Find new opportunities for domestic smart chips to "start at the same time" in the five major industries of AEIMR

Recently, a team of professional analysts from Beijing Huaxing Wanbang Management Consulting Co., Ltd. went to Nuremberg, Germany to attend the "Embedded World 2019" and exchanged ideas with many global and domestic leading technology companies participating in the exhibition. The theme of this exhibition is "Embedded Intelligence", which is very consistent with the purpose of Huaxing Wanbang analysts' trip, that is, to explore how to use the development opportunities of the rapid evolution of intelligent technology and applications to promote the innovative development of the integrated circuit industry and achieve a win-win situation for the application industry and the integrated circuit industry.

Huaxing Wanbang launched a research report for this conference, revealing that in the process of evolution from embedded computing or real-time control technology to intelligent technology, the new generation of smart cars (Automotive), edge computing (Edge), industrial electronics (Industrial), medical electronics (Medical) and robotics (Robotics) and other five major industries will achieve value reconstruction and provide domestic smart chips with the opportunity to "start at the same time". This development trend also provides a number of suggestions on how to seize these opportunities.

The report will start with analyzing the value reconstruction brought by intelligence, and put forward four suggestions: innovation in intelligent chip architecture, making good Chinese chips, expanding the ecosystem for master control, and paying attention to the trend of AIoT intelligent network integration, to communicate with colleagues in the industry. The following is the text of the report:

Walking into the "Embedded World 2019" from the East Gate of the Nuremberg Exhibition Center (NICC Ost), the first company that Huaxing Wanbang's senior analysts saw was Solectrix from Germany, a design company specializing in embedded systems, developing various high-end designs for application markets such as camera systems, medical technology, industry and automobiles. In addition to the industrial cameras, robotic arms and car models displayed by the company, what was impressive was the various system-level module (SoM) products launched by the company for these applications.

Control panels or system-level modules for the AEIMR industry are a large market

In fact, if you visit this exhibition carefully, you will find that there are many design companies or system-level module providers like Solectrix. Their module products often use high-performance processing and storage chips such as FPGA+DSP+DDR, plus peripheral chips such as high-end clocks and power supplies from companies like Silicon Labs to form hardware modules or small circuit boards. They also combine their professional experience in reliability and application optimization in related industries, as well as software and algorithm engineering development for specific applications, to launch high-value-added products and sell them to world-class automotive, industrial equipment and medical electronics manufacturers.

Intelligent networking reconstructs value composition

But it is worth noting that whether it is "traditional" electronic systems such as smart cars (Automotive), industrial equipment (Industrial) and medical electronics (Medical), or emerging industries such as edge computing (Edge) and robotics (Robotics), they all need a large number of such modules, or ASIC chips that are very similar in architecture. This is a proven market, but it does not mean that it is a cash cow market that existing module manufacturers can always sit back and enjoy the fruits of their labor. This is because new technologies and demands are coming.

If we combine the automotive, embedded, industrial, medical and robotics markets into the AEIMR industry market and observe them, we will find that the development of its underlying technology is evolving from the previous platform period to a new architectural innovation period. This is because the five AEIMR industries are all experiencing re-innovation in intelligent networking. In other words, in the AEMIR industry, the control and processing circuits in existing chips are important, but new intelligent technologies and functions are reshaping their value composition.

Commercially available chips may not necessarily support the emerging intelligent applications in the AEIMR industry, and the demand for smart chips will become more prominent

From the current situation, artificial intelligence chips are the best field for cultivating unicorn companies. For example, companies such as Cambrian Technology have become unicorns in the industry. However, artificial intelligence chips have entered the second stage of practical application from the early stage of solving the problem of whether there is one or not. The unicorn companies that will emerge will be those manufacturers that can quickly provide chips and solutions to the final market. In addition to scene recognition technology on mobile phones, the intelligent needs of the five major industries of AEIMR will give birth to a new batch of intelligent chip unicorn companies.

Value reconstruction brings opportunities to the chip industry

This value reconstruction is bringing new innovation space and product market. System-level modules can be replaced by application-specific chips (ASIC) or system-level chips (SoC) after the market scale reaches a certain amount. For this reason, Huaxing Wanbang believes that the demand for these architectural innovations, coupled with the technical integration of system-level modules into new ASICs and SoCs, is ripe. The time has come for domestic chip manufacturers to "start at the same time" with global manufacturers in the AEIMR industry market.

