of deep learning in recent years is closely related to the popularity of cloud computing, the improvement of chip computing performance, and cost reduction. Traditional CPU processors implement serial instructions, and even modern improved multi-core CPUs target instruction set parallelism and task parallelism with limited computation speed; GPUs optimized for graphics processing and large-scale parallel operations will perform large-scale operations. The speed has been increased ten or even hundreds of times, making it possible for computers to quickly process massive amounts of data. On the other hand, the price and cost of GPUs have also been greatly reduced. Nvidia GPU (GTX 1080) has 9 TFLOPS performance for only $ 700, which means only 8 cents per GFLOPS, and in 1961, enough IBM 1620s needed more than 9 trillion per 1 GFLOPS. At the same time, the special chip technology dedicated to accelerating artificial intelligence is constantly being introduced. In addition to traditional manufacturers, large companies also actively reserve chip research and development power through self-development or mergers and acquisitions.

1.2.3 Technical facilities

Technical facilities refer to the technical aspects of processing and converting source data into structured data or intelligent results with intelligent value. 1)生态— 开源框架。 The technical facilities of the intelligent Internet mainly include the following three levels: 1) Ecology-open source framework. Tensorflow, Torch, MxNet and other deep learning frameworks.

2) Call — General API Voice, natural language, image processing and other functional modular APIs or general algorithm APIs.

3) Assets—Algorithm Model The proprietary algorithm model of the enterprise / developer transforms the important link of intelligent value and unique competitiveness.

1.2.4 Consumer layer

There are many types of business around the smart Internet ecosystem, including chip vendors at the infrastructure level, cloud service providers, and data transaction platforms at the data level. However, the "consumer layer" here refers to the business model where the core of the transaction comes from the intelligent value after data processing. Currently, there are mainly three forms: 1) General Function Module Trading

Sales of functional APIs such as speech, natural language, and image processing.

2) Algorithmic economy

The "algorithm economy" was proposed by Intel at IDF 2015 (Intel Developers Forum). When data becomes an asset that society, enterprises, and individuals cannot ignore, algorithms are the carrier of the valuation and exchange of these data assets, which will promote Data becomes the new currency-the currency of the digital world. That is, a business model with algorithms as its core competitiveness and barriers. The algorithm economy mainly comes in the form of software, including BI (Business Intelligence), customized algorithms serving individual / family quantified self (quantifiedl self), and proprietary algorithm providers. From the logic of the output results, it is difficult for the algorithm to form a monopoly, because unlike traditional commercial complex payment measurement factors, the 1% improvement in the effect of the algorithm economy has significant advantages in substitution; and from the support behind the algorithm—the data source Seeing, the data of the feeding algorithm is continuously accumulated and extended, and it is easy to form a data source barrier in a specific or vertical field, thereby establishing a data source monopoly advantage. 3) Smart upgrade application

Smart upgrade applications refer to products and applications that are consumer-oriented and have products in the corresponding scenarios, but iteratively innovate them through smart Internet technology. Smart upgrade applications include smart upgrade applications on the mobile side and comprehensive smart upgrade applications in other scenarios. For example, today's headline is an upgrade of the original news reading client through a recommendation engine and machine learning technology. Prisma uses an intelligent picture paradigm to upgrade traditional P-picture software; autonomous driving is a comprehensive intelligent upgrade of travel scenarios. With the further improvement of open source platforms and the accumulation of callable APIs, smart upgrade applications will first usher in a wave of development on the mobile Internet. In addition to technical capabilities, the ability to sharply capture scenarios and needs will play a more important role. .

智能互联网的技术成熟周期 1.3 Technology Maturity Cycle of Smart Internet

The following figures are Gartner's latest 2016 emerging technology maturity curve and 2016 IoT maturity curve.

Among them, Jiuhe focuses on the technologies that can reach the mature production stage within 2-5 years as follows:

