Chips have always been a pain point for China's semiconductor industry, and it is also a time bomb buried between China and the United States. After ZTE was fined, the state has increased its support for domestic chips, supported by large funds, built IC industrial parks, and promoted project incubation. , can be said to be in full swing, and companies specializing in chip design are also emerging. However, as a reporter in the semiconductor industry, the author has found in many years of interviews that it is really difficult to engage in chip design. Technology is a major threshold, requiring many years of technical accumulation, and capital is a major threshold. The limelight is more keen to make quick money. Projects, such as chip design, are cautious about projects with a long payback period. Therefore, we see that many startups are more willing to start with applications. The drones, VR/AR, and smart cameras that were popular all over the world back then are good examples.
However, chips are not completely without opportunities. The drive of a hot topic will make a certain field rise rapidly. For example, artificial intelligence has led to the entrepreneurial tide of AI chips. Companies such as Shenjian Technology, Horizon, and Cambrian are also standing in IC entrepreneurship. , and has been favored by VCs. In the fifth roadshow of Anchuang Growth Camp, the author also met several chip startup companies, and their technical capabilities and team members were unanimously affirmed by the judges on site.
Last year, AI chip entrepreneurship was in a mess. Is there any chance to enter the market now? It is estimated that this is a question that many startups are thinking about. For start-up companies, only by keeping up with the hot spots can they get the attention of venture capital and get financing smoothly. Of course, if the team is good enough, it is another matter. For example, Shenzhen Dapu Microelectronics, which was conducting a roadshow in the Anchuang Growth Camp, chose a unique entrepreneurial project called a data storage processor chip (DPU).
Regarding why he chose this entrepreneurial point, Yang Yafei, CEO of Dapu Microelectronics, explained, "With the evolution of technology, the operation speed of the main control chip is getting faster and faster, and the storage capacity of the memory is also getting larger and larger, but the data processing needs Putting data into the cache from the memory for processing, the main control chip and memory have made rapid progress, but the intermediate storage management and control technology is still relatively backward. Judging from the current enterprise-level SSD market, Samsung, Western Digital, Intel In control, most data centers in China use these products. Our DPU has made great innovations, putting computing and many intelligent elements on the chip, realizing the integration of computing, storage and intelligence. Previously, computing It is separated from storage. Both storage and display are external devices of the processor. We believe that in the era of artificial intelligence and big data, storage and computing are integrated, so for the first time, some computing has been implemented on storage disks. Of course, we started very Be cautious, and do the most basic and general computing well. For example, data format algorithms, unstructured data query on disk, and data sorting are done on disk. We think this is a future revolution in the storage field."
In the selection of entrepreneurial projects, Yang Yuxin, co-founder of Anchuang Space, said, "We are looking for applications related to scenarios. Previously, artificial intelligence and the Internet of Things were basic technologies. Now, basic technologies are more integrated with actual scenarios. At the same time, we Looking at the development trend of basic technologies, last year we made AI chips and IoT chips, and this year they are all storage. Because you can see that in the process of technological development, the amount of data is increasing, and storage is a bottleneck, so we mainly Look at two trends, one is the trend of technology, the other is the trend of the industry.”
Selling IP or selling chips?
Does a startup company get financing if it is successful? In fact, getting financing is just the beginning of entrepreneurship. What they need to consider later is how to make the product launch and how to make a profit. Almost all startups will struggle between life and death. In front of chip entrepreneurs, there are basically two choices: selling IP, or selling chips by themselves. Arm can be described as smooth sailing in the field of IP licensing. It is a typical case of success. Is it applicable to entrepreneurs who are entering the market now? In this interview, I heard two voices from non-net reporters.
Yang Yuxin believes that from the perspective of venture capital, it is difficult to survive only by making IP, and he must make his own chips. He analyzed, "ARM's success is due to its emergence in a specific historical period. The early development cycle of chips is very long, and ARM's licensing fees range from hundreds of thousands of dollars to tens of millions of dollars. After authorization, each chip will have to wait for shipment. Two years, about a few cents per chip. IP becomes a chip to prove to customers that the chip can work. No one dares to use unverified IP. It takes about half a year to get verified. And IP authorization is not available to small companies. The licensing fee may only be hundreds of thousands or hundreds of thousands of dollars. The smaller the IP, the less the core will receive, and then wait for a year and a half to make the chip, and the start-up company may have already been dragged down. For example, AI chip companies, if they only do chips without algorithms and scenarios, will immediately encounter problems with the ecosystem, because they only do neural network acceleration and specific acceleration capabilities, they must develop enough algorithms and applications. I don’t think an entrepreneurial A company can launch an ecosystem, so it must have its own scene, its own algorithm, and its own chip. AI chips must be vertically integrated, and when there are scenes and algorithms, they can do it themselves, so it is not established for a startup company to only make IP, and make chips. At least if you sell it, you can receive money directly.”
