Yang Yuxin, Black Sesame: Present and Future of Domestic Computing Chips

"When it comes to smart driving, traditional car chip manufacturers are slowly falling behind. I asked these traditional automobile chip manufacturers. What do you think of these powerful chips? We give up strategically. "

On September 14, Black Sesame Technology held a technology open day in Beijing. At the meeting, Yang Yuxin, chief marketing officer of Black Sesame Intelligence, made several important points on the current development of automotive chips in China:

He believes that in the future, China and the United States will become the most important parties in the automobile chip game; In addition, 2025 will be a key node, and chip manufacturers who did not get on the bus before 2025 will lose in this round of automobile chip development. In addition, Yang Yuxin also told Automotive Industry and Commerce that in the future, only three or two companies in China will manufacture computing power in automotive chips.

Under the multiple factors of intelligent wave impacting the automobile industry and geopolitics affecting the chip industry, as a new domestic automobile chip enterprise, black sesame technology has attracted more and more attention from the industry, media and investors. Nowadays, the self-driving chip with large computing power has become the "standard" of intelligent electric vehicles, and the A 1000 chip of Black Sesame Technology has been on the bus.

The following is a summary of the speech.

Expanding applications define the entire intelligence foundation. Everyone likes to talk about the base, and we also talk about the base. Looking back at several major stages of the development of the whole electronic industry, the PC era is mainly characterized by large-scale computing, advanced and miniaturization, and PC has appeared. The architecture of the whole chip is like CPU+GPU, and the basic intelligence is based on the chip.

In addition to the miniaturization of complex computing, the image function of smart phones has also become very important, so the smart base of mobile phone chips, like CPU+GPU+ image processing, has become a very critical and indispensable part. Automobile is a master, integrating many technologies accumulated by the electronics industry in the past decades. The whole application data platform, innovative electronic and electrical architecture, including a powerful basic computing platform, and then mature and reliable components, software can be upgraded, and the requirements for the whole intelligent base and chip are becoming more and more complicated. CPU+GPU+ISP+NPU neural network accelerator, plus a lot of signal processing, the chip will become more and more complex. The chip made now may be the largest integrated circuit in China, and its complexity is directly proportional to its size. The chip we have now is about 100 mm, and it will be several hundred mm in the future. The chip is becoming more and more complex, and the scale of integrated circuits is becoming more and more complex, which is also to meet the requirements of smart cars for function and performance.

"China and the United States will become the main players in the future big computing chip game."

In fact, with the development of intelligent drive technology and upstream core supply chain, people have been very concerned about the lack of core in recent years, and the most important thing is chips like MCU. Where are the suppliers of these traditional automobile chips mainly concentrated? It is concentrated in Europe, America, Japan, Renesas in Japan, Infineon in Europe and TI in the United States. Their growth is inseparable from the growth of these European, American and Japanese automobile companies. The development of these enterprises has sold cars all over the world and cultivated the upstream core supply chain, especially the core chips.

When some parts with relatively concentrated functions appear in cars, there is a demand for SoC chips, such as on-board parking systems. A few years ago, buying a big-screen car was enough. Behind this big screen is an application processor, including the SoC of IDAS sold by Mobileye. Indeed, foreign manufacturers have their inertia and occupy a dominant position.

But when it comes to intelligent driving, we see that this core has begun to change. On the one hand, these traditional automobile chip manufacturers are slowly falling behind. I asked these traditional automobile chip manufacturers, what do you think of these powerful chips? We give up strategically. Because of several aspects: first, they are far away from the market, and second, long-term industry inertia will restrict their sensitivity to innovation to some extent.

Looking at the future of autonomous driving or intelligent driving, what will the global chip pattern become?

On the one hand, the United States remains a highland of global innovation. The United States is a very important force, and this force has been iterative: now the chip giants in the non-automotive field in the United States are the fastest. For example, NVIDIA doesn't make cars, Qualcomm doesn't, and Intel buys Mobileye. But they saw the importance of high-performance computing in the future process of automobile intelligence, so they began to enter this market, and what they were good at broke out in this market.

Some domestic companies represented by Black Sesame Intelligence have grown into unicorn enterprises. On the one hand, they are constantly struggling to survive, on the other hand, they are constantly embodying our innovations in the products of actual customers. I believe that in the future, China and the United States will become the most important players in intelligent driving.

The development of the next generation electronic and electrical architecture will not happen overnight. Although many people are talking about the concept of central computing, the evolution from domain control architecture to central computing will have different stages, including the integration between domains. I believe that there will be multi-domain fusion chips, not in one step. Finally, there will be a huge chip to integrate all the functions, but it will start to integrate one by one. This multi-domain integrated chip, we saw the chip company begin to lay out.

We are optimistic that after 2025, the central computing architecture will come. I feel that if China enterprises do well, there is really hope to share the world with these enterprises in the United States in the field of smart cars in the future. This is an optimistic estimate. On the one hand, thanks to its position in the future automobile industry of China automobile enterprises, it will definitely be a very important position in this field. Ten or fifteen years later, half of the top ten auto companies in the world are from China. Our growth, our gradual development in the industry, also depends on the development of domestic car companies.

NVIDIA restrictions in the United States have a great influence on "ideology"

First of all, American restrictions on NVIDIA do not involve automobile products. But where is its impact on the industry? Everyone suddenly found a knife on their heads, although it is possible that the person who raised the knife said that he would never put it down in his life.