物联网超过8 项技术有望在2 - 5 年内到达生产成熟期,未来5 年内物联网技术将快速发展,为智能互联网联打下坚实的「连接」基础,采集到更多之前不可触达的线下数据和产业数据,辅助机器学习的突破。 More than 8 technologies of the Internet of Things are expected to reach the mature stage of production in 2 to 5 years. In the next 5 years, the Internet of Things technology will develop rapidly, laying a solid "connection" foundation for the intelligent Internet connection, and collecting more previously inaccessible lines. Data and industry data to assist in machine learning breakthroughs. 正在走向泡沫破灭低谷期的物联网集成技术,能促进企业认识到物联网集成不只是简单的M2M(Machine to Machine)一体化,在此之前要做好底层互操作性和硬件集成工作。 —The IoT integration technology that is heading for the trough of the bubble burst can promote enterprises to realize that IoT integration is not just a simple M2M (Machine to Machine) integration. Prior to this, the underlying interoperability and hardware integration must be done well. I T / OT 集成撩拨着供应链和生产链实时打通的巨大想象,但回望更底层的物联网集成正经历低谷期,IT / OT 的集成仍必先实现物联充足的数据采集布点。 I T / OT integration evokes the huge imagination that the supply chain and production chain are connected in real time, but looking back at the lower level of IoT integration is going through a trough period, IT / OT integration must still achieve sufficient data collection and distribution of the IoT. 预测性分析与IT/OT 集成技术正处于同一阶段,事实上,IT/OT 集成的想象力直接能转化为丰富实时的数据源,喂养预测模型,两者呈正相关。 —Predictive analysis and IT / OT integration technology are at the same stage. In fact, the imagination of IT / OT integration can be directly transformed into a rich real-time data source and feed predictive models. The two are positively correlated. 与智能制造、工业大数据造密切相关的物联网集成、IT/OT集成、预测性分析、机器学习等多项技术同时处于2 - 5 年内到达成熟区的临界点,可以预期未来五年内工业领域将快速发展突破。 —Many technologies such as IoT integration, IT / OT integration, predictive analysis, and machine learning, which are closely related to intelligent manufacturing and industrial big data technology, are at the critical point of reaching mature areas within 2 to 5 years. It is expected that the industry will The field will develop rapidly. 自然语言问答技术行走于泡沫低谷期, 此前bots 疯狂的投资热将逐渐恢复理性,企业开始思考将技术应用于实际产生价值的场景而非想象场景里试错。 ——Natural language question answering technology walks in the trough of the bubble. Prior to this, the crazy investment fever of bots will gradually restore its rationality. Enterprises have begun to think about applying technology to scenarios that actually generate value rather than trial and error in imaginary scenarios. In other words, in the past two years, C-side bots and other entrepreneurs will gradually cool or switch to B-side application scenarios.

League of legends ipl5

(2) Business Paradigm in the Intelligent Internet Age:

From flow economy to efficiency economy

Dr. Sun Xiyou in 2003, "Flow Economy", defined "flow economy" as "the various types of economic existence and development in the economic field that rely on the flow of factors or production to bring economic benefits and development." In the Internet context, traffic is the number of customers / users (earlier PC Internet browsing / clicks). The evolution from the traditional Internet to the intelligent Internet. The "connection" as the core value driver of the traditional Internet focuses on the action itself (that is, whether it is used), which corresponds to the traffic economy based on the number of users / the economic value of a single user; and The intelligent Internet with "data" as its core driving force focuses on the output (ie, the effect of use), which corresponds to the efficiency and economy of cost savings and decision-making efficiency. After a complete explanation of the concept, characteristics and core components of the intelligent Internet, Jiuhe will analyze the smart Internet from two perspectives: the development trend, core characteristics of the ToB and ToC markets in the era of the intelligent Internet, and industry changes under the transformation of the intelligent Internet. Business social formation, exploring future market trends and evolution laws.

智能互联网的市场发展趋势 2.1 Market Development Trend of Smart Internet

2.1.1 ToB: From management costs to decision costs, from customer assets to data assets

In the era of mobile Internet in the pursuit of traffic economy, due to the network effect of C-side and the characteristics of rapid capacity lifting, mobile smartphone applications first exploded in the C-side field. After the C-side application cultivated user habits and accumulated a certain user base, mobile SaaS applications are only beginning to rise. However, for the efficiency economy, the improvement of efficiency directly affects the increase in revenue and efficiency figures. The overall development of ToB companies in the era of intelligent Internet is ahead of C-end companies.

For enterprise services in the era of the intelligent Internet, titles such as "AI-aaS" (AI as a Service), "DasS" (Data as a Service), and "Algorithm Economy" are popular in the industry. Jiuhe believes that the core characteristics of the evolution from the traditional Internet to the intelligent Internet are: 1) The core role of enterprise services for enterprises has evolved from reducing management costs to reducing decision costs. Traditional enterprise service software reduces the labor and resource costs required for information and personnel management through visualization and auxiliary management of workflow and information flow. In the value chain of data-driven decision-making, data can be huge, multi-dimensional variables that exceed the computable range of the human brain; decisions made by machine processing output can achieve real-time feedback. In other words, the reduction of decision-making costs in the era of the intelligent Internet refers not only to the complex problems that machines can replace humans to make decisions beyond the scope of the human brain, but also to the real-time nature of the decision cycle. The cost reduction and real-time nature of complex decisions make it possible to provide a dynamic supply chain from downstream to upstream, and also promote the development of technologies such as autonomous driving intelligence that have high requirements for real-time feedback and massive data processing. 2) Customers are not just revenue, they are core data assets.

