On July 9th, Microtek announced that it has partnered with China Mobile to complete the NB-IoT R14 standard rate enhancement test and become the first chip manufacturer to pass the test. This test uses a development board based on MediaTek MT2625 chip. , R14 measured uplink peak rate up to 150kbps, downlink peak rate reached 102kbps, is 4-6 times the R13 standard rate, indicating that R14 has been in the remote meter reading, over the air firmware upgrade (FOTA), push to talk (Push to Talk or The application capabilities of multiple scenarios, such as Voice over Message, help push the Internet of Things market to a broader development space and create more value for the industry chain. With the development of the Internet of Things market and the abundance of business, the R13 standard is The peak line rate can no longer meet more and more IoT service requirements. In order to ensure that it can achieve the same rate level as GPRS, it can better meet the high requirements for remote upgrade and NB-IoT data transmission rate and delay. Power-over-Large-Overlay (LPWA) service scenario. The evolving NB-IoT R14 standard increases the uplink and downlink peaks by increasing the transmission block (2536 bits TBS) and adopting the dual-process (2 HARQ) technology. The rate can reach more than 100kbps. At the same time, the NB-IoT R14 standard has also been enhanced in terms of mobility, positioning, multicast and multi-carrier technologies, and further provides better standards and technical support for the development of the Internet of Things market and industry maturity. . Deputy CEO of MediaTek Technologies and General Manager of Intelligent Device Business Group Zeng Jie said: 'MediaTek is pleased to be the first chip manufacturer to pass the China Mobile NB-IoT R14 interoperability test. MediaTek actively promotes the development of standards for the Internet of Things. Technology R&D plays an important role in the development of the Internet of Things. At present, we have two chips that support R14 – single-mode MT2625 and GSM+NB-IoT dual-mode MT2621. These two chips will be widely used in industry With the consumer Internet of Things (IoT) market, accelerate the comprehensive arrival of the Internet of Things. 'MediaTek and China Mobile have worked closely together on the Internet of Things. In June last year, they jointly launched the industry's smallest (16mm x 18mm) NB-IoT universal model based on the MT2625 chip. Group, to help companies quickly and easily achieve the development of IoT devices. After that, both parties cooperated in NB-IoT R14 interoperability testing. Internet of things development It is necessary to actively cooperate with the network-side and chip-end vendors to ensure a good experience of the IoT device. The cooperation between China Mobile and MediaTek at NB-IoT is an example. In the future, the two parties will cooperate and explore in more areas, and advance together. The win-win development of the Internet of Things industry.
2. Samsung's first public investment in blockchain, layout as early as three years ago;
Samsung's first blockchain investment begins with the Internet of Things.
Pang Qian Recently, blockchain startup Fliament announced a $5 million Series A round of financing. The investors are Bullpen, Capital, Verizon Ventures and Samsung Ventures.
Blockchain startups get financing is no news. It is worth noting that the appearance of consumer electronics giant Samsung in this investor.
According to public information, Samsung Ventures is known as Samsung Ventures. It is an investment company of Samsung Group and is committed to investing in early-stage capital companies. It focuses on investments in technology, IT, hardware, software and other industries.
The Samsung Ventures formally appeared as an investor, marking Samsung's first public investment in the blockchain industry.
In fact, as early as 2015, Samsung began to quietly deploy the blockchain industry.
Samsung did not openly invest in the previous three years. One of the major investment methods is to follow IBM's participation in the ADEPT blockchain project. ADEPT uses Bitcoin and Ethereum Networks to enable devices to communicate with each other, aiming to connect consumer devices to the Internet of Things.
Another investment method is to invest in the name of a subsidiary company.
In 2016, Samsung SDS, the largest subsidiary of the Samsung Group, announced that it will invest in a local company that provides companies with a blockchain development platform. The reason is that blockchain technology is a financial innovation technology that can change the world.
In January 2018, Samsung Data Systems announced the investment in two start-up companies, one of which is Blocko, a Korean blockchain company, and the other is Darktrace, a UK network security company.
