Golden opportunity: Can domestic semiconductor IP take advantage of the current situation?

With the rise of more and more domestic chip design companies and terminal manufacturers joining the chip manufacturing army, the domestic IP industry is welcoming a rare opportunity. However, having a huge market is only a starting point. Only long-term accumulation and stable output can truly usher in the final breakthrough.

The prosperity of IP market

As the global semiconductor industry continues to grow, the entire IC design market continues to grow rapidly. According to data released by SIA, global IC sales in 2020 were US$439 billion, a year-on-year increase of 6.5%. As an upstream link in IC design, semiconductor IP has also ushered in vigorous development due to its high performance, excellent power consumption, moderate cost, and shortened design cycle.

IPnest conducted statistics on the global semiconductor IP market in 2020, and the results showed that global design IP sales increased by 16.7% in 2020 to US$4.6 billion, which was the best growth since 2000 and one of the biggest highlights of the entire IP market last year. In addition, the growth rate of the IP market has exceeded that of the semiconductor industry itself in the past 10 years, showing a trend of rapid growth.

By product category, the processor IP market is expected to reach US$6.255 billion in 2027, with a compound annual growth rate of 10.15%; the mixed analog and digital IP market is expected to reach US$1.332 billion in 2027, with a compound annual growth rate of 6.99%; the RF IP market is expected to reach US$1.124 billion in 2027, with a compound annual growth rate of 8.44%.

The fastest growing segment will be interface IP. According to IPnest, interface IP vendors will have a total revenue of $1.068 billion in 2020, compared to $872 in 2019, a year-on-year growth rate of 22.4%.

Industry insiders believe that the focus of interface IP is also shifting from mobile phones to data centers. In 2010, smartphone SoCs integrated protocols such as USB, storage controllers, HDMI, DP, SATA, and MIPI, but not PCIe or Ethernet. Currently, data-centric applications account for the largest share of the interface IP business. Advanced memory controllers (DDR4, HBM2, GDDR6), PCIe, and Ethernet that require high-speed SerDes, as well as emerging chip-to-chip (D2D) solutions, can be found in data centers, servers, wired networks, and 4G/5G base stations.

Why is IP becoming more and more important? The main reason is that it can speed up the time to market. IP reuse and hardware-software co-design have become the mainstream of current chip design. Most SoCs are designed based on a combination of multiple IPs.

Another reason is the continued diversification of IC applications. Li Zhiyong, director of IP technical support at Cadence Electronics, said: "The driving forces behind IP development are AI, machine learning, big data analysis, advanced models and simulations, including COVID-19 drug development testing, autonomous driving, climate analysis, etc."

The increase in design costs brought about by advanced processes is also an important reason for the rise of IP. Advanced process nodes have continuously increased the complexity of design, thereby increasing design costs. According to the IBS report, taking the design cost of the first process node in the mainstream application period as an example, when the process node is 28nm, the design cost of a single chip is about US$41 million, and when the process node is 7nm, the design cost quickly rises to about US$222 million. If the use of IP is increased, the chip design cost can be reduced accordingly. In addition, with the evolution of the process, the number of transistors in the chip has increased significantly, and the number of IPs that can be integrated in a single chip has also increased significantly. Taking the 28nm process node as an example, the number of IPs that can be integrated in a single chip is 87. When the process node evolves to 7nm, the number of IPs that can be integrated reaches 178.

"Companies that once had their own IP departments are starting to adopt third-party IP." Industry insiders have observed that this trend is accelerating, and the comparison of total cost of ownership between research and development and purchase ultimately makes chip companies more supportive of third-party commercial IP.

Yu Chengbin, co-CEO of Xinyao Technology Co., Ltd., believes: "The sooner we can provide IP to IC design companies, the more their sales and profits can increase, development costs can be significantly reduced, and product life cycles can be extended. This is why IP has been able to accelerate the development of the semiconductor industry."

Recently, the new chiplet in packaging has also brought huge imagination space for the rise of IP. According to Omdia data, the global chiplet market is expected to reach US$5.8 billion in 2024, with an average annual compound growth rate of 44%, and is expected to gradually expand to the entire semiconductor market.