As an entrepreneur, Dr. Yang Yafei shared another example. Dr. Yang worked for Qualcomm for eight years. As the world's largest smartphone chip manufacturer, Qualcomm began to develop neural network IP in 2010, and formed a team of 20 people. The goal at that time was to put it in the SOC of mobile phone chips. However, due to power consumption issues, this goal was not achieved until the team left in 2014 to establish Nervana Systems, which was later acquired by Intel for nearly $400 million. At present, artificial intelligence is still in a state of melee, and it is easy to copy and plagiarize only by doing algorithms. Therefore, entrepreneurs get financing and instinctively say that they need to protect themselves with chips. In the future, large chip companies may look for such IPs, which is equivalent to doing A module is used as a carrier, and Huawei HiSilicon uses third-party IP in this mode. As an entrepreneur, if you have sufficient funds, you can boldly design chips. If you have limited funds, you must take into account the implementation of the chip, and it may fall into your hands when you make it. Starting from IP is also a development path.
Dapu Microelectronics' strategy is to walk on two legs, Dr. Yang said, "DPU is still in the early stage, and we will not make a big step in each generation of products. For the consumer field, it already belongs to the Red Sea, and we fight very hard. The authorization method is used by consumer-grade SSD hard drives and notebook computer manufacturers, so that we will make a commission from it; for the enterprise-level market, we will make our own products. Qualcomm also makes money through two channels: one is products, the other is Patents. We have also been thinking about how to implement technology and convert it into value. We can see that we have a very strong business scenario, and now we are going to optimize this business scenario. We have thought of using machine learning and deep neural networks. innovation in this field.”
It is also a good destination for a startup company to be acquired by a large company
To become a unicorn in a field is the vision of most startups, but after all, there are usually only one or two unicorns. Ordinary startups also have to find their own way of survival, and can get a stable business to survive the survival period. It's what all companies strive for. In the chip field, acquisitions have been very frequent in the past few years, big fish eat small fish, and small fish eat shrimp. Is an acquired startup a good destination?
Yang Yuxin analyzed, "Large companies may not have enough motivation to do an innovative technology, and even small teams in the company may not be able to obtain resources, while startup companies are desperately trying to make a technological innovation, and when they do well, large companies will not be able to obtain resources. May choose, if you are optimistic about this market, you may acquire it, which is why small companies are bought. For a startup company, the founder is the biggest ceiling of the company, and the company can grow as strong as the founder's ability. Therefore, founders must clearly recognize their own upper limit, or the upper limit that they can break through. Entrepreneurship is actually related to life. It’s also a good place for a company to be acquired by a big company.”
Dr. Yang Yafei expressed his opinion based on his own experience, "There is a big difference between China and the United States in terms of corporate acquisitions. After a certain level of development, American companies will be acquired and merged. This is a high probability event, and after mergers and acquisitions All large companies are doing technology integration. After two years, they have begun to integrate the original system; China is very lacking in this regard. In many cases, domestic so-called acquisitions are based on revenue, and listed companies need to acquire Or the main purpose of holding the company is to consider on a consolidated basis, that is, to consider how much the company’s revenue will increase when it enters a new field, because the original business can no longer tell a new story, and a new theme needs to be found. There are many demanding companies. Relatively speaking, these large companies in the United States are more lenient. In fact, they do not require the acquired company to generate much revenue alone. I think Huawei will follow this path in the future, and will continue to look for good companies. of companies are acquired and then integrated within the division.”
Chip entrepreneurship is destined to be difficult. Fortunately, the government is strongly supporting it. There are also institutions such as Anchuang Growth Camp providing resource docking support for startups. It is a good time for domestic entrepreneurs. We will see more chip entrepreneurship in the future. The company grows.