At the same time, Samsung’s data system stated in an official statement:
'In order to improve the competitiveness of the company and promote the development of the business, the future will continue to invest in more start-up companies in the international scope. 'This also shows that this type of investment will become the norm.
These investments are mainly focused on technology introduction. The purpose is to introduce cutting-edge technologies in the field of network security and blockchain.
Although the amount of this investment in the Fliament was small, it showed Samsung's consistent strategy in the blockchain.
Fliament is similar to ADEPT in that its IoT device will be able to communicate securely, execute smart contracts and send micro-transactions, which will greatly increase the transaction efficiency of large companies in all industries. That is, Samsung Investment and the previous investment in ADEPT are similar. Expect it to help Samsung break through the IoT scene.
Next, Samsung will continue to pay attention to the blockchain field and invest in the project.
In March 2018, Brendon Kim, managing director and investment director of Samsung's wind investment company 'Samsung NEXT', once suggested that blockchain is crucial for ensuring communication between devices, and said that the blockchain is currently Samsung's NEXT One of the investment themes. 36氪
3. Toshiba's success or failure in China? Semiconductor memory business is the key;
'Global Network Technology Reporter Wang Huan', 'Nihon Keizai Shimbun' reported on April 9th that Toshiba’s new operating system has started to sail under the risks associated with the sale of semiconductor memory business. The sale of semiconductor memory business is a prerequisite for business reconstruction. , Failed to complete the plan before the end of March as originally scheduled, and there is still no prospect until April 3. This is because it failed to pass the anti-monopoly review of the Chinese government. If the sale of a business with a price of up to 2 trillion yen is delayed for a long time , It may affect Toshiba's financial strategy.
In September 2017, Toshiba and the U.S. investment fund Bain Capital signed a sale agreement for the storage subsidiary 'Toshiba Memory'. The sale is premised on the antitrust review of each country.
Success or failure depends on the Chinese government
The review of the United States, Japan, and other countries has successfully concluded, but the review of the Chinese government, which was regarded as the biggest obstacle, was not launched until December 2017. Generally speaking, the review time is 4 months. The target for the sale was originally scheduled to be completed by the end of March. At the time, it was worrying. In fact, it was not passed by the end of March.
The Chinese government targets companies that sell a certain number or more of products in global and Chinese markets, and conducts anti-monopoly reviews at the time of merger and transfer of business rights. China has defined semiconductors as a policy industry, and sold them around Toshiba Memory. There are many opinions that the review will be long-term.
If the sale has been stalemate, the previously planned investments of Toshiba and Toshiba Storage may be forced to adjust.
Toshiba Memory is building a new factory at its main Yokkaichi plant (in Yokkaichi City, Mie Prefecture). It is also planning to build a new factory in Kitakami, North Iwate, and to start building a plant during the year. A Toshiba executive revealed that 'we will not be able to afford huge amounts in the future.' The risk of investment' is one of the reasons for the sale of the storage business.
Bain Capital and others are planning to inject investment funds after the acquisition of the memory business. The more delayed the transaction, the more likely it is that Toshiba’s burden will increase. In the context of a huge investment by rival Samsung Electronics, if the investment is slow , Toshiba Memory's corporate value may be affected.
According to the transfer agreement between Toshiba and Bain Capital, after April, Toshiba will acquire the right to terminate the contract under certain conditions. Bain Capital will also be given the right to terminate the contract in July. If the transaction is delayed for a long time, the opposition will be required. Shareholders' interest in this transaction has drawn more attention.
On April 3, Che Guchang, chairman of Toshiba’s president and chief executive officer (CEO), said in an interview on the 3rd, 'We endeavor to complete the sale as soon as possible’, and also stated that '(such as the storage of such business) is not a bad thing”.
Toshiba’s shareholder capital is 460 billion yen at the end of March, and the shareholder capital ratio is expected to be 11%. However, Toshiba’s end-December 2017 was bearing 1.1 trillion yen interest-bearing liabilities due to bank borrowings.