Machine vision is an important entry point to intelligence. IPs such as Imagination's PowerVR series GPUs can help chip design companies quickly enter the market.

"From the perspective of chip architecture, chips for AEIMR and other industry applications have many things in common. First, they all need to develop high-performance accelerators for intelligent applications. Second, they also need higher system reliability/internal security, as well as the ability to handle multiple protocols and data format conversions." Liu Zhaohui, special chief analyst at Beijing Huaxing Wanbang Management Consulting Co., Ltd., said. "If domestic chip companies can define chip products based on architectural requirements, and then combine the latest semiconductor intellectual property (IP) and software algorithms and industry experience, then they can embark on a path of independent innovation."

How big is the market for AEIMR's five major industries?

The five major domestic AEIMR industries are developing rapidly or have huge stocks. At the same time, intelligent networking has become a major trend in the development of these industries, which will bring huge opportunities to domestic chip manufacturers. Below, Huaxing Wanbang makes some simple analysis on the situation of the relevant industries:

Automotive: In 2018, my country's automobile production and sales reached 27.809 million and 28.081 million respectively, once again ranking first in the world, but the growth rate has slowed down significantly; at the same time, the market share of Chinese brand cars declined in 2018, and the cost/price leadership model began to be challenged. Intelligent network technologies such as advanced driver assistance systems (ADAS) and self-driving cars have brought new opportunities to both domestic automobiles and the domestic chip industry (please refer to Huaxing Wanbang's article: "Facing the turning point, starting from the "core" to promote the collaborative innovation and development of the automobile and chip industries").

China Association of Automobile Manufacturers released the 2018 national automobile production

Edge Intelligence: Edge computing is a widespread and extremely diverse field. Applications such as smart cities and smart security have created huge demand for artificial intelligence chips. Many regions and departments are updating their video surveillance systems and replacing traditional IP cameras with cameras with video recognition capabilities. At the same time, intelligent applications such as face recognition at related checkpoints such as airport security checks and specific places such as hotels are also leading the world.

Industrial applications: Industrial applications not only refer to manufacturing, but also include transportation, mining, logistics and other industries. The intelligent demand of manufacturing alone is at the trillion-level scale, so Premier Li Keqiang mentioned the concept of building an industrial Internet platform and "intelligence +" in this year's government work report. Industrial applications involve a variety of sensor fusion, control systems, intelligent processing and communication protocols. From the "Embedded World Exhibition" in February to the Shanghai Munich Electronics Show in March, it can be found that industrial cameras and their graphics post-processing systems are a huge industry, and this is just a small application of industrial intelligence.

Many manufacturers are promoting innovation around the five major industries of AEIMR, which is an important opportunity for domestic smart chip design companies.

Medical Smart Healthcare: As my country enters an aging society, the government increases its investment in people's livelihood, and individuals pay more attention to their health, coupled with the accelerated construction of a community-based, health-oriented, and balanced urban and rural medical service system, the demand for smart medical and health care devices in my country will continue to be strong, and it will also be the focus of innovation in industries such as artificial intelligence. The application of using artificial intelligence technology to assist doctors in interpreting medical images is developing rapidly, and remote surgery trials based on 5G technology have also been successful; in the future, miniaturized medical-grade 3D image and video processing and transmission equipment will also be a huge new market for smart chips.

Robotics: In China, robots are an industry with a wide range of categories and huge potential demand. In addition to the industrial robots with an annual demand of hundreds of thousands, there are also automated guided vehicles (AGVs) that are constantly entering various industries, and unmanned delivery, cleaning and service vehicles that will generate huge demand once they enter the commercial market, and of course various service robots, consumer robots, etc. Not every robot needs a reducer, but every robot needs a controller, and many types of robots require machine vision and artificial intelligence post-processing, which brings huge market opportunities for the development of smart chips.

AEIMR's five major industries' smart hardware requires smart chips

There is no doubt that the five major industries of AEIMR are welcoming a new era of intelligence. It seems that today's semiconductor industry has provided various chip products such as CPU, GPU, FPGA and DDR6 to support its development. Everyone only needs to develop and optimize software for applications, but this is not the case. A very clear example is that when manufacturers develop autonomous driving solutions, the top priority is still the new autonomous driving "brain" - controller, domain controller or multi-domain controller, rather than using commercial processing blocks or modules on the market.