In the past, it was basically intensive to receive a large number of calls from customers, whether it was to step up cooperation and strengthen the requirements of project promotion, or just to contact customers who had not officially started cooperation before. We found that car companies are asking group leaders about the localization and replacement process of large computing chips or core computing chips. Even if there is a possibility of 10%, I will make a backup of 100%. Although we have no qualitative influence on the automobile industry in the short term. However, I believe that for domestic chips, not only chips, but also the upgrading speed of the whole smart car will be accelerated.

You never know when there will be more or less restrictions. In the past, it was judged that the United States would not necessarily start work in the civilian market, but in fact, the restrictions on chips involved the civilian market, which was a bit unexpected. So I don't think this will affect the automobile industry as a whole in the short term. But ideologically, I think it is still quite big. Everyone thinks that it is still necessary to find backups of various schemes in China as soon as possible.

The Chip Act includes EDA tools that were previously restricted. The impact of the Chip Act on China's semiconductor industry, the United States hopes to bring the global semiconductor industry back to the United States in the future, and the cost may increase, so it needs to make up the money. Although political correctness is important, it is also necessary to settle accounts. It will not affect China in the short term, but it will make the competition between China technology and American technology more important. For example, EDA tools below 7 nm have an impact, and more advanced processes below 5 nm have no impact. When China's head or big chip industry is turning to more advanced technology, does it replace China's own system? This is a question mark that everyone needs to type. In addition, many other chip bills are still filed in France, which has a good influence, but on the whole, it has sounded the alarm for the domestic automobile chip semiconductor industry, and the restrictions will only increase.

How long can the domestic automobile chip market be hot?

Everyone pays attention to many companies that make automobile chips. Car chips are difficult to make, and chip design used to go around.

Because, first, the investment is large. The cost of IP, talents and R&D is much higher than that of consumer-grade and industrial-grade chips. The second cycle is long, from the research and development cycle of the chip itself to the cycle of customer certification, and the cycle of developing products after customer certification is very long. Enterprises need to have sufficient capital reserves, lasting endurance and strong strategic firmness. In addition, automobile chips are not like other industrial or consumer markets. The time window of the latter is relatively loose. As long as your product has certain advantages, you can find customers and start selling. But car chips are different, because the car industry is very conservative and unwilling to use the products of new suppliers, not to mention that you are a startup company.

However, I think the opportunity allocated to local auto chip companies is very obvious, and there may not even be a chance before. Because of the iteration of new technologies, there are fewer and fewer choices. On the one hand, car companies have to choose products from more innovative technology companies. On the other hand, in order to ensure that there is no shortage of chips in the future, it is necessary to cultivate local suppliers.

Although automobile chips are booming now, I think it's crucial to get on the bus in 2025. I think 2025 will be a very important time node. By that time, there will be no local suppliers for most key automobile chips. The so-called exhaustion is not a batch boarding or a mature large-scale application. This means that the chip is already on the market and customers are already using it or developing products.

Because when car companies cultivate their own supply chain system, when your key supplier already has a domestic supplier, its motivation to replace another domestic supplier will decrease geometrically. For car companies, it takes a lot of manpower and material resources to cultivate a mature supplier and start bulk shipment, especially for our complex large-scale chip supplier. It took him a long time to become a supplier with manpower and material resources. If there is no special reason, there is basically no motivation to change.

So everyone must run quickly. If we can get on the bus and enter the supply chain system of automobile manufacturers before 2025, there will be many opportunities in the future. If you can't get on the bus in 2025, and everyone else in your field gets on the bus, the chances are basically very small.

In fact, we are lucky. We entered this market earlier and have been doing it firmly. We have chips now, and then gradually enter the stage of mass production. But I believe that many domestic chip companies will come out in this field, but time is precious.

Black Sesame should be the global leader in autonomous computing chips.

Introduce black sesame intelligence. We are positioned as a global leader in autonomous computing chips. We hope to empower the industry to form different vehicle and road solutions through our leading chip technology, including the ability to combine software and hardware algorithms based on our chip technology. We have invested about 500 million dollars, and with the tide of this era, we can have enough funds for technical iteration.

Black Sesame was founded on 20 16, and its core team is basically from Tsinghua, from two fields: automobile and chip. 20 18 was commercialized and the first generation of chips was released in 2020, which verified our technology and products. At the same time, our plan has been working closely with car companies, and 202 1 and 2022 are both stages of rapid development. 202 1, many ecological cooperation began to land, our roadside scheme began mass production, the latest A 1000 series chips were released, and the C round of financing was completed. Xiaomi investors voted for us. In 2022, the business continued to be promoted and the products began to be mass-produced.

Our release with JAC is actually the point we set on 202 1, because we respect the wishes of car companies very much. If we don't decide to cooperate, we will promote the car company. Car companies will be willing to release PR together after they decide to recognize your ability for a period of time. We released the cooperation between Jianghuai this year, and there should be more cooperation between car companies this month, as well as the C+ round of financing announced some time ago.

Our team is actually very distinctive. I have seen many experts in this field in China, either in the automobile industry or in the chip industry. Few experts like us can combine the two fields of chip circle and automobile circle. In the field of chips, we have more than 20 years of experience in chip design. Our team 10 has developed hundreds of SoC slides and IP designs over the years, and developed the first image processing IP. The automobile team also has more than 20 years of experience in the automobile industry, with previous experience in mass production of automobiles, and at the same time, it has achieved 5 years and 3,000 times business growth in other automobile product fields.