In the past, to design an SoC, you needed to purchase multiple IPs from different IP suppliers and integrate them together. With Chiplet, users only need to buy small silicon wafers and package them into a SiP (System-in-Package) to realize an SoC. Therefore, Chiplet can also be regarded as a hard-core IP provided in the form of a chip. This is the chipization of IP, which will become a major direction of industry development in the future.

It is not easy to build a good IP

From the perspective of global market regions, the Asia-Pacific region currently occupies the world's largest semiconductor IP market and is expected to become the fastest growing region in the world. Among them, China's IP market has the greatest potential.

The five-year compound growth rate of newly established chip design companies in mainland China reached 24.7%. The planned chip design projects increased significantly, and the overall growth rate far exceeded the global average, bringing growth space to the domestic semiconductor IP industry.

"The development of domestic consumer electronics and Internet applications is booming. Many system manufacturers need to make their own chips. Alibaba, Tencent, and ByteDance all need to make their own chips, which naturally leads to a strong demand for IP." Yu Chengbin said: "In terms of processors or AI engines, manufacturers have to make their own chips. In this situation of flourishing, IP is more needed."

However, the large domestic IP market does not mean that domestic IP is strong. IP giants such as Arm and Synopsys still occupy an absolute leading position in the market, and domestic IP only occupies a small share.

Someone has summarized the three main difficulties faced by the development of domestic IP: first, the rapid upgrading of various interface IP standardization protocols, which determines that IP research and development requires long-term investment; second, IP research and development needs to continuously track the process evolution of various chip manufacturing companies, and must have the ability to quickly transplant process technology; finally, after integrating the protocol standard into the IP and customizing it for each process technology, it is also necessary to jointly conduct chip verification with downstream chip design companies and mass production to ensure the marketization of IP products.

Two of them are related to the process, which shows the close connection between the process and IP. "Only IP that has been verified by the process can be accepted by the market." An industry expert told Jiwei.com.

Yu Ke, president of Shanghai Xinlianxin Intelligent Technology Co., Ltd., pointed out the importance of IP verification with a vivid metaphor, "Some IPs have no problems during simulation, but it may not be the case when they are actually running, because the timing problem is very important in the chip. It is not as simple as A+B=C. If A reaches B but not B, the chip will not run."

Yu Chengbin holds the same view: "Frankly speaking, the threshold for making IP is very low, but its interoperability and reliability require long-term iteration and verification. Only IP that has been mass-produced can be accepted by the market."

"Moreover, the demand for IP is increasing with increasingly advanced processes. There are also many requirements on driver specifications, functions, and costs, and we are very concerned about reducing development risks and product cycles," Yu Chengbin added.

The market calls for advanced process IP technology and complete IP solutions to support more and faster chip innovation. However, due to various high-tech barriers, the development of domestic IP technology has been lingering in the mature process and has not been able to go deep into the advanced process field. Therefore, the development of IP is not only the business of IP companies themselves, but also relies on the entire industry chain to provide more support.

Moreover, since most IPs are sold through licensing, it is difficult to obtain scale benefits like terminal products. Therefore, although IP is important, it is a difficult road. Industry insiders pointed out that the slowdown in the growth of the number of nodes, industry integration, and IP standardization will increase the profit pressure of IP companies.

Therefore, doing a good job in IP is not only a competition of technology, but also a competition of market understanding, business model, etc. Zeng Keqiang, Chairman and Co-CEO of Xinyaohui, believes that there are two key factors for domestic IP to gain market recognition: first, the company needs to be able to make firm long-term investments and focus on technology research and development and product polishing, which requires the company to provide sufficient guarantees in terms of strategy and capital investment; second, the talent team. A successful IP team must have more than ten years of experience in mass production, running-in, iteration and industrialization.

All of this is based on long-term honing, so there is no shortcut for domestic IP. It can only rely on continuous accumulation and making more "allies" to truly achieve a breakthrough.