'Tragedy' Li Zaisheng took over from Samsung | Overseas investment in new stocks

1.ST China Summit: In 2017, STM32 MCU shipments exceeded one billion units, and the Chinese market made a huge contribution. 2. After Li Zaiyi took over Samsung informally, the overseas newly-created stocks all suffered losses; 3. Intel introduced FPGA technology for servers. ; 4. Intel and NVIDIA change the market strategy AMD or can benefit in the consumer market; 5. Moore's Law ends up plagued Intel NVIDIA anti-benefits

1.ST China Summit: In 2017, STM32 MCU shipments exceeded 1 billion, and the Chinese market has contributed a lot;

Set micro network April 28 news

There is no doubt that the Internet of Things era has arrived! And the data shows that STMicroelectronics has become one of the beneficiaries.

From April 25th to 26th, STMicroelectronics (hereinafter referred to as ST) held the third 'STM32 China Summit' in Shenzhen. The theme of the summit was 'Wireless Connection and Cloud Access'. The main topics discussed today include sensing. With data processing, security, smart industry, etc. Including Microsoft, Ali cloud, wit cloud and many other partners brought with the ST cooperation program.

In the specific arrangements for the meeting, April 25th will be a high-level keynote speech by STMicroelectronics and the STM32 Ecosystem Partner for Sensing and Data Processing, Wireless Connectivity and Cloud Access, and Security and Smart Industry Forum; and will be launched for the first time on April 26. Shanda's STM32 Fan Carnival line party, which allows participants to participate in seminars, observe the latest technologies and exhibits of ST and its partners.

In the first day of the conference speech, the ST executives shared the achievements of the STM32 in the past year. As one of the world's largest semiconductor companies, ST has a wide range of products, sensors, power devices, automotive products and embedded processor solutions. The programme has played an important role in the ecosystem of the Internet of Things. Among them, MCU is one of the most important businesses. Official data shows that in 2017, ST occupied approximately 19% of the general microcontroller market, and the company has more than 800 STM32s. Products, more than 50,000 customers.

In 2017, the company shipped a total of 1 billion STM32 global shipments, equivalent to selling 32 SCMs per second.

Among them, the share of the ST general-purpose microcontrollers in the Asia-Pacific market accounted for more than half of the total. Especially in China, according to research firm iSuppli data, ST is the second-ranking standard microcontroller manufacturer in China in 2017, accounting for 16% of the market. And has the most extensive Cortex®-M microcontroller product line on the market, more than 800 STM32 products. Since 2007, ST MCU in China achieved a compound annual growth rate of 27%.

ST officials also expect that the company's MCU market share in China will exceed 20% by 2020.

Currently, ST is developing a new generation of STM32 wireless products, STM32WB, which uses dual-core (CMO+ core responsible for the protocol stack), with dual modems (BLE and 802.15.4), with Cortex M0 zone security features, and facilitates user-based CM4 hard core. Development and Application. The product is expected to be available in the second half of 2018.

In terms of industrial chain ecology, ST can build smart homes, smart cities and industrial Internet of Things from a chip to a module, cloud, base station, network management and security, etc. At the same time, ST also relies on a wide range of product lines. Provides a powerful sensor portfolio for wireless connectivity.

Cao Jindong, Director of Markets and Applications of STMicroelectronics China's Microcontroller Division, said that the ecosystem is the key to success in the future. It has driven 2017 ST MCU (8-bit + 32-bit) sales to 2 billion. In the future, China will continue to be promoted. The close cooperation. At the same time, ST also developed STM32 talent development plan. Including TTT plan, excellent course, STM32 design competition, to help China's education development.