In the past, the core indicators of a business services company were the number of customers and the corresponding payments / revenues. In the era of the intelligent Internet, customers are no longer simply a sign of revenue, and behind them are more valuable data assets. Data is the basis of the algorithm, and the accumulation of data assets can help companies continuously extend new markets and business opportunities to create new profit channels. At the same time, sufficient or unique data sets are also important conditions for the algorithm model to stand out.

Enterprise services can be observed from two perspectives: one is functional enterprise service software, that is, customers are not limited to industries, and focus on solving pain points in a certain link of the enterprise production chain, such as customer service software, human resource management software, recruitment applications, etc. One type is industry-based enterprise service applications, which are rooted in providing services in an industry. The data accumulated by these two types of services are horizontal data across industries and vertical data focusing on industries. IBM and Salesforce, which started with CRM, respond to the acquisition strategy of the intelligent Internet, which respectively reflect two data development ideas: IBM is rooted in the medical industry to accumulate data to feed Watson, and Salesfore actively acquires tool-type artificial intelligence to tap the value of its cross-industry data. On the other hand, due to the core value of data, industry giants with rich data sources or companies with unique data sets may enter the field of enterprise services with data resources.

In addition, due to the importance of data assets, security and stability requirements will be more urgent than traditional enterprise services. The security market is an important market component of smart Internet toB applications. Gartner pointed out that if the IoT security can reach 7-8% of the annual security investment in the IT industry, the global IoT security scale will reach 91 billion to 104 billion US dollars in 2019, which is almost the same as the entire traditional network security market size; according to CEDA According to research, the scale of China's Internet of Things industry is expected to reach RMB 930 billion in 2016, and it will be close to RMB 183 billion by 2020. Based on a conservative ratio of 1%, China's Internet of Things security will reach RMB 9.3 billion in 2016 and RMB 18.3 billion in 2020. yuan.

九合根据App Annie 的数据,整理了从2010 年到2015 年间中国区App Store 和Google Play 下载量爆发的应用和国内智能手机出货量数据的对比关系: 2.1.2 To C: The degree of application of personal / family data assets determines the development stage of the market . According to data from App Annie, Jiu He has compiled apps and domestic smartphones that have experienced downloads in the App Store and Google Play in China from 2010 to 2015. Comparison of shipment data:

1) The development of mobile Internet has been along the horizontal axis of time, and has experienced several waves of development of tools — social — business transaction platforms (including e-commerce, O2O) — entertainment consumption.

2) The characteristics of the mobile Internet market development stage are positively related to smartphone shipments:

In the early days, there were not many smartphone users, and a number of basic tools such as input methods, weather, and billing emerged to help users use smartphones better. When the use of mobile phones reached a base that could form a certain network effect, social applications emerged and Drive more people to accept smart phones; social and tool applications have cultivated good mobile usage habits, and mobile phone users have continued to grow near their peak. Based on a sufficient number of users, a new wave of transactions under platform migration opportunities has begun to emerge Opportunities for the platform; when mobile phone shipments started to grow smoothly, that is, when the mobile Internet reached maturity and saturation, emerging products focused on further expanding the user's possible stay time, cultural and entertainment applications flourished, and various applications became more and more detailed Divided into vertical and sinking trends. The core factors surrounding the development curve and stage characteristics of different types of mobile Internet applications are users / traffic: the "traffic" (even usage) of smartphones affects the development stages of different types of applications in mobile Internet; the traffic of a single application (ie users Number) and economic value determine its market share. From the flow economy to the efficiency economy of the intelligent Internet, data is the core asset and value driver. Regarding the market development rules and stage characteristics of the smart Internet C-end market, Jiuhe believes that the application level of individual / family data assets will play a key role. The smart Internet C-end market will go through the following four phases:

Mobile smart upgrade application (using public data sets and some individual behavioral data to upgrade the scene experience) — Smart home / family (individual / family data assets really play a role in smart life) — Smart consumption (upstream and downstream data of the industry chain and personal consumption data in real time Interaction) — Personal / family data platform and security application (formation of personal / family data asset protection awareness).