Alan Kay, Turing Award winner and the most important figure in today's computer field, once said: People who pay close attention to software should create their own hardware. This sentence is very appropriate in the current intelligentization of the AEIMR industry. Therefore, the final kings in these industries are the top hardware experts. Of course, creating the top intelligent hardware is inseparable from innovative intelligent chips, whether it is developed by itself or jointly defined or developed with professional chip design companies.

To this end, Chen Hao, senior analyst at Huaxing Wanbang, added: "AEIMR is an important application field for new technologies such as artificial intelligence, 5G and the Internet of Things. ASIC chip or system-level module suppliers in these fields and their end users, no matter how successful they have been in the past, are facing the challenge of re-innovating their architectures driven by these technologies. Domestic chip manufacturers entering the market from the direction of intelligent networking can achieve the same level as the traditional advantages of existing manufacturers. This is a good opportunity for innovation-driven development of my country's integrated circuit industry."

However, it is not easy to develop smart chips and solutions at present, which is why there are so many unicorn companies in the field of artificial intelligence and autonomous driving chips. An intelligent chip must meet the application needs of various industries in AEIMR. It must not only design hardware acceleration based on the sensor characteristics and data features of each industry, but also consider constraints such as power consumption and volume. It must also undergo safety and reliability certifications such as industrial and automotive standards, and it must also maintain sufficient flexibility for the evolution of standards and protocols.

To this end, Huaxing Wanbang recommends that domestic chip companies start from four aspects: innovative smart chip architecture, making good Chinese chips, expanding the ecosystem for master control, and paying attention to the AIoT intelligent network integration trend, so as to develop corresponding cost-effective chips for the AEMR industry market.

Smart chip architecture innovation

Different from the traditional processing mode of sensor data by industrial control units, ECUs or PLC processors in the AEIMR industry, many future applications require real intelligent functions, that is, to add new functions such as training, reasoning and low-latency decision-making to the existing processing functions, that is, on the end-side devices, which will make the chip architecture design more difficult. This is because while computing power has increased significantly, it is also necessary to consider constraints such as storage bandwidth, internal and external data channels, volume/area, overall power consumption, information security and reliability.

It is precisely because of these constraints that many extreme applications cannot be built entirely with commercial chips and modules on the market, so it is inevitable to develop your own smart chips, and these chips require innovative architectures and the highest-performance IP. In order to ensure that the newly developed chips have high performance and low latency for smart applications, it is not enough to introduce conventional GPUs or embedded FPGA (eFPGA) processing units in SoC or ASIC designs. Hardware accelerators such as "embedded super accelerators" or "hardware accelerators with multi-level accelerators" are needed. At the 2019 Embedded World Exhibition, analysts at Huaxing Wanbang discovered these two types of artificial intelligence or smart artifacts.

Acceleration and re-acceleration will be a basic feature of embedded intelligent chips, and super neural network acceleration or multi-level hardware acceleration will become standard.

The PowerVR Series3NX series of neural network accelerators (NNA) released by Imagination Technologies at the end of last year is a series of embedded super accelerator IPs that can support embedded or edge devices in the AEIMR industry to achieve extremely efficient reasoning. The PowerVR AX3125, which has the smallest area and the lowest power consumption in this series of IPs, can provide 0.6 trillion computing operations per second (TOPS), and the highest performance PowerVR AX3595 can achieve an amazing 10 TOPS; by integrating multiple processing cores, the PowerVR Series3NX can also achieve 20, 40, 80 or even 160 TOPS, so the AI ​​chips built with this series of IPs can support a variety of intelligent applications from smart door locks with cameras to self-driving cars.

Imagination Technologies showcases its advanced GPU and NNA IP portfolio at Embedded World

Smart chips using "hardware accelerators with multi-level accelerators" IP or technology will play an important role in the AEIMR industry. The archetype of such accelerators is the embedded FPGA (eFPGA) that is "not just an FPGA". Take the fourth-generation Speedcore eFPGA IP recently launched by Achronix, the industry's leading eFPGA provider, for example. Compared with the previous generation of Speedcore eFPGA products, this product has a 60% performance increase, 50% power reduction, and 65% chip area reduction; more importantly, its new machine learning processor (MLP), known as the "accelerator on the accelerator", provides 300% higher performance for artificial intelligence/machine learning applications.