The main transaction bank has been urging Toshiba to sell its storage business in order to ensure the recovery of loans and credit lines. If Toshiba receives a sale of 2 trillion yen, it plans to use it to secure growth and repay bank loans. If the transaction process is stagnant Long-term, it may affect the financial strategy.
Lose the main cash cow
Toshiba is scheduled to hold a regular shareholders meeting in late June. In order to curb the spread of some shareholders’ opposition to the sale, it hopes to complete the sale by the end of June.
However, even if the sale is completed, Toshiba's reconstruction will face many problems due to the loss of its memory business as the largest cash cow.
Toshiba still can't find any other business with annual operating profit of over 100 billion yen like storage. The social infrastructure business such as elevators and railways, which is positioned as the profit pillar of 'New Toshiba', is also facing saturation in the domestic market.
In overseas countries, the thermal power generation related businesses such as gas turbines, which are Toshiba's strengths, are experiencing sluggish growth due to renewable energy such as solar energy and wind power. Che Gu also said that 'we need to strengthen our basic operating capabilities'.
In addition, corporate governance reform, which has become a hotbed of business crisis, is also a topic. The nominating committee of the management team is composed of independent directors in principle. The majority of the directors are changed to independent directors, and external supervision functions are strengthened. In November 2017, Toshiba will The number of executive directors has been reduced from 23 to 15 and the management system has been streamlined.
In 2015, the Toshiba financial fraud scandal was exposed. Since then, the internal management system has improved, but a Sun company discovered the problem of accounting for non-compliance costs in December 2017. To promote business restructuring, it is also necessary to ensure transparency in corporate culture reforms. . Global Network Technology
4. Behind the financing of 600 million U.S. dollars: Shang Tang accelerates fission;
On April 9, Shangtang Technology SenseTime completed a $600 million Series C round of financing, led by Alibaba Group, the Singapore Sovereignty Fund Temasek, Suning and other investment institutions and strategic partners. The new round of investors will provide more for Shangtang. Rich application scenarios and stronger overseas distribution capabilities, accelerating the landing of AI technology and ecological construction. In the field of artificial intelligence with strong rivals, the window period of startups is only three or five years. Shang Tang has been accelerating for more than three years since its establishment. Evolution to an AI platform company.
Author | Sun Jing Yan Dongxue April
Editing | Zhao Yanqiu
Sources | AI Finance (ID:aicjnews)
The most in-depth media in China's new economy
The industry does not understand the business model.
For more than three years since its establishment, the artificial intelligence startup has burned a lot of money, and has done the bottom technology that only giants such as Google will do; it has laid out a big show and claimed to apply artificial intelligence technology to hundreds of industries; When the domestic business has not yet fully stabilized, it has extended its tentacles overseas; even when the company itself is still producing blood, it has begun capital operations and external blood transfusions.
But it is enough to attract attention. A startup company, which has monopolized more than 150 doctoral degrees in the direction of deep learning in artificial intelligence, won over 400 heads of customers at home and abroad, and its annual business growth has exceeded 400% for three consecutive years. More than US$4.5 billion, it has become the highest-valued artificial intelligence startup in the world, and has repeatedly set a record of global AI financing.
Before and after the establishment of Shangtang, the field of computer vision in China has been turbulent. In 2011, Tsinghua graduates Inaki and his classmates founded Contempt; in 2012, MIT postdoctoral researcher Zhu Tsang returned to the country to establish plans to follow; in 2015, Zhou Ye from the Chinese Academy of Sciences set up Yun Cui from Chongqing. By 2017, Shang Tang, Despise, Yun Cong, and Yi Tu, emerged from a crowd of start-up companies and became the unicorn under the artificial intelligence field with computer vision as the core. .
Record the growth path of these companies, as an analysis of fissile business samples. Take Shang Tang as an example. In the years before and after its establishment, it basically did a few things: From scratch, lay the foundation, develop the underlying technology; from zero to First, to do projects, to place artificial intelligence technology; From one to one hundred, to form a mature product in the industry. Now, it began to take the platform to do ecology.