  • Smart upgrade application for mobile terminal

The smart upgrade application on the mobile terminal is built on the basis that the mobile Internet has been widely accepted and used. The most important thing for users is that the intelligent value realized by the technology is compared with the efficiency of existing applications. The accuracy of the output results and the user ’s accuracy Sex / efficiency tolerance is an important factor affecting its market acceptance.

1)移动智能应用最先获得市场广泛接受的将是文化娱乐、信息消费等拥有趣味性,用户对的准确度和效率不是特别看重的工具类应用,比如prisma。 Jiuhe believes that the market development of mobile smart upgrade applications will go through the following sequence: 1) The first mobile smart applications to be widely accepted by the market will be cultural entertainment, information consumption, etc., which are interesting, and the accuracy and efficiency of users are not Special emphasis on tool applications, such as Prisma. 2) It has long-tail value, but the traditional business model cannot balance the conflicting needs of human costs and scale, and can find corresponding solutions and obtain business value in the era of intelligent Internet. This type of application usually focuses on a field or solves a problem, and there is no killer App competition. Therefore, in the face of the desire for "solutions", users usually reserve a certain period of time for the maturity of accuracy of results. For example, car purchase consulting is a requirement for many novice car buyers. Many car new media entrepreneurs have also tried paid consulting models. However, with the increase in the number of users and consultations, revenues cannot basically cover human costs. NLP bot can) and has the potential to accumulate core data assets in vertical fields. 3) Comprehensive personal assistant application services that serve high-frequency needs such as taxi rides and takeaways have extremely low user tolerance, and the technology is still in progress, and there is still a relatively long distance from widespread market acceptance.

  • Smart home, home

1) Smart home / family is a data ecosystem consisting of hardware products (home robots, digital entertainment equipment, security monitoring, etc.), software systems, and cloud platforms based on the living environment, and personal / family behavior data in life And health data will accumulate vertically in the data ecosystem over time to feed customized intelligence based on individuals / families, and intelligent hardware used to direct the surrounding individuals to assist individuals and family members to live healthier lives. Intelligent hardware is not the core of smart home / home, the core is based on the data intelligence collected and cultivated by the home Internet of Things. The intelligent life in the ideal state can realize the cycle in which personal / family data assets truly act on the individual / family life, play the accumulation and training of personal / family data in the vertical dimension of time, and tap the intelligence of time and machine “customization” value. 2) The realization of smart home / home depends on the development of chips and the underlying technology of the Internet of Things, as well as the rise of solution providers and the construction of home data centers. In the long run, the development of smart homes / families needs to experience "local intelligence (some smart appliances such as cleaning robots bring convenience)-part of the connection (part of the connection of life data and two-way action on smart appliances to help intelligently serve individuals and families. Member life) — Global intelligence based on personal / home data centers (home automation, intelligence based on personal / home data centers) is still in the local intelligence stage. 3) China ’s smart home market has a low penetration rate and huge market space.

  • Smart consumption

1) In the era of the intelligent Internet, there are new consumption scenarios and experiences. Unmanned stores, machine shopping guides, smart dressing mirrors to select the clothes to be tested based on gesture operations, VR / AR-based immersive shopping, etc. The platform will become increasingly blurred and move towards online and offline integration. 2) The intelligent Internet era enables customized shopping, which is based on a smart factory and dynamic zero inventory supply chain that can meet customized needs and dynamic performance. 3) The ID of offline shopping transactions may gradually be replaced by identity data (faces / fingerprints / voice prints / genes ..), and offline ID innovation will push back the ID of online trading platforms.

  • Personal data platforms and security applications

1) When the smart home / family develops to a certain degree, it will promote the market's emphasis on personal data assets, and personal / family data platform products will appear, which exist in private clouds or other forms. 2) Just as the security products of the PC Internet era set off by the Melissa virus, the security products that protect personal data assets will also be driven by an iconic data security / privacy time. 3) The personal data platform and security product development design on the C side lag far behind the B side.

2.2 Industry Innovation in the Intelligent Internet Era

The efficiency and economy of the intelligent Internet era penetrates all walks of life, reflecting the obvious cost reduction and efficiency improvement that can be directly converted into economic value. Goldman Sachs has analyzed from the perspective of statistical economics that the United States in the mobile Internet era did not reflect a significant increase in social productivity on economic data, but the social productivity promotion effect brought by artificial intelligence has presented quantifiable economic benefits.