Achronix's Speedcore Gen 4 eFPGA multi-tier accelerator IP can be used in a variety of markets

Compared with cloud-based artificial intelligence technology, the intelligent technology of the AEIMR industry is still an embedded intelligent technology, which means that the resources of the entire system are still very limited, and the main learning and inference operations must be completed independently by these processors. Therefore, when developing intelligent chips for these industry applications, in addition to choosing a reasonable chip architecture, another important factor that can quickly implement chip design is to choose an embedded super accelerator such as PowerVR 3NX, or a hardware accelerator with multi-level acceleration such as Speedcore Gen4 eFPGA.

Make good chips for China

In addition to achieving the highest computing power under various constraints, the smart chips used in the five industries of AEIMR also have one thing in common: they need to face the reliability, internal security and protection capabilities of various complex environments in industry applications. This requires these chips to pass the corresponding industrial or automotive grade specification certification, and also requires the use of newer technologies to achieve internal security and protection against external attacks.

New tools and IP are needed to solve the system-level complexity faced by smart chips

The reliability required for AEIMR industry applications is the advantage of several traditional automotive chip providers, who have rich experience and knowledge accumulation in this field. But this is not an insurmountable difficulty. In addition to introducing corresponding mechanisms and designs in chip design and manufacturing, the most important thing is to have the necessary patience to screen and tune the relevant chips. In recent years, many domestic chip design companies have already acquired this ability and patience. Some of them have even passed the automotive-grade certification for mixed-signal SoCs including temperature-sensitive RF units.

For example, Taidou Microelectronics Technology Co., Ltd., a leading domestic provider of satellite positioning and navigation technology, chips and solutions, announced in November last year that its RF baseband integrated satellite positioning chip TD1030 passed the AEC-Q100 Grade2 certification. It not only became the first domestic RF baseband integrated Beidou/GNSS satellite positioning chip manufacturer to pass this certification, but TD1030 also obtained the highest level of automotive certification for similar chips. Its operating temperature range of -40℃~+105℃ can support automobile manufacturers, Tier-1 suppliers and intelligent application solution providers to develop more reliable in-vehicle navigation systems.

Taidou Microelectronics' TD1030-Q3003AB Beidou/GNSS RF baseband integrated chip was the first in China to pass the AEC-Q100 Grade 2 certification

In addition to reliability, chip internal security and external defense capabilities are also important indicators required by the AEIMR industry. Many technologies developed for this purpose in the traditional embedded control era can be used as reference, such as the lock-step technology in automotive controllers, that is, the two control cores in the same chip run synchronously, which can ensure that the processor is in normal working condition at any time.

But these are far from enough. Some new demands brought about by intelligence are changing the definition of chip security and protection, such as full life cycle monitoring of heterogeneous multi-processors on chips. If Chinese chip design companies want to surpass the accumulation of traditional automotive and industrial chip providers in this field, they can only use the latest technology to quickly pass these technical barriers, complete their own high-security and protection chips, and have capabilities such as full life cycle monitoring that traditional chips do not have, so as to design truly good Chinese chips.

UltraSoC, a provider of critical system safety and security protection headquartered in Cambridge, UK, announced the launch of lockstep monitor IP at the end of last year. Chip design companies only need to introduce this IP and solution in their designs to support the implementation of lockstep functions, thereby monitoring in real time whether the SoC processor is running reliably, safely and without hidden dangers. At this year's Embedded World Congress, the company also announced a major expansion. The new function supports SoC designers to build on-chip monitoring and analysis systems with up to 65,000 components, thus providing seamless support for systems with thousands of processors.

UltraSoC demonstrates advanced multicore debug technology at RISC-V booth at Embedded World 2019

The new UltraSoC architecture can efficiently monitor the countless internal building blocks used to construct the most complex SoC products and analyze the impact of their interactions on system-level behavior, which can play a huge role in realizing the artificial intelligence and machine learning technologies required in cutting-edge applications such as driverless cars. Esperanto, which realizes the most energy-efficient high-performance computing systems for artificial intelligence, machine learning and other emerging applications, has placed a thousand RISC-V processors and artificial intelligence/machine learning (AI/ML) accelerators on the same chip. UltraSoC's IP products support the monitoring, analysis and debugging functions of this large-scale processing chip. [page]

To make the main control chip, we must build a large ecosystem

When talking about intelligent applications today, in addition to the chip performance indicators such as how many TOPS, there are two groups of words that often appear at the same time. One group is software and algorithms, which is one of the focuses of the existing intelligent chip development companies; the other group is the industrial ecology related to the application. Behind these words closely related to intelligence are also ecosystems. In addition, each industry in AEIMR has its own unique ecology. A successful main control chip developer must be an operator or organizer of a large ecology, which is also the key to examining whether a company can become a unicorn in the next stage.