2. After Li Zaiyi took over Samsung informally, the overseas newly-created stocks had full losses;

After Li Zaijun, the vice president of Samsung Electronics, informally overseen the group in 2014, he actively promoted the global new investment plan. However, these operators suffered a total loss in 2017. Although the short-term results are not satisfactory, the new entrepreneurs Whether it will be possible to achieve greater synergy with Samsung in the future remains time-tested. A number of Korean media reports have pointed out that according to Samsung’s 2014 to 2017 business report, it revealed that 14 Samsung’s investment overseas operators’ operating status did not The industry has made a beautiful report card and all losses have occurred. It is worth noting that 2014 was exactly the time when Li Zaijun took over the job and started to invest in overseas start-up companies. Li Zaiyu’s investment insights have attracted attention from all walks of life. Among the industry players, the biggest loss was in the intelligent doorbell manufacturer Bot Home Automation, with a net loss of 154 billion won (approximately US$140 million). Soundhound, which produces smart audio, also suffered a loss of 22.4 billion won, coupled with high-speed data transmission provider Keyssa, and laborers. Sense of the new (AI) entrepreneurs Graphcore, chip maker Inida, Internet of Things (IoT) new Afero and other 6 companies, net loss of more than 10 billion won. Among the 14 companies, Samsung holds shares ranging from 0.89% to 17.88%. The amount of investment is between 1.02 billion won and 7.06 billion won. The industry believes that Li Zaiyu seems to be over-investing. For Samsung, investment in new entrepreneurs can be diversified. The latest technology, to make up for the lack of self-developed technology, and to bring into full effect, so Li Zaijun has been actively looking for potential global players since 2014. Generally speaking, large companies invest in new ventures in order to enhance the technology of related businesses and use the current situation. Look, it shows that Samsung's cooperation with these companies is still not very effective. However, there are other opinions that it takes a lot of time for new start-up companies to enter a stable growth stage. Therefore, these investment cases should be observed from a long-term perspective, as pointed out by industry relations sources. Many large companies invest in unique technologies developed by new start-ups, but the technology or business can start to generate profits, and the time can be long or short. However, AI-related companies such as Soundhound have considerable potential. Soundhound In addition to Samsung Capital, it has also received investment from other South Korean companies. Although investment record for new start-up companies needs to be strengthened in 2017, Samsung still does not reduce investment speed. 2018 Earlier announced the investment in Israel’s newly identified voice recognition of US$4.6 million. The related technology is expected to be used to improve the maturity and completeness of Bixby technology. In the future, whether Li’s investment perspective and business practices can lead Samsung to a new peak remains to be seen. Time verification DIGITIMES

3. Intel introduces FPGA technology for servers;

Intel (Intel) announced that its OEMs will launch servers equipped with Arria 10 GX FPGAs for data centers. According to the review, the move will help the company create a breakthrough for the FPGA market in the server market, and once successful, it will be for others. Opponents set up obstacles. According to EEJournal, after Intel bought the Altera for several decades that dominated the server market, the outside world predicted that the former would intentionally bring about major changes to the development of data center computing, that is, the introduction of FPGA-based accelerators. Mainstream. Nowadays, Intel announced that OEMs including Dell and Fujitsu will launch programmable accelerator cards (PAC) servers preloaded with Arria 10 GX FPGAs. Earlier Intel also Said that the Acceleration Stack software development stream enables third parties to build maximized accelerators for Intel Xeon CPUs, and announced third-party financial acceleration, data analysis, cyber security, genomics and video transcoding, and artificial intelligence (AI). Applying Sales to Accelerate IP Ecosystems. Comments pointed out that from the implementation of financial risk analysis and database key workers In terms of quantity, the results are quite persuasive. Take financial risk analysis as an example. Compared to traditional Spark implementations, each symbol-based algorithm accelerates up to 8.5 times, and the simulation time is doubled. In database acceleration, Intel claims that Data analysis is faster than 20 times faster, data storage is more than 2 times faster and storage compression is more than 3 times. In the financial risk market, performance is of utmost importance because Terabyte (TB) grade data must be analyzed immediately and analyzed more quickly The return is quite large, because options trading, insurance and regulatory compliance will spur the anticipatory risk analysis market to multiply in the next five years. Therefore, Intel financial analysis partner Levyx has launched and tested a platform using Intel products. In addition, it is expected to accelerate from many companies. The application will be more, and the strength of the relevant ecosystem will become the key to the success of the new server. The comments pointed out that due to the use of low-cost Arria components, the FPGA accelerator will not have a major impact on the overall cost of the server, but for Companies that already have matching applications will see even greater benefits. With FPGA pre-maximized applications, most vendors will Significant benefits can be gained in performance improvement and power saving. This will also bring new developments to Intel's Programmable Solutions Group (PSG). FPGA companies have become accustomed to competing server value models. Engineers have chosen to design FPGAs for their systems. Previously, Fmax, number of LUTs, MAC size and performance, design tool execution time, and a host of other considerations were introduced. In the past, FPGAs were dedicated components that were often used by experts, and now many companies will be able to see large-scale purchases of Arria FPGAs, equivalent to servers. The FPGA shadow will be commonplace within the industry. The commentary believes that once FPGAs are massively added to the server, it will provide a good environment for third-party developers of high-value accelerators based on Intel Acceleration Stack, and once a large number of application developers make good use of the Acceleration Stack, it will allow Intel FPGAs have the potential to become x86 in accelerators. DIGITIMES