From this fission process, we can see how Shang Tang is using the high-speed iterative artificial intelligence technology to shake the industry market from multiple dimensions in the industry window period, while avoiding giants' range with 'escape speed'. Development builds a business territory with great potential.
Read business soup
One day, Yu Kai, the founder of Horizon Robots, could not help but ask Shang Li CEO Xu Li: Do you really invest so much in building your own supercomputer center?
The Supercomputer Center has burned money. Dr. Yan Junjie, a researcher at Shangtang, once made an estimate. Pressing the button labeled 'run' would cost the whole training session at least 500,000 yuan. 'We have more than 150 Ph.D. On the algorithm platform, press 'run'. Xu Lixiao said that this is not counting the annual GPU cluster purchase budget of several hundred million yuan.
Not only that. At the level of deep learning, the vast majority of domestic AI companies rely on Google, Facebook and other artificial intelligence open source frameworks such as Tenserflow and Torch. However, business users have to pay for themselves to use the underlying framework. It is like a software company announced that it wants to Outside the development of a separate operating system, this idea is somewhat bizarre.
Shang Tang’s early investor, IDG Capital Partner Niu Kuiguang was the first investor to perceive Shang Tang’s aspirations. He remembers that Xu Li’s use of “cat brain” and “monkey brain” analogy explained to him The current rise in artificial intelligence can be compared to 'using data to build the brain'. If the current level of artificial intelligence technology is in the 'cat brain' stage, most companies are doing is 'how to use the cat brain to catch the perfect Mouse'; and the excitement of Shang Tang is to create a platform to train the 'cat brain' into 'monkey brain', which essentially enhances the ability of the 'brain'. After the brain is upgraded, the machine can learn more quickly and master more complex skills.
Xu Lihui will not hesitate to emphasize, 'We are an artificial intelligence platform company with Chinese original technology, not a face recognition company.' Their goal is ambitious, so more needs to be done.
Niu Kuiguang still remembers the unique impression the company initially gave him. In 2014, he saw a dissertation of a team of Prof. Tang Xiaoou from the Chinese University of Hong Kong: Applying the DeepID algorithm they developed to face 20,000 faces of data sets, computers. For the first time, the recognition accuracy rate exceeds the human eye. This is an important signal - meaning that computer vision technology may be commercialized. He also discovered that during the period from 2011 to 2013, the top conference of the global computer vision ICCV/CVPR 29 of the top conference papers, Tang Xiaoou team accounted for 14. Niu Kuiguang immediately flew to Hong Kong, visiting Professor Tang.
In Hong Kong's Chinese multimedia lab, he saw demos of five or six computer identification technologies. The research direction was consciously combined with practical applications. For example, a demo was used to turn pictures into impressionist styles. This is now the popular Mito function. 1. The research team also repeatedly emphasized originality, 'We are the source of the ground-breaking algorithm'. These have left Niu Kuiguang a deep impression.
In 2014, IDG Capital raised tens of millions of dollars to help the research team step out of the lab and find the key to AI changing the world in real business. Niu Kuiguang took a big risk. After all, in 2014, the AI industry was still unpopular. No one can tell whether technology can be successfully transformed into commercial value.
Going out of the lab, how the new generation of technology companies should go is a question that the founding team has to think about. In the field of artificial intelligence, there have been previous lessons.
For example, in the field of speech recognition, technology has long passed the window period of technology pioneers, and the accuracy rate of each PK is higher than 1%. There is no substantial meaning. Since BAT collectively realized, intelligent speech is the next generation of human-computer interaction. After the entrance, the market that must be seized, the use of funds, and especially the advantages of large platforms, it is possible to crush the startup companies in minutes. In the field of speech recognition, there is no longer a new giant company with unicorns or hundreds of billions of market capitalization.
