The 5th Oceanpine Investment Salon was successfully held in Beijing

On September 23, the 5^th Oceanpine Investment Salon series "Opportunities and Challenges of AI Chips Investment" was successfully held in Beijing. As the cornerstone of the development of the AI industry, AI chips have developed rapidly in recent years, and many companies have deployed. Our salon invited three heavyweight keynotes speakers from the chip field — Mr. Yajun Li, Chairman & CEO of Sunic Capital and Director of Montage-Tech; Dr. Kai Yu, CEO of Horizon Robotics and Mr. Yalin Zhang, COO of Enflame Tech. The audience guests included Dr. Bo Zhu, the General Manager of Lam Research Capital Asia Pacific; Mr. Mingyao Wang, the General Manager & Managing Partner of LegandStar; Mr. Wu Quan, the Partner of Sunic Captial and several representatives of well-known leading investment institutions in the industry had a lively and in-depth discussion on the current topic of investment in the chip industry.
  Regarding the issue of the valuation of AI chip companies, Mr. Yajun Li，Chairman & CEO of Sunic Capital said that investment valuation can be divided into four stages. The first stage can be called the “market dream ratio”, that is the invested company has a reasonable development plan for its own company in the future. However, there is no profitability for the time being; the second stage is the “market-sales ratio”. If the company already has stable income, you can refer to the company’s market-sales ratio. The lower market-sales ratio, the greater value of the company’s investment; the third stage is the “price-to-earnings ratio”. When the company is already profitable, we will refer to the price-to-earnings ratio. Generally, the lower the price-to-earnings ratio, it means that investors can buy the stock at a relatively low price. If it is too high, it means the value of the stock is overvalued. Big. But this is not absolute. Different industries have different price-earnings ratios. Finally, look at the price-to-book ratio. When the market price is higher than the book value, the quality of the company’s assets is better and has development potential. On the contrary, the asset quality is poor and there is no development prospect.
 
Regarding the future investment in the semiconductor industry, from a big logic point of view, the trend of my country’s semiconductor industry can be based on the communications industry or the automobile industry 20 years ago. Both have lasted for 20 years. In the future, my country’s technological development will continue for 20 years. Integrated circuits are among them. The leader will continue for 20 years. As for the problem of market bubbles mentioned, market bubbles exist at all times, and we also need to coexist with market bubbles. In the future, market bubbles will exist for a long time. Finally, choose truly powerful products. In the future, 80% of my country’s deep-tech companies will eventually be eliminated, 10% can barely be maintained, and the remaining 10% are truly powerful companies. Therefore, it depends on whether you have the ability to do so. Find the truly valuable high-tech companies to achieve China’s Intel and Nvidia.
  Dr. Yu Kai shared the reasons for choosing to cut into the car chip. Internet companies have funds and resources and have a high probability of cutting in from cloud chips. In order to stagger the competition, Horizon chooses end-side computing chips. Yu Kai founded Baidu Autonomous Driving and knows the key points of the vehicle side. Unlike data centers, the vehicle chip has no heat dissipation system and the external operating environment is uncontrollable. Therefore, vehicle devices have extremely high requirements for power consumption and heat dissipation, and technical thresholds high. It is predicted that in the future, cars will become supercomputers, and in-vehicle sensors will far exceed mobile phone sensors and bring massive amounts of data. Time threshold. It will take at least 5 years from the start of the project, to mass production, and to sales. The chip industry is a comprehensive competition of software + hardware. Horizon has complete software and hardware capabilities.
 
Regarding the auto industry and Tesla, Dr. Yu said that the first half of the auto industry's competition-electrification, the second half of the competition-intelligence. The real game is in the second half. In the field of autonomous driving, Tesla has been ahead of its competitors for about 5 years. Tesla's speed has boosted the development of the entire industry. Horizon will seize the opportunities of the times and focus on the field of autonomous driving to become the world's leading company.
  Mr. Yalin Zhang, COO of Enflame Tech, believes that many start-ups are not falling from 0 to 1, but from 1 to N. Therefore, he currently focusing on cultivating team stability, and 99% of his time is spent internally. At present, the single-point training chip promotes the inference chip, which also verifies the strong execution of Enflame.

He said that he commented on the current cloud AI, not just talking about chips and architecture. Nvidia is not only doing cards and chips, but also systems and clusters. Nvidia's acquisition of Israel's Mellanox Technologies and ARM is a "home run" for servers. Last year, Nvidia’s data center revenue surpassed that of the game segment for the first time, which also verified this. Therefore, when discussing cloud AI, we are talking about systems-interconnection, ecology, and clusters. IDC clusters in the cloud, currently only a few companies in the world can do, in addition to Enflame, there are NVIDIA, Google and Huawei. The need for clusters, take the game business of Enflame’s core customers as an example, which requires dozens of servers for cluster training, so the chip needs to support clusters. The difficulty faced by cloud chip start-ups is that existing NVIDIA users are unwilling to transfer existing development and KPI models to the new architecture. Therefore, Enflame’s strategy is to first follow in the footsteps of NVIDIA, but through the advantages of product cost and localized services, gradually grasp the right to speak, and finally win the to B market.
 

Source: Oceanpine Capital