Most semiconductor enterprises in China are focusing on their survival and profit. It's impractical to force them to keep following Moore's Law or trying the IDM model for memory volume production.
Professor Bo Zhang of the University of Electronic Science and Technology of China recently suggested that "More than Moore" will become an important opportunity for the Chinese semiconductor industry. Professor Zhang's unique insights on this topic are now attracting interest from the industry.
The “More-than-Moore” concept seems new. It's not. Professor Zhang said that it is, actually, nothing more than one of the three directions of Moore's Law.
Professor Zhang refers to the “More-than-Moore process” as “non-size dependent.” It describes the improvement of the value or performance of the device, not by the size reduction, but by the increase of functions.
Moore's Law has governed the semiconductor industry for more than 50 years, shrinking the size of transistors and enlarging the diameters of wafers, with the former playing a critical role.
But Moore’s Law has its limitations. It worked in an era when increasing the density of transistors reduced costs. Moore’s Law, however, failed to predict the greater power consumption at the cost of improved transistor performance. This has prompted the industry in recent years to look at other dimensions of the Law – namely, “non-size dependent" processes.
Why More-than-Moore now?
Global semiconductor sales reached $200 billion in 2000. It took 13 years to hit $300 billion in 2013, but just four more years to add another $100 billion. By 2017, the chip industry grew to $400 billion, and then to $470 billion in 2018. It's very likely that semiconductor sales will reach $500 billion by 2020. In short, this illustrates the fast rise and value chain materialization of silicon in global electronic product markets.
On the other hand, the effects of shrinking process size are waning as Moore’s Law is approaching its physical limits, exploding the cost of IC design. Fewer manufacturers are capable of continuously pursuing shrinks. That's why More-than-Moore has become the weapon of the semiconductor industry in waging a new war.
Most industry experts believe that More-than-Moore is aiming at analog, RF, power, MEMS, as well as the 2.5D and 3D stack packaging, which has become so hot recently and may become one of the most critical driving powers in the coming 20 years.
Industry consensus suggests More-than-Moore may focus on 8-inch wafers and some portions of 12-inch wafers, covering a process geometry of above 28nm. When the 12-inch silicon wafer was just emerging in 2000, the dividing line was 90nm. The gradual upgrade of 8-inch silicon wafers has now expanded to 65nm. It all depends on economic factors, as investments are required to upgrade 8-inch equipment and purchase mask aligners for 90nm or smaller process.
If you employ 8-inch wafer equipment in More-then-Moore process, you can enjoy the following advantages: 1) All equipment has fully depreciated and requires no additional cost; 2) The 250nm-130nm process is relatively mature and stable; 3) You won’t require software upgrades for the manufacturing equipment, usually controlled by equipment manufacturers like Applied Materials or ASML.
More-than-Moore technology is ideal for distributed markets, diversified product lines, robust application relations, and none dominant players, noted Professor Zhang.
A closer look into the global top 10 suppliers of analog, RF, power, and MEMS devices suggests that they are topped by established companies with long history. They include NXP, Infineon and STMicroelectronics from Europe; Sony, Toshiba, Mitsubishi, Fujitsu, and Renesas from Japan; as well as TI, Microsemi and ON Semi from the United States. Many of these manufacturers still stick to 8-inch or even 6-inch wafers, but they have a pivotal position.
More-than-Moore production lines in China
Huilian Wang, general manager of the Xiamen Semiconductor Investment Group, said that in 2018 there were 33 fab lines in construction or planning. Of these, 21 are for 12-inch lines, 11 for 8-inch lines, 16 for More-than-Moore, nine for logic, and eight for storage. The SEMI data suggests that the Chinese production capacity of wafers accounts for 16% of the world total and may soar to 20% in 2020.
SMIC, the government-backed foundry, is not only developing 14nm and smaller logic processe, but is also fast expanding its 8-inch capacity at Tainjin, Shenzhen, Ningbo, and Shaoxing to name a few. Meanwhile, second-tier manufacturers like Huahong, China Resources Microelectronics and Shilan — which have been enjoying quite healthy margins — are pushing into the 12-inch field from the 8-inch one. Both are aiming at greater product diversification and penetrating the More-than-Moore process segment.
Decisions made by these companies reflect the Chinese industry’s realization that they must catch the More-then-Moore trend quickly and deeply. The emergence of so many More-than-Moore production lines means that, at least for the time being, they best fit the development requirements of the semiconductor industry in China.
'Impractical' for China to pursue Moore’s Law
In terms of the industry’s interests, it's necessary for SMIC and Huali Microelectronics to pursue 14nm or smaller logic processes, Yangtze Memory Technologies to develop 3D NAND flash memory and Hefei Changxin Memory Technologies to focus on 19nm DRAM. Yet they face an uphill battle because they are not just starting everything from scratch, but must also think about later expansion. This is a long and bumpy road with heavy investments, long cycles, mounting technology obstacles and demanding competitiveness in product markets.
Most semiconductor enterprises in China are focusing on their survival and profit. It's impractical to force them in following the Moore’s Law [in pursuit of finer geometries] or try the IDM model for memory volume production. It's inevitable for them to adopt the More-than-Moore technologies, instead.
The Chinese semiconductor industry dominates, to some extent, the global 8-inch process OEM markets while it lags in case of analog, high voltage, MEMS and IDM model production.
Data provided by Professor Zhang suggest another benefit of the More-than-Moore process: it helps in raising the portion of localization. Average prices of ICs imported by China are a mere $0.75, less than one dollar. It imports a large number of high-end CPUs worth hundreds of dollars each, and also imports a large number of discrete devices, power management ICs, and microcontrollers that do not require advanced technology. Therefore, the More-than-Moore process can leave hefty room in the market for domestic companies to play.
No short cut for China
It's clear that there is no shortcut for the Chinese semiconductor industry to outrun the competitors. More-than-Moore is opportunity and risk at the same time. We are chasing a market full of powerful and experienced competitors. There are no natural advantages in existence except our larger global market with plenty of opportunities, diversified product mix and good association with applications. These are our unique position to cut a piece in this demanding market.
Success of business is determined by market space, technology capabilities, timing and most importantly, a competent CEO.
The impact of the current trade war on China's semiconductor industry — still in development — cannot be underestimated for two reasons. One is that the electronic industry’s supply chain has partially moved out of China, and the other is the mounting intimidation by the United States.
The sum of these two developments may bring quite a big pressure for China who plans to maintain its average annual growth in the semiconductor industry at 20%.
Yet, everything comes like a "double-edged sword" which may make something "bad" into "good." The trade war is bringing talent back from abroad and the external pressure may lead to a greater performance by evoking a fighting and cooperative spirit.
‐ The author Mo Dakang is an alumni of Zhejiang University, prominent scholar and industry critic who witnessed the development of China's semiconductor industry in the last 50 years.