Technology start-up companies only use the window period for technology to start, enter the industry, or combine with hardware, or combine with data, or combine with traffic to form products or even platform services, and establish barriers before they have the opportunity to reap dividends.
Moreover, the premise of all this is fast enough. Silicon Valley technology innovation in the United States also follows the principle of 'escape speed' - in astrophysics, the escape speed is the minimum speed required for an object to escape its gravitational pull and fly to space. 'If the technology With the degree of innovation and future development, you can achieve an 'escaping speed' and you will be able to seize the time window and form a big platform. ' Xu Liwen's speed, his thinking is extremely fast, like this company's development rhythm.
The establishment of this 'escape speed' is closely related to Shang Tang’s early-stage reserves. Co-founder Xu Bing used 'regardless of cost' to describe the investment of the year. After the financing was in place, they burned a lot of money in two places: First, 'talents' Monopoly', from 2014 to 2016, Shang Tang used packaging solutions to find all the top scientists in this field. He has accumulated a large number of deep-learning talents, including graduates, and formed a strong talent team. The second is to build hardware. Computing platform, buys more than 6000 GPUs from Nvidia, builds a supercomputing center, and builds powerful production tools.
Xu Li believes that although Moore's Law in the field of hardware has been invalidated in some sense, Big Data, the algorithm has entered a new era of Moore's Law, and the Supercomputing Center is necessary to maintain the algorithm's ability to iterate with Moore's Law. The underlying infrastructure.
Shang Tang’s two investments were initially extremely risky. At one time, the company faced the risk of capital chain fracture. However, with the deepening of commercialization, the advantages are also highlighted. For example, artificial intelligence is seen in an industry before others. Break through the timeline of the industry's red line, prior to other people investing resources in cooperation with an industry head company to engage the industry to land; prior to other people to make open source systems such as Google Tensorflow could not support features.
Xu Li cited an example. Smart phones have the innovative function of 'dual cameras for real-time photo taking'. Shang Tang is the only company in China that uses end-to-end deep learning technology to provide technology. Because the technology involves a number of newly designed networks, it is Existing open source frameworks are not supported.
When we entered many fields, it seemed that there was competition, but as the demand deepened, technical requirements increased, and doing it became a blue ocean. ' Xu Li said.
In addition to the uniqueness of technology, technology landing, especially commercialization of scale, is the biggest test for all AI companies.
Co-founder Yang Fan analyzed that: Shang Tang's strategy is '1+1+X', and at the same time, diversified layout, rather than just betting on a certain subdivided field. This is not only beneficial to the overall iteration of the algorithm, but also facilitates the card position competition. After all, security, finance, mobile internet, mobile phone, retail, and autopilot are the focus of competition, contempt, and cloud-based services. Although each has a business focus, the above-mentioned lines are the ideal scenario for commercial development.
In the vertical field, Shang Tang’s approach is to find the head company first, to explore the AI technology landing, from the project to the product; and then to open up an industry ecology vertically and do the platformization. “In the field of security, for example, when we put the front end The camera, the backend cloud platform and even the whole chain of the chip are open, then it can be the largest industry technology platform.
After accumulating a number of industries, the cross-industry cross-industry appears, and there is also the possibility of making a big platform.
AI Financial Club asked Xu Li whether he feared that this strategy would become a public enemy of the industry. 'Jin Yong Fan' Xu Li talked about martial arts: There are many schools in the arena that focus on training, you practice knives, and he practices sticks. They are incompatible with each other. Suddenly, there is a spirit of internal skill called Jiuyang Divine Power. After focusing on internal strength cultivation, swords, guns, sticks, and sticks can all be enhanced.
'Shangtang' is to make Jiuyang's magic power into the underlying technology, redefine the practice of knives and sticks, provide it to the industry, then continue to practice knives with knives, continue stick training with sticks, but all have more progress.' Xu Li explained that this formed a strong barrier, while providing the industry with martial arts, rather than enemies. Shang Tang positioning is to empower Baiye.
