Increased 40 times in five years, we have witnessed the miracle era of semiconductors

It is commonplace for the semiconductor market to be treacherous. As the stock market is soaring, the semiconductor industry is experiencing historic integration. While Intel is struggling, AMD and Nvidia are using huge stock gains to make large transactions. Time never said a word, but returned to all the problems. Looking back at the semiconductor stock price trend in the past five years, the performance of some companies is really shocking. AMD’s performance is the most eye-catching, each stock price has increased by 40 times, NVIDIA has also prospered because of AI, and TSMC has been so successful as an industry changer. It can be said that we are the witnesses of this age of miracles.

　　“Headwind” AMD: Soaring 4150% in 5 years

The biggest increase was undoubtedly AMD. In October 2015, AMD struggled to maintain the level of US$2 per share. At that time, with the decline of the personal computer business, it seemed to have no reason to exist. But after a long period of operational struggle, AMD finally got out of the predicament in terms of economic value creation under the leadership of its outstanding leadership team. Today, AMD’s stock price has soared all the way to around $85. AMD’s stock price has doubled 40 times in the past five years, an increase of 4150%.

AMD stock price list in the past 5 years (Source: Yahoo Finance)

In 2014, AMD appointed Lisa Su as its new CEO. Under her leadership, the company began a large-scale transformation. Su makes the company shift from PC products to diversification. Game console chips, data centers and virtual reality (VR) have become increasingly important revenue drivers. These moves enable the technology giant to directly compete with NVIDIA and its graphics chips. And in the server chip market, AMD and Intel go hand in hand. AMD said on its analyst day in March that it expects data center revenue to account for 30% of total sales by 2023, up from 15% in 2019. Coupled with the acquisition of FPGA leader Xilinx, it has further consolidated its reputation as one of the long-term winners of data center chips.

AMD has just announced its latest financial results for the third quarter. AMD’s computing and graphics divisions (CPU and GPU) achieved revenues of $1.67 billion, an increase of almost one-third over the same period last year (precisely an increase of 31%) . This figure also increased by 22% from the previous quarter (Q2). Part of its strong performance is due to sales of Ryzen processors.

As key industries such as gaming, virtual reality and data centers maintain their growth trajectories, AMD should still be one of the most important beneficiaries. Five years from now, AMD’s revenue may not reach a growth rate of more than 50%. Nevertheless, these growth will allow AMD’s stock to continue to rise.

　Nvidia, the “darling” of the AI ​​industry: a 5-year increase of 1796%

In the past five years, NVIDIA has benefited from the rapid development of the gaming and data center industries and has become a star in the stock market. This year alone, NVDA stock has risen nearly 134%. Looking back over the past five years, Nvidia’s stock price has risen from a range of $20 to $30 five years ago to $550 today. In the past five years, the increase was 1796% (refer to the stock price of $29 on October 26, 2015).

Nvidia stock price list in the past 5 years (Source: Yahoo Finance)

Nvidia has a place in five end markets, which are data center, gaming, automotive, OEM/IP, and professional visualization. In the past few years, its gaming channel has historically contributed the most to overall revenue, but the company’s acquisition of Mellanox led to rapid growth in data center revenue in the second quarter, making it the largest source of revenue in the five end markets.

Nvidia recently announced the acquisition of chip designer ARM for US$40 billion. This acquisition may be an important turning point for Nvidia as it will expand its hardware coverage from GPU and industrial hardware to consumer computing, of which mobile computing is the growth so far. The fastest area. If Nvidia can complete the acquisition of ARM, then Nvidia will become a firm foothold in the consumer and industrial hardware fields. It will also make good use of the huge offline AI boom, which will make Nvidia the most powerful AI platform among all semiconductor manufacturers.

AI is the future of Nvidia. Nvidia’s core business of consumer and industrial hardware is the backbone of artificial intelligence applications. Artificial intelligence will become the next technological frontier.

　　TSMC has enough process and packaging dividends: an increase of 319% in the past 5 years

In 2015, TSMC stood out by virtue of its own InFO packaging technology, and monopolized Apple’s mobile phone processor orders until 2020. Driven by advanced processes and advanced packaging, TSMC has been the best in the foundry industry in recent years. In October 2015, TSMC’s stock price was approximately US$21, and five years later, TSMC’s stock price reached a maximum of US$88, an increase of 319% in the past five years.