When it comes to software and algorithm ecology, traditional artificial intelligence such as AlphaGo relies on cloud computing, because cloud service companies have a large number of excellent software talents and computing power. However, when we talk about embedded intelligent hardware for the AEIMR industry, we also quote Alan Kay's famous saying, but without software, intelligent chips still cannot work. In the face of the intelligent needs of the AEIMR industry, not every chip company has to gather a large team of engineers proficient in Python or Tensor to engage in vertical integration of the industry ecology. The division of labor in the industrial chain and horizontal cooperation will also become a way of development.

At the Electronica Industrial Vision Exhibition held in Shanghai not long ago, senior analysts at Huaxing Wanbang also saw some local visual post-processing algorithm providers who were willing to cooperate with chip and platform developers to provide complete video solutions. At the same time, the S company mentioned at the beginning of this report is a potential international partner. Of course, if the path of industrial integration is taken, then the smart chip must be open enough, and the selection of software and hardware platforms is very important.

We see the rise of more and more local professional algorithm providers, providing ecological support for the development of intelligence and smart chips.

For software platforms, in actual operation, it is necessary to analyze the development toolkit SDK of the IP or chip provider, which should not only help developers create custom neural network applications based on the accelerator, but also provide support for open specifications such as SYCL. The SYCL specification developed by the Khronos group supports developers to use C++ language to develop any application supported by OpenCL or a specific neural network API.

Taking the PowerVR Series 3NX neural network accelerator from Imagination Technologies mentioned earlier as an example, this NNA IP supports both customized applications and mapping each SYCL function library to its dedicated IMG DNN API library, thus providing a non-proprietary open standard development environment.

The choice of hardware ecology is also very important. If there is a hardware evolution process for developing chips for intelligent applications in the AEIMR industry, not only can the development speed be greatly accelerated, but also a lot of costs can be saved. Therefore, it is best to look for manufacturers that can provide hardware flexibility. Take the aforementioned Achronix as an example. The company provides Speedster high-density, high-capacity FPGA chips, as well as system-level packaging (SiP) FPGA bare die cores and Speedchip packaging services, as well as Speedcore eFPGA IP products.

From using independent FPGA chips for development, to using FPGA bare chips for SiP packaged chips, and then using eFPGA for fully integrated SoC or ASIC, it is also a good model. Hardware ecosystem support is very important.

Since these products all use the same FPGA architecture and are supported by Achronix's ACE development tools, smart chip developers can use Speedster FPGA chips for software and algorithm development immediately when starting a project. If the AEIMR industry has other hardware requirements such as dedicated processors or interface driver chips, the applications on Speedster can be migrated to FPGA bare chips and packaged together with these chips as SiP chips. During these processes, you can ship while using Speedcore eFPGA IP to develop your own SoC hardware. Once the development is completed and the market is in volume, you can launch high-performance and low-power smart chip SoC or ASIC.

Pay attention to the trend of AIoT intelligent network integration

In today's intelligent trend, different industries and fields are using different languages ​​to describe the same trend. Some industry leaders are pushing for the integration of artificial intelligence and the Internet of Things, namely AI + IoT = AIoT; in addition, in many occasions, intelligence is linked to topics such as V2X, industrial Internet of Things and 5G, and intelligent networking has become the same concept. The development of embedded smart chips and the integration of innovative Internet of Things connection technology have clearly become an important trend in the development of intelligence.

In the development of autonomous vehicles, the connection between internal sensors and control data transmission lines, as well as the communication between the vehicle and the external environment are also very important.

Of course, for embedded smart chip design companies, there is currently no need to integrate communication circuits using RF mixed signal technology into the main control chip. However, for system-level connection design, considering connection is no longer just a simple technical decision on selecting data rate. Innovation in connection technology is also affecting the development of intelligent applications. At the same time, various connection standard entities and manufacturers are actively expanding their own ecosystems.

For example, the all-solid-state contactless connector technology developed by Keyssa uses ultra-high frequency wireless transmission at a rate of 6Gbps to replace traditional connectors that need to be plugged and unplugged, which can greatly optimize the connection of robots, automobiles, internal connections of circuit boards, and medical electronics, as well as the performance of smart chips. At the same time, because the connector and the corresponding openings are eliminated, the system reliability is greatly improved, the design flexibility is improved, and the manufacturing cost is reduced.

