Summary
* What is new - Intel announced recently that it plans to merge Tower Semiconductor. In addition, the news reported that Intel may license its x86 CPU cores to chip vendors in order to grow its foundry business. We believe these moves could have limited impacts for TSMC in the long term.
* TSMC’s profound platform - TSMC’s open innovation platform (OIP) has a strong intellectual property (IP) library and design ecosystem for customers’ product design (please see chart 1/2/3 for details) while Intel lacks a complete product design platform and IP library. In addition, TSMC’s design for manufacturing (DFM) provides more than 50% reduction in time-to-market for customers’ products vs. tier-two foundries due to the verified design rules which are key for wafer fabrication yield rate improvement pace and product design cycles. These core competences have been built up in the past 30+ years, which are more than 10 years ahead of tier-two foundries (including Intel).
* Intel’s cost disadvantage - our industry surveys suggest that TSMC’s wafer fabrication cost could be 25%+ lower than Intel’s. We attribute this to 1) TSMC’s shorter production cycle time, 2) a faster yield rate improvement pace for advanced nodes at TSMC (TSMC’s 3nm mass production schedule to be 1 year ahead of Intel’s), and 3) TSMC’s high utilization rate in a slow season due to a strong customer base. Even with the US government’s latest subsidy support for the US semiconductor companies, TSMC’s wafer fabrication cost could be 10-15% below Intel’s in the long term given TSMC’s continuous improvement in operational efficiency and scale strength.
* Conflict of interest and weakness in x86 CPU power efficiency - Intel could continue to compete with potential foundry customers in computing, Internet of things (IoT), autonomous driving chip, GPU (graphics processing unit), and FPGA (field-programmable gate array) in the long run given its exposure to these areas. Intel has been trying to penetrate the foundry industry in the past 10 years. However, foundry customers have been concerned over the technology leakage issue if these companies were to co-design products with Intel’s foundry division. Notably, chip vendors may prefer ARM-based or RISC-V CPU architecture rather than Intel’s x86 CPU architecture. This could stem from 1) x86 CPU power efficiency issue vs. ARM-based and RISC-V CPU and 2) chip vendors’ and system houses’ (Apple, Amazon, Microsoft, Meta and others) increasing product designs for ARM-based server/PC CPU and consumer/mobile devices and RISC-V for IoT applications. As such, it could remain challenging for Intel to attract potential customers for its x86 CPU cores and gain share in the foundry market in the long term.
* Geopolitical risks - we believe Chinese chip vendors may be reluctant to adopt Intel’s x86 CPU cores to design their products and outsource foundry fabrication to Intel in the long term in light of US/China trade tension. In addition, Tower Semiconductor may lose Chinese customers to TSMC and other tier-two foundries given Chinese chip vendors’ growing concerns over potential supply disruption of an US foundry in the long run if Intel were to successfully merge Tower Semiconductor.
* A solid re-rating story - we expect TSMC to generate a 20-25% earnings CAGR and stock compound return in 2022/26. This could be driven by its enhanced technology leadership in front-end wafer fabrication and back-end packaging services, better pricing power, market share expansion, and secular growth of high-performance computing, automotive, computer visualization, and IoT markets in 2022/26.
郑重声明：以上内容是作者基于公开信息所做的判断，仅代表个人或嘉宾观点，与本网站、任何公司与任何机构立场无关。股市波动与很多因素有关，投资决策是个人基于自己的研究分析所做的决定，本文目的在于事实、观点分享，不构成任何投资建议。本文未经本营和作者书面许可，任何机构和个人不得以任何形式转发、转载、翻版、复制、刊登、发表、修改、仿制或引用文章的全部或部分内容。本营对任何第三方的未经授权行为所产生的影响不承担任何责任，同时保持实施法律行动的权利。
更多详细信息，请联系：michaelchou@acecamptech.com
