Rapidus does not snatch up on electric customers? Strategies behind 2-nanometer "small crowd strategy" in Japan

Rapidus' recent developments have attracted high international attention, and major Japanese media such as Asahi News have also reported the company's major information on the production of 2 nanometer (nm) chips under the big title. According to Rapidus' instructions, the company will adopt different business models when entering the market, such as the "Rapid and Consolidated Manufacturing Service (RUMS)" model that covers the design, manufacturing, and packaging.

Executive Director Atsuyi Koike said the company would not compete head-on with NTEC because Rapidus would not aim for super-large production. He believes that some small AI new customers want advanced chips tailored to their needs rather than general purpose chips produced on a large scale, so Rapidus will focus on working closely with them to achieve the results they need. In addition, in order to win customers, its senior managers have also tried to connect with Broadcom, the United States and other technology factories such as Google, Amazon, Facebook, Apple, Microsoft (commonly known as GAFAM), in order to supply chips to them.

It has only been less than three years since its establishment in August 2022, and Rapidus has reached this level. This is a very admirable stage achievement. Nevertheless, the company's future development seems to be facing many uncertainties. This article wants to explore from the perspective of strategic thinking of enterprise management to further clarify the possible impact of the company on our semiconductor industry.

Strategy Group

When discussing policy topics in the management domain, strategic groups are often mentioned. It refers to a group of manufacturers that adopt similar pricing, market positioning, product characteristics, and adopt similar strategies to compete for the same target market. Usually they are considered to have direct competition.

According to Xiaochi's statement, Rapidus aims to target some small AI new companies, hoping to compete with large manufacturers in different markets. If it can be done, it can be separated from each other, which means that Rapidus and other large factories are affiliated with different strategy groups, so it may face different opportunities and threat combinations without causing direct competition, and reduce the intensity of competition by design.

But in the industrial competition structure of semiconductors, is it possible that there are different strategic groups in the advanced process chip market? In addition, the question worth pondering is: How many small AI new companies can get involved in cutting-edge 2nm chip products? Are these small niches enough to support such a large 2nm investment cost for the company?

To answer these questions, you can think about them from the following aspects: Can Rapidus make a big profit from a small niche?

Judging from the actual actions of its senior manager trying to engage with Broadcom and GAFAM in the United States, Kochi's strategic thinking is far more ambitious than expected, showing that the company is not only satisfied with a small niche market.

Xiaochi may probably find that although these new companies can be charged high prices for custom production, if the order or market size is too small (for example, there may not be many small companies that are capable of involving 2nm chips in the past few years), it may not be enough to make Rapidus profitable and grow. If you cannot make a significant profit, the source of funds required to enter the next stage of competition will be full of variations, so you need to contact these technology factories to win big orders.

Multiple target markets in the same group

In fact, according to the theoretical viewpoint of "multi-varied target markets within the same group", in some industries, enterprises within the same strategic group can still be classified according to the characteristics of different customer groups, and position different customer groups according to differences in products, pricing, services, etc.

For example: Both Shi and BMW belong to high-level automobile strategy groups, but the target groups that are set by the two are slightly different. Mr. Bin mainly attracts successful people who pursue high social status and mature stability; while BMW mainly attracts high-income groups who are enthusiastic about driving and pursue personalized high-income groups.

The author believes that in the semiconductor industry structure, the "advanced process group" can be divided into different markets. That is, NTD, Samsung Electronics, Intel, and Rapidus are all the same strategic group, so there is still a direct competition between them; however, Rapidus locks (but is not limited to) this market segment of small AI new companies, in order to reduce the intensity of competition with the other three major factories.

Rapidus needs to consider the configuration of core resources and unique capabilities

Nevertheless, the strategy group should not only rely on externally visible market characteristics to plan, but should further think about the "configuration of core resources and unique capabilities".

If the manufacturer's resources and capabilities are of high homogeneity, such as similar technical capabilities, R&D capabilities, management experience, brand awareness, etc., it will make the company likely to adopt similar competition strategies; if the resources and capabilities are different, it will have the opportunity to lead to a diversification of strategies or business models. In other words, if a manufacturer wants to serve customers in different market segments, it may need to establish different core resources and unique capabilities combinations.

To serve small AI new-creator companies that pursue speed and specific innovative technologies, Rapidus' "Innovative Integration for Manufacturing (IIM) factory" needs to think together with customers about how to innovate design. These companies are often focused on specific applications with strong computing power, AI acceleration performance and low power consumption, such as the rapidly developing edge computing domain. At this time, Rapidus' design ability and flexible and flexible production ability are important.