Fresco Logic's F-One technology provides a new way of interconnection for the emerging AIoT. F-One technology includes a series of highly integrated aggregation control chips that can aggregate a series of communication protocols in the same F-One serial channel in a flexible and dynamic way. It can be widely used in multi-sensor environments, industrial Internet of Things, network security, automotive wiring harness reduction and mobile device accessories.

In the five major industries of AEIMR, connections are very complex and diverse, so the needs of vertical industries must be considered, such as the CAN bus of automobiles and various existing protocols in the field of industrial automation, as well as the connection needs for intelligent processing of data transmission, low latency and high speed, as well as cost and ease of deployment. More importantly, various connection protocol entities are creating a larger ecosystem around their connection technology. For example, 5G technology brings not only speed, but also new value and a new 5G ecosystem.

Another factor driving the hot intelligent network or AIoT today is that IoT connection technology is bringing more new value to intelligent applications, such as providing time and location tags for intelligently processed data to support many business innovations. We can use Semtech's LoRa technology for analysis. LoRa has become a leading technology in the field of low-power wide area networks (LPWAN) with its low power consumption, long distance and easy deployment, and has been widely used. It is one of the fastest-growing IoT technologies in recent years.

LoRa is the first to enter the new stage of connection + data service in the LPWAN field, and also represents the new trend of intelligent networking.

At this year's Embedded World Congress, Semtech demonstrated LoRa's applications in smart cities, smart transportation, and smart homes. Shortly after the conference, the company launched the LoRa Cloud™ geolocation service, a new cloud-based geolocation service compatible with the LoRaWAN protocol and most network servers. It can be accessed through Cloud API calls, simplifying geolocation access for all LoRaWAN-based devices and enabling popular and intelligent applications.

Innovations in the field of IoT connection technology are also emerging. At this year's Embedded World Congress, many domestic companies exhibited their IoT communication chips and modules, and we can see the comprehensive rise of Chinese communication module manufacturers. We also visited Silicon Labs, a chip provider that can provide the most comprehensive short-range wireless communication protocol support, at this conference. The company's products support Bluetooth low energy, Wi-Fi, Z-Wave, Zigbee/Thread and private protocols, as well as wireless SoC chips that can support multiple communication protocols on the same chip.

Silicon Labs showcased the most comprehensive wireless connectivity chip portfolio at the 2019 Embedded World Conference and is a great ecosystem partner for smart chip companies

Innovations from leading manufacturers such as Silicon Labs have further shortened the distance between the Internet of Things and embedded smart chips. Last year, the company launched a new Wi-Fi product portfolio, the WF200 transceiver and WFM200 module, which reduced the power consumption of Wi-Fi connections by half, supporting the further popularization of intelligent applications including IP security cameras and points of sale (PoS). In January this year, the company released new Bluetooth software for its Wireless Gecko product portfolio, taking the lead in using the direction finding function in the Bluetooth core specification 5.1 to provide positioning solutions with an accuracy of less than 1 meter for indoor positioning and mobile positioning applications. Z-Wave is the company's exclusive protocol and is popular among customers who want to build private mesh networks.

Conclusion

On March 5, Premier Li Keqiang delivered the Government Work Report at the Second Session of the 13th National People's Congress. When talking about the government's work tasks in 2019, the report included "promoting the transformation and upgrading of traditional industries" in the key tasks. The report pointed out: Focusing on promoting high-quality development of the manufacturing industry, strengthening industrial foundations and technological innovation capabilities, promoting the integrated development of advanced manufacturing and modern service industries, and accelerating the construction of a manufacturing power. Build an industrial Internet platform, expand "smart +", and empower the transformation and upgrading of the manufacturing industry.

Embedded World 2019 venue - Nuremberg Exhibition Center

At the same time, many friends in the industry are talking about the next step in the development of artificial intelligence, and the core elements are rapid application and implementation. Therefore, the Huaxing Wanbang analyst team believes that the five major industries of AEIMR will be the key breakthroughs in the next step of intelligent development, and they also provide domestic chip design companies with the opportunity to start at the same time as their overseas counterparts. Starting from the innovation of smart chip architecture, making good Chinese chips, expanding the ecosystem to make master control, and paying attention to the trend of AIoT intelligent network integration, there will be opportunities to achieve rapid development in the emerging market of smart chips.