This kind of positioning is like Bosch in the automotive world and Intel and Qualcomm in the chip industry.
However, the 9 Yang magic was not so good. For some time, in knowing the truth, there were too many former researchers evaluating Shang Tang's researchers. They were not doing the rhythm of a commercial company, and the transition from technology to product was not timely.
At the beginning of the venture, Shang Tang experienced various difficulties. Later, under the leadership of Niu Kuiguang, Shang Tang received a multi-million-yuan loan from Pufa Silicon Valley Bank; Ren Quan's StarVC 'snowboarded' to cast A+ rounds. With the help of these two emergency funds, Shang Tang plans to build an engineering force to fight the market.
Until the end of 2016, the concept of artificial intelligence continued to heat up, and a large number of artificial intelligence technology requirements emerged in the market. Business technology broke out before it broke out in many industries.
Difficult to land
In terms of commercialization of scale, technological advantages are not a master key. In the new field of artificial intelligence, startups have been detouring in the early stages of commercialization.
Shang Tang co-founder, employee No. 001, Tsinghua graduate Xu Xiheng recalled that in 2014, Shang Tang, who had just come out of the lab, first made a 'software development kit' as a product program promotion. But when they contacted China Mobile, UnionPay and Loan Bao found that 'the demand presented by the other party is not completely related to technology and is an application method or function point derived from the technology.'
In the process of going to the market, changing from technical thinking to product thinking is a challenge for Shang Tang team.
The cooperation with Lending Po was a great beginning for Shang Tang’s transformation. Initially, this Internet finance company was very satisfied with the use of face recognition technology for registration and authentication. However, after half a month’s discovery, there were hackers. Through the illegally obtained ID data and photos, the attack system has received subsidies for online promotion activities.
Attacks are ever-changing. There are Gif animations to confuse people's faces. There are holes in the eyes that make the eyeballs move.
However, Shang Tang’s technical team can quickly update the algorithm and destroy the new attack mode within two or three days. After almost two months of attack and defense, Shang Tang almost defended all attacks on the market.
Shang Fan co-founder, Yang Fan, who is responsible for the engineering team, summed up two experiences in this process: First, it is necessary to have the mentality of serving customers, and to quickly integrate products with customers; Second, it is necessary to maintain strong artificial intelligence originality. R & D capabilities.
In the past era of computer virus defense, it may take several months for each new attack method to emerge. However, thanks to the rapid iterative advantages of artificial intelligence algorithms, if the engine is powerful enough, it must be equipped with powerful data. The super-computing capability may solve the problem overnight. This speed, and the design of algorithms based on expert knowledge are not at all in one order of magnitude.
In the first half of 2016, Shang Tang wanted to promote this online face recognition registration and identity authentication program to the entire industry market. At that time, other competitors could not have a good enough anti-attack plan. In the cooperation and interaction, Shang Tang accumulated key technologies and data. Thanks to this, Shang Tang occupied 70% of the market after two years.
Bonding the head customers to innovate in the industry is a mature game of Shangtang. In 2017, Shangtang and Shenzhen Public Security Bureau launched the application of urban scale security, which will video of thousands or even tens of thousands of cameras in the city. Put together, if necessary, find the city trajectory data of a particular person in the past 3 months.
In the past, the industry could only make video queries on one or more than one camera. The challenge to the technology is comprehensive. The Shenzhen Public Security Bureau will combine the needs of the next three to five years with Shang Tang's technology. When the product is polished After a certain degree, it can be standardized and promoted to other large cities in the country.
The exploration of commercial scale continues. In the traditional software industry, there are currently three major game rules: selling products, selling services, and selling products and services. However, in the field of new artificial intelligence, there is also a state of chaos.
'In a sense, what we provide is intangible. It is a new type of technology implementation in itself. It is not a figurative product.' explained Liu Gang, vice president of business team for Shang Tang.
Is it to provide a common technology, or to create a product that can include different components, or to connect different algorithms to different industries in the underlying platform? In the early days, the team also had brainstorming, but did not come to a standard answer.