However, if you still want to serve a design company like Broadcom, Rapidus must provide high-quality OEM services to produce stable, reliable and efficient chips for application in enterprise-level equipment and data centers. Furthermore, because of its large order size, Rapidus must be able to provide large-scale production capacity and excellent production capabilities with high yields.

Generally speaking, to serve these two different types of enterprises at the same time, manufacturers may need to have different core resources and unique capabilities, but sometimes this is not easy. For example, the entire single-dragon process carried out by Rapidus, which produces only one wafer at a time, may have its own advantages, including being more accurate in control, so that the overall process can achieve higher accuracy and consistency; however, this processing process will result in a significant increase in the production cost of a single wafer, and a large-volume production is more consuming, which is not suitable for the rapid production of large quantities of wafers.

From this we can see that Rapidus' current planning and resource configuration should be more suitable for serving small AI new companies. Firstly, not only because the company was just established, the resource and capability base was weak, but also because it involves the above configuration problems, it is not easy to adjust significantly in the short term. Therefore, can Rapidus have both fish and bear's paw? Worth thinking about.

Rapidus should think about the degree of difference and conversion cost issues

On the other hand, even if Rapidus is really capable of doing it, it does not mean that it will work. The main reason is related to the degree of difference and the switching cost.

Rapidus should constantly ask himself: Compared with competitors, is there really a difference in my skills (how much difference is there)? Can you provide unique services (how unique is it)? Can we produce products and services that are more in line with customer needs due to differences? Is it possible to continue to promote the next generation of innovative technologies?

If the degree of difference between the competitors is not very high, it will not be able to occupy a favorable position in a specific market structure and will face greater challenges and competition.

According to Rapidus' planning, the company will adopt different business models when entering the market, and its details include integrated manufacturing and packaging automatic operation systems, etc., mainly to distinguish them from other large factories, hoping to bring their customers unique value, thereby improving profit and competition.

Although the Rapidus test is differentiated on the above-mentioned layer surfaces, the effectiveness of the RUMS model remains to be verified, and the operating pressure of its integrated manufacturing and packaging may also be limited on large factories.

In addition, judging from the current situation, its competitive strength is too strong, which will also weaken the value that the company's differences bring to customers.

With the years of hard work and deep cultivation of Taiwan Power, its outstanding production capabilities, excellent TSMC-Online™ customer service system, a complete open innovation platform, a complete ecological system, etc. have been well received and highly competitive. Many customers have not only signed long-term supply contracts with NTEC, but have been tightly trapped by them. If the manufacturer is traded, it may bring high uncertainty, conversion costs and business risks to it. Therefore, unless they have to, they may not try it easily.

It is probably not so easy for large customers who want to take Taiwan electricity.

From the above explanation, we can see that Rapidus cannot clearly isolate different strategy groups and will face greater competitive pressure. In addition, if Rapidus' technical level and operation process are not much different from those of large factories, it is also more unique and difficult to replace core capabilities and resource energy competition, because it involves customer conversion cost issues, it is not easy for the company to seize the large customers of Taiwan Electricity as expected.

Although the author is very respectful to the mission and his past understanding of Rapidus' senior manager, he also believes that the company's breakthrough in semiconductor technology can help Japanese scientific research and other scientific and technological industries; however, the author believes that:

The impact of Rapidus on our semiconductor industry at this stage is limited. As for whether the company can produce smoothly in 2027? And what are the production yield and chip performance after mass production? It is worth our continued observation. From the perspective of enterprise business, in addition to the above issues, the company's future will still be full of hardships and challenges due to the above issues such as customers, quantity production, yield, huge funds, continuous research and development investment, and the extension of outstanding talents.

Can Rapidus kill a bloody path and build a beach for survival? This is a test of the business wisdom of his senior manager and the fighting spirit of the whole staff.

Former Intel CEO Gelsinger has some advice for Japan’s Rapidus Japan’s Rapidus in talks to supply chips to U.S. tech giants: CEO ラピ　ス, basic parts trial success　Funds・Customer…Product Subscription QuestionはShan Punishment Extended reading: Can Rapidus' unique operating mode be removed? Can Japan's Rapidus 2 nano chip production be so smooth? Only by breaking through the three difficulties can you create a miracle