Jiuhe will analyze the innovations and opportunities brought by the intelligent Internet from the five basic industries of industry, agriculture, medical care, finance and retail.

  • industry

Industry accounts for the core proportion of GDP in most countries. The previous industrial revolutions have also promoted the evolution of human socio-economic and organizational forms. Industry in the era of the intelligent Internet is experiencing the fourth industrial revolution in the history of mankind-Industry 4.0. The State Council issued the "Made in China 2025" in May 2015, setting the national goal for China to achieve Industry 4.0 transformation by 2025.

In his book "Industrial Big Data", Professor Li Jie defined "Industry 4.0" as: "The fourth industrial revolution is an industrial value creation revolution with intelligence as its core, and its ultimate purpose is to achieve a high degree of integration of production activities. Make systems think and work together like people. "

The transformation of the intelligent Internet to the industry is mainly reflected in the following two aspects:

1) Intelligent production realizes the production chain from downstream to upstream based on flexible supply chain

Intelligent production refers to the creation of intelligent factories through advanced technologies and hardware products such as industrial robots, 3D industrial printing, unmanned logistics, and the Internet of Things, which can realize a reverse-customized production chain that meets the characteristics of Industry 4.0. 工业4.0 区分于其他三次工业革命的核心特点之一便是其基于用户端的生产力需求,打破了传统的刚性供应链和计划生产模式,能够实现下游推动上游的生产链条。



从宏观角度来看,工业大数据包括资产/设备数据、产品数据、用户数据、供应链数据;工业大数据可用于资产可靠性和设备故障监测、能源效率的管理、产品质量管控、生产流程优化管理、上下游和产品周期预测-管控-追踪、生产流程重塑、定制化生产等。每个环节1% 的效率提升作用到整体经济收益上都是一个可观的数字。但从现阶段来看,狭义的工业大数据主要集中在机器数据和生产流水线产品数据上;工业大数据的应用也主要集中在设备监测、能效管理和质量把控上,贯穿上下游和产品生命周期的价值大数据还有赖于产业的整体成熟和打通。

就目前国内而言,作为工业大数据底层基础的物联网普及、传感器的分布仍处于较为粗燥的阶段,万物互联基础下的工业4.0 理想状态仍需时间。

  • 农业



智能互联网发挥数据的智能价值,能够实现农业“种植(机器选种,自动播种etc.) - 施肥(基于环境和植物数据状态的精细化施肥etc.) - 灌溉(精细化自动化灌溉etc.) - 病虫害防治(疾病识别etc.)- 收割(自动化收割etc.) ”各个环节的效率提升与成本控制(资源节约),促成单位面值农作物产值的提高。高盛在一项研究中指出,智能互联网能够对美国农业生产各个环节实现7% - 30% 不等的效率提升,预计至2050 年,美国每英亩耕地的产值能够达到281 蒲式耳(对比2016 年的165 蒲式耳)





智能互联网理想状态下的农业,有望实现作物都有其对应数字参数所形成的“digital twin”,流转到市场里的每个农作物都能查询到完整的质量检测数据以及回溯作物生产成长过程中的各项数据,实现真正全数字化的食品体系。


  • 医疗



计算机辅助诊疗(computer aided diagnosis,CAD)在诊断中的应用已经有40年的历史,但近年来随着神经网络的出现,机器学习给CDA 带来了突破性的进步,尤其体现在医疗影像诊断方面;建立在自然语言处理技术基础上的机器问诊也处于快速发展期。IBM Watson可以在17 秒内阅读3469 本医学专著,248000 篇论文,69 种治疗方案,61540 次试验数据,106000 份临床报告。





传统药物研发周期长、成本昂贵、成功率回报率低。但智能互联网从一定程度上打通了新药研发各个环节的壁垒:在试验前期新药筛选时,可以通过综合算法获得安全性较高的几种备选物,提高成功率;在进入动物和人体试验阶段前,可以综合成分分析和既有已知药物的副作用数据库,选择生副作用几率最小的药物进入动物实验和人体试验,节约成本、提高安全性;此外,还可智能模拟和检测药物进入体内后的身体指标与剂量、浓度等用药指标之间的关系,推荐试验中的最佳用药方案;进入试药后期,可综合前期数据推算研制成功率,选择放弃成功率较低的药品种类,减少成本浪费。高盛一项研究指出,通过算法对后期试药的成功概率推算能够帮助美国每年至少减少10 亿美元的试验支出。


