The status and influence of global foundry in the semiconductor industry has been increasingly deepened, thanks to Zhang Zhongmou's second comeback in 2009, pushing the development concept of OEM to 'ultimate'.
The history of OEM
In 1987, the concept of foundry was started in Taiwan, China. It was a big leap in the semiconductor industry chain. It first promoted the progress of the fabless model. Today, it can be seen that the advanced process technology of OEMs has improved The IDM model is comparable, leading to the increase of OEM's weight factor in the semiconductor industry chain. For example, in the global chip (integrated circuit) output, one in every two blocks is processed by TSMC.
However, when TSMC, or UMC, just started, the semiconductor industry is actually dominated by the IDM model.
Until 1994, the output value of global fabless was only US$3.6 billion, reflecting that at that time, in addition to IDM, the foundry's process technology was far behind IDM. Therefore, the foundry was only the second supplier of IDM plants. When the capacity was in short supply, it could be used as In the IDM plant, the design industry was still very weak.
In the beginning, IDM and OEM were different in their concepts. IDM plants will certainly strive to pursue the ultimate in craftsmanship. Moore's Law is one of its driving forces. For example, Intel will not spare any effort to promote the continuous shrinkage of its process and its adoption. New technology, so its research and development costs reach about $10 billion a year, which is almost the highest in the world.
The foundry is a service system, which provides customers with the services of process technology, to meet the needs of customers. The birth of foundry is highly sought after by the fabless, which can save the huge burden of investment in factories. Therefore, from a logical analysis, Workers should not pursue the most advanced process technology. The risks and costs are too high. To develop the individual process for the customer's needs is rare (depending on the customer's cost for the process development), it considers several customers. The use of the same type of technology to save costs, so the industry has a view that the technical content of the previous foundry is not too high.
Zhang Zhongmou pushes the OEM to the extreme
TSMC is also not likely to be a one-time climber. It was observed that it has two key points. It is one of the most significant investments in the late 1990s. It almost took off the hat of the OEM only as a second-rate technology. The other is Zhang Zhongmou in 2009. The second comeback, it stands tall, looks far, is determined to push the concept of OEM to 'ultimate', that is, OEMs should also actively develop the most advanced process technology, and keep pace with IDM.
Therefore, from the aspect of operations, it strengthened R&D to collect top talents. The average annual investment in equipment is about US$10 billion. After several consecutive years of hard work, we have been at the forefront of the most advanced technological processes, such as 7 nanometers since 2018. Production, its Tainan Jingyuan Eighteenth Plant was grounded in January 2018, and will use the Extreme Ultraviolet (EUV) lithography technology to produce a 5nm process on a large scale. It is expected to start trial production in 2019 and mass production by 2020. , It is expected to produce 1 million 12-inch silicon wafers a year, with a total investment of NT$700 billion. It is the world's first production line with an energy production capacity of 5 nanometers. The industry is even more admired for OEMs.
At this stage, TSMC has no doubt with Intel and Samsung. As a result, TSMC is fully capable of attracting the world's top fabless, including Qualcomm, Apple, Huawei, Nvidia and other orders.
After Zhang Zhongmou assumed the post of CEO in 2009, he first doubled the capital expenditure in 2010 to 5.9 billion U.S. dollars, led Taiwan Semiconductor Manufacturing Co., Ltd. to spurt the industry’s cutting-edge 28-nanometer process chips, and the 28-nanometer process chip has become the mainstream of the smartphone era. In the same year, TSMC won the order of Apple, which had been exclusively owned by Samsung, and became a giant in the foundry industry.
According to "The World Magazine" reported in October 2017, TSMC’s share price has increased by 237 percent in the seven years that Zhang Zhongmou has regained control in 2016. TSMC’s cumulative investment from 2010 to 2018 reached US$79.6B.
Today’s foundry, long since it was not UMC, the foundry positioning of TSMC was not a so-called fab-lite, but an open platform model that not only had IDM capabilities, but also adhered to OEM services. Networking, AI era migration, market demand IC design companies must establish a deeper cooperative relationship with foundry. It is reflected in the fact that both parties are not only increasingly dependent on manufacturing, but also have to maintain close communication in market definition. The relationship even extends to the terminal link, forming a 'fate community' of multi-party cooperation. In this way, we can resolve future challenges, ensure efficiency, quality, yield, and various commercial outcomes.
Conclusion
There have been many recent articles praising Zhang Zhongmou for his 'Lifelong Learning for Happy Work, Watching Success and Failure for Moderate Life' and it is indeed worth our studying.
Regarding the development of China's semiconductor industry, Zhang Zhongmou said, 'In the next five to 10 years, Continental Semiconductor will make great progress, but TSMC will make even greater progress, and mainland companies will still fall back to back-selling for five to seven years.'
Compared with the progress of TSMC, there may be an increasing gap. It is not that we do not work hard enough, but the progress of TSMC is really too fast. Its investment efficiency is higher and it is at a different level. It cannot be compared with TSMC.
Today, when a great man retirees, he certainly has a big influence on TSMC, and Zhang Zhongmou has long thoughts and plans to replace it with a double-headed system.
Future global OEMs may have passed the “golden” era. It is already very difficult to achieve a growth rate of 10%. Due to the decentralization of future market forces and the fact that process technologies are approaching the limit, the performance and power consumption of products, etc. More demanding, many new applications such as AI, autopilot, and Internet of Things are bright prospects, but the actual application of the market will take time to develop, so the foundry will face greater challenges.