4. Intel and NVIDIA change their marketing strategies AMD may benefit from the consumer market;

Intel is implementing a strategy that was developed in 2016 with a focus on PC-to-Internet (IoT), data-centric markets such as cloud computing and FPGAs, and NVIDIA focusing on artificial intelligence (AI). Autonomous vehicle and data center applications; The new revenue sources of the two companies are already equivalent to the sales of consumer PCs. As major competitors change their market strategies, diversify their development, and move to a more profitable computing market. , AMD has the opportunity to profit in the consumer computing market. According to Tom's Hardware report, there have been some major events in the industry between 2017 and 2018, including the Meltdown and Spectre chip security breaches that have the biggest impact on Intel; AMD Radeon Technologies Group executives, and announced plans to return to the high-end independent GPU market; NVIDIA became the second IC design company to enter the top ten chip sales list. Tom's Hardware believes that the current AMD should be prepared, Competitors profit from the consumer computing market when they are distracted. Unlike their major competitors, AMD’s target market still focuses on half of the consumer PCs and game consoles. Plans to contribute most of the revenue. AMD continued to focus on rebuilding the competitiveness of x86 architecture processors over the past few years. It was surprisingly successful. Supermicro seems to be gaining momentum from the Ryzen processor, sensing Intel threatened by Intel. The corresponding products will also be introduced, but AMD's new Ryzen 7 2700X will be even more competitive than its predecessor, and AMD's Ryzen will also greatly simplify the cooperation with the platform. Its platform design can provide the same for all compatible processors. Performance. However, AMD has not yet been able to replicate the same successful model on the GPU. However, in order to gain a dominant position in the consumer market, AMD needs to strengthen the competitiveness of the GPU. The latest rumors indicate that AMD is ready to release a Ryzen-class graphics card. , But it has been observed that it cannot be achieved in 2018. Although no new products have been introduced, the latest report shows that the cryptocurrency boom may contribute to AMD's market share improvement in GPUs. However, NVIDIA should be waiting for the right time to introduce new GPUs. , may expand the leading AMD. The situation favoring ultra micros is the friendly relationship with consumers, especially those who are loyal to their loved ones. Competitors are struggling to achieve a breakthrough in the data center market. In 2017, Vega's graphics architecture was introduced first, followed by e-sales retail. AMD has challenged NVIDIA's dominance in the AI ​​market when marketing Epyc and Vega Instinct. Some progress has been made so far. Intel is eager to counter Ryzen. NVIDIA has pushed the GeForce Partner Program (GPP). It has gone so far as to take a monopoly on the edge. The desktop processor launched by Intel in 2017 has not reached. Expected goal, so that AMD has a desperate opportunity. The product lineup is fragmented, including the rush to replace the 4-core processor with 6 cores; the positioning of the new X299 high-end desktop (HEDT) platform is unknown. , and the performance is entirely dependent on processor pairing; the new Coffee Lake mainstream desktop platform seems to cross three very different chip sets. It is said that it will bear the cost of developing a separate GPU, Intel will The graphics market is fully chasing AMD, including the computing market and gaming market where AMD performance is significantly weaker. The consumer market is less sensitive to power and efficiency, and users are more concerned with Intel. The initial generation of GPUs will be more tolerant. If Intel's independent GPUs are rumored to be introduced in 2020, they will be the only ones that have the opportunity to benefit from self-control. Fabless NVIDIA and AMD will be handed over to TSMC and GlobalFoundries respectively. The application of AI and deep learning in data centers is becoming increasingly important. Large-scale parallel processors such as GPUs perform far better than CPUs in processing related workloads, and the demand for processors in data centers has changed. Investing more heavily in designing GPUs that are more suitable for the relevant workloads poses a serious threat to Intel in the data center. In the case of stand-alone GPUs, the last GPU product that Super Series beats its competitors at the time of launch is Hawaii in 2013. The most successful design was Polaris in 2014, but it has entered the second update cycle by 2018. AMD's GPU product line is constantly re-using products with old designs and new designs but no competitive products, so NVIDIA has pure The technological advantage, in the independent GPU market is a major threat to the weak AMD. DIGITIMES