Shang Tang tries to unify his ideas with another path - empowering Baiye. While walking closely to the needs of the industry, he gropes.
But for a start-up company that has been established for less than four years, one of the first realities encountered is: Among the major industries it has attempted to enter, some have long been entrenched in traditional giants.
For example, in the mobile phone industry, traditional visual companies have established a mature cooperative relationship with mobile phone manufacturers. The latter will need to come up with something different.
One of the needs of Xiaomi’s mobile phone is the baby album—when the mobile wallpaper switches, it automatically picks out all the pictures of the baby from the album and replaces them in turns. This is the face clustering. The “one to one comparison” with face recognition. Differently, face clustering needs to gather the same person's photos in thousands of photos. After understanding the requirements, the research side and the engineering side develop together and finally come up with the products that Xiaomi is willing to pay for.
From face clustering, Shang Tang has signed a strategic cooperation with OPPO to create an OPPOAR developer platform that uses Shangtang SenseAR platform as the engine to promote the application and popularization of AR in mobile phones.
In addition to these large-scale industries, Shang Tang is also expanding into new industries, such as smart medical care, robotics, remote sensing, and financial big data analysis.
How can startups take care of so many industries? Liu Gang, the vice president of the lead business team, said that priority must be given first.
He also believes that the 'T'-shaped framework is more suitable for its own company: For example, in the security industry, after completing a project and doing the next project, it is necessary to extract common patterns from similar businesses and accumulate a set of 70% of the industry. ~80% common solution, which can greatly increase efficiency. This is the horizontal of the 'T' word, creating a horizontal product matrix.
Even among different industries, technology versatility can be abstracted. An interesting example is that the technology of 'scarling people' in the security market can also be used to 'grasp people', such as new retail. August 2017, Suning The first unmanned store 'Suning Sports Biu' opened.
Customers only pay for the first 'brushing' step from the store to checkout payment. 'You will find that there are a lot of key technologies in this area that are very versatile with the security identification technology we already have.'
On the horizontal side of the 'T' word, around each of the vertical subdivision areas, 'have to walk a little further'. This is the deep understanding of the industry. For example, in the Internet industry, face recognition technology, Plus augmented reality technology.
Sometimes, this half-step more than just technology, it may be business.
In the live broadcast platform, advertisers’ resources have been placed on the Shangtang platform. Shangtang, the broadcast platform and third-party advertising providers can build an ecosystem to better embed ads in the live broadcast. Another example is an Internet cafe or hotel. It takes a few seconds before the face authentication takes place. Putting advertising resources under the line, putting it on the brush face machine will produce extra value.
In this line of thinking, Shang Tang tries to promote identity verification machines at Internet cafes free of charge. Subsequent advertising costs can also be used to open a market faster.
“The more industries that are doing, the more cross-maturity effects will emerge between the industries. The more opportunities there are,” Yang Fan, co-founder of Shang Tang, said. Under such considerations, if they enter a new industry market, they will not bring about excessive costs. Nor will it lead to excessive dispersion of resources.
Shangtang is also using and improving the automated tool chain. 'High-productivity tools represent the speed and ability of innovation to achieve sufficient growth for limited personnel.'
However, there is still a bottleneck in the market. The landing of artificial intelligence in various industries is an ongoing issue. This is an iterative and gradual process. PricewaterhouseCoopers has a rough estimate of the market for AI visual identification technology companies. They believe that the security plate is the largest. From the 2016 to the 2020 years, it will maintain a compound annual growth rate of 90%. But in other industries, it is difficult to reproduce a full bloom.
In this context, Shang Tang has turned his sights overseas to explore a wider margin.
Battle of overseas start
At the academic annual meeting in 2015, Tang Xiaoou and Xu Li met Lao Shizheng. Lao Shiyi is also a heavyweight expert in computer vision. He was sent to Japan to study in the 1980s and was responsible for the Japanese Omron Face Visual Group Technology. people.