基因测序是一项极其复合摩尔定律的技术,随着基因测序技术的更新换代,基因测序的成本不断下降。2001 年平均每兆数据量基因测序成本是5292.4 美元,单人类基因组测序成本是9526.3 万美元;2006 年新一代测序技术推出,平均每兆数据量基因测序成本下降至581.9美元,单人类基因组测序成本下降至1047.5 万美元,而2014 年1 月Illumina 推出HiSeq X Ten 更是将单人类基因组测序成本降至1000 美元以下。随着基因测序成本的不断降低,基因有可能成为新的“健康代码”,通过基因检测、基因编辑等技术不断革新健康检测与治疗方式。

  • 金融





开年的一则新闻引起了金融圈一片哗然:由于工作都被自动交易程序接管,高盛在纽约总部的美国股票交易柜台雇佣的600 名交易员已被削减至2 名。根据《金融时报》,在2000 年,纽约证券交易所的场内交易者超过5500 名,而现在则不400 名。麦肯锡全球研究院在1 月推出的报告中称,金融和保险领域的工作,有43% 的可能性会被自动化替代。目前机器在金融交易领域可承担的角色主要有:自动化生成报告;海量数据库精准搜索;预测模型。自动化报告生成和精准搜索解放了大部分初级分析师的工作。而基于卫星数据、人流分布热力图等多维数据角度的算法模型,突破了人脑能处理的知识和数据边界,多数情况下能够精准指导市场交易,减少人为误差。但深度学习的“黑箱效应”是典型墨菲定律的践行者,机器微小的认知误差造成可能带来连锁多米诺骨牌效应;此外机器只能基于已有数据给出结论,但机器很难理解一家尚未盈利的公司该如何估值或一种新商业模式的价值在哪里。


早在1952 年, Markowitz 就提出了“投资组合理论”(Portfolio Theory)并因此获得了1990 年诺贝尔经济学奖。智能互联网时代,通过机器进行投资组合的搭配,又称为智能投顾(Robo-advisor)。智能投顾相较传统的理财顾问或个人投资理财,质量稳定、收费低,并且能计算超过人脑可处理范围内的丰富金融产品和收益影响因子,关联和平均做得足够好的智能投顾应用一般能够实现不错的收益,有望逐步发展为中产阶级理财的重要方式。

  • 零售



以Amazon Go 为代表的无人商店成为今年创投圈又一热词,但“无人商店”从字面意义上仅体现了智能化购物场景的自动化和人力成本降低,智能互联网给零售带来的想象力还包括:智能化购物导购机器人、手势操作和AR 结合的智能试衣镜、基于视觉系统的特定人群识别与营销、智能购物车;你可能在线下完成网上下单,可能在线下试穿线上挑选好的衣物,可能填完定制化的数据传输给工厂…智能互联网时代智能化购物场景的核心特点是:用户动作由「消费」转移为「体验」;用户消费逐渐模糊线上线下的边际。


7-Eleven 便利店创始人铃木敏文曾在《零售的哲学》一书中提到7-Eleven 垄断日本便利店过程中采用的方式:从社会环境变化预估消费者行为;创造出“单品管理”概念解决滞销品问题…上述手段所面向的都是传统零售行业决定成败的核心点— 库存,而智能互联网的推动目标,则是通过用户导向的供应链体系和实时反馈的物流仓储,实现动态零库存的理想状态。

在工业部分我们提到,工业4.0 理想状态能够实现从下游到上游的定制化生产,而位于制造上游的零售行业,以此对应的是以动态零库存为目标的供应链和物流革命。这里的“动态零库存“指的是基于大数据的实时特征、能够满足预测数量并随时调配的库存状态,而非时刻空仓的绝对零库存。实现动态零库存需要基于销售预测、客户偏好预测与精准营销、快速响应定成本的智能物流、动态定价等技术组合。


Reference ·Goldman Sachs: 《Artificial Intelligence : AI,Machine Learning and Data Fuel the future of productivity》·Jay Lee:《工业大数据》,机械工业出版社·Gartner: 7 Technologies Underpin the Hype Cycle for the Internet of Things 2016, http://www.gartner.com/smarterwithgartner/7-technologies-underpin-the-hype-cycle-for-the-internet-of-things-2016/·铃木敏文,《零售的哲学》,江苏凤凰文艺出版社·Michael Sacasas :Kranzberg's Six Laws of Technology, a Metaphor, and a Story, http://thefrailestthing.com/2011/08/25/kranzbergs-six-laws-of-technology-a-metaphor-and-a-story/·孙希有,《流量经济》,中国社会科学出版社








Spider pond