5. Moore's Law ends up plaguing Intel NVIDIA

Since its introduction in 1965, Moore's Law has been the guideline for the development of the global semiconductor industry for several decades. But gradually seeing weaknesses in these years has the potential to end. If so, Moore’s Law has followed Moore for decades. Intel’s law is bound to need a new strategy to continue the trend. In addition, it is thought that the end of Moore’s Law is good news for NVIDIA, but the CPU will not disappear or be eliminated. It will still exist, but it is not so amazing. According to Forbes, Bruce Pile, a SeekingAlpha columnist, points out that in the current global semiconductor industry, Intel is facing increasingly greater challenges and is gradually not seen as unbreakable. In the past, CPU played a major computing role in the PC era, so the clock rate and crystal density were very important. Since then, with the emergence of more meta-computing requirements, companies such as NVIDIA have invented graphics chips (GPUs). The architecture's new type of processor, with thousands of parallel cores, enables a greater number of simple tasks with less transistor density, making it ideal for use in In the field of play and other fields, the next shift is underway. GPUs are considered as computing devices. Nowadays, they are also suitable for emerging technologies such as artificial intelligence (AI). The graphics cards are also really suitable for a large number of parallel digital operations. The ability to linearly increase computing power, bypassing today's Moore's Law limit, is different from the case where the CPU's internal crystal increases by a factor of 1 when the performance is only 10%. Therefore, as NVIDIA CEO Huang Renxun had previously stated that Moore's Law is dead, the GPU will replace CPUs produced by Intel and other chip makers do not believe that the CPU will disappear in the future, and that PCs and other computing functions will still only have CPUs to provide computing needs in the future. However, these PC type functions will not be surprising to the industry. In part, it will form a stunning part of the industry. For example, if GPUs provide more computing needs in emerging technologies, it can be seen that development APIs convert the computing task to the type of parallel drawing operations performed by the GPU, or to the data. The center is seen to unload more computing tasks from the CPU as much as the GPU provides operations. Even this will help NVIDIA's GPU vendors continue to be new. The technology sector benefits, such as AI, data centers, automotive electronics, medical platforms, etc. However, Pile believes that Intel’s current size is still significant. For example, Intel’s market value is still nearly one time more than NVIDIA, and its sales performance is also better than any other publicly listed products. There are many chip companies and they still dominate the global CPU market. In addition, even if the 28- and 40-nanometer wafer process can produce useful chip products, the Intel 14nm wafer fab is still a competitive advantage, with any 28- and 40-nanometer process. Production of chips manufactured on a 14-nanometer process will help reduce the size, cost, speed, performance, and power consumption of the chips. In this case, if only the design can not afford the 14-nanometer process and only design Both 28 and 40nm chip companies are likely to become potential acquisition targets for Intel. In this case, Pile believes that with NVIDIA’s current market value of US$137 billion, Intel will find it difficult to buy NVIDIA. Currently NVIDIA and Samsung Electronics Cooperate to produce NVIDIA chips on Samsung's 14nm process, but whether Intel can produce a lower-priced, faster product than NVIDIA GPUs in the future is worth watching. And at NVIDI As A continues to grow, it is estimated that NVIDIA will be large enough for Intel to fail to acquire in a few years. DIGITIMES

2016 GoodChinaBrand | ICP: 12011751 | China Exports