List of TSMC’s stock prices in the past 5 years (Source: Yahoo Finance)

TSMC is the world’s largest fabless semiconductor company integrated circuit manufacturing plant, using 261 different technologies to produce 10,436 different products for 481 different customers. TSMC currently has more than 50% of the market share in the US$42 billion semiconductor foundry business, and Samsung, which ranks second, has approximately 18%. Leading fabless companies, including Nvidia and AMD, are leveraging TSMC’s advanced technology capabilities at the 7nm and 5nm nodes. Intel is also turning to TSMC’s 7nm, pushing revenue to exceed $2 billion.

It is understood that two-thirds of TSMC’s 5-nanometer production capacity, with a total of about 180,000 pieces, have been packaged by Apple. With the addition of AMD, Qualcomm, MediaTek and other major customer card capacity, TSMC’s 5-nanometer capacity is fully loaded ahead of schedule. In order to meet the needs of huge customers, TSMC is also actively purchasing land and factories to accelerate investment. Among them, Nanke Plant P1~P2 has started 5nm mass production, P3 is expected to be put into production at the end of the year, P4 will start civil construction, and P5~P6 will be used for 3nm. For advanced packaging, the Zhunan plant broke ground in July and will be mass-produced in the second half of next year.

The semiconductor industry is at a turning point. With the slowdown of CMOS scale, advanced packaging technology has become a necessary option for high-performance chips in the post-Moore era, and it is also regarded as the key to the continuation of the life cycle of Moore’s Law. The timing and grasp of advanced packaging technology is now an important part of TSMC’s dominance, and it is also the main difference between TSMC and Intel.

Since TSMC introduced CoWoS as a high-end advanced packaging platform for heterogeneous integrated silicon interfaces in 2011, from InFO (and its multiple versions of InFO-os and InFO-aip) to SoIC, and then to 3D multi-stack (MUST) A series of innovations such as system integration technology and 3D MUST-in-MUST (3D-mim fan-out packaging). Last year, TSMC successfully trial-produced 3D IC packaging processes such as 7nm System Integration Chip (SoIC) and 16nm Wafer Stack Wafer (WoW), and is expected to enter mass production after 2021.

FPGA leader Xilinx: an increase of 159% in the past 5 years

For Xilinx, it has not been easy in recent years. The continued impact of the trade war between China and the United States has exacerbated the normal cyclical decline in semiconductor sales. In recent years, Xilinx has continued to transition, and platform construction is a well-known path. Xilinx has transformed from a pure field programmable gate array (FPGA) card manufacturer to a “platform company.” Xilinx’s stock price was $47 in October 2015. It is now around $122, a 159% increase in the past five years. Now, after being acquired by AMD, what kind of sparks can the two make?

Xilinx stock price list in the past 5 years (Source: Yahoo Finance)

Xilinx specializes in Field Programmable Gate Array (FPGA). Although typical semiconductors cannot be programmed after being put into use, FPGAs have one major difference. Users can actually delete and replace software without changing the hardware, which is a huge advantage compared to CPU and GPU. Use cases for FPGAs are even more ubiquitous. Microsoft uses FPGAs in its data centers, Amazon provides them in its cloud services, and even the military uses them for F-35 joint strike fighters. The data center is one of its most important end markets, and the competition with AMD and Nvidia also makes AMD very interested in this.

Kurt Marko, principal analyst at Marko Insights, believes that AMD and Xilinx “are about the integration of CPU, GPU, and other accelerators on the system or motherboard.” “FPGA provides the flexibility to implement multiple functions, which can be offloaded And accelerate specific features to enhance the x86 kernel.”

Concluding remarks

In fact, it is not just these large foreign factories. The entry of large domestic funds and the opening of the science and technology board have opened up the “two channels of supervision and supervision” in the semiconductor industry. Semiconductor companies with a market value of 100 billion are constantly rising, and one miracle in the semiconductor industry is taking place. .