October 19, 2018 - 1:16pm EST by
2018 2019
Price: 38.54 EPS 13.59 0
Shares Out. (in M): 25,930 P/E 0 0
Market Cap (in $M): 198,220 P/FCF 0 0
Net Debt (in $M): -631,682 EBIT 0 0
TEV (in $M): 0 TEV/EBIT 0 0

Sign up for free guest access to view investment idea with a 45 days delay.



In the epic telling of the history of risk and markets “Against the Gods”, Peter Bernstein attempts to unlock the story of how capital and industry developed into what is today’s financial markets.   His description of risk as a balance of past and future captures the essence of how one can look at Taiwan Semiconductor


“The issue boils down to one’s view about the extent to which the past determines the future.  We cannot quantify the future because it is an unknown, but we have learned how to use numbers to scrutinize what happened in the past.  But to what degree should we rely on patterns of the past to tell us what the future will be like? Which matters more when facing risk, the facts as we see them or our subjective belief in what lies hidden in the void of time? Is risk management a science or an art? Can we even tell for certain precisely where the dividing line between the two approaches lies?”


Taiwan Semiconductor is a tale of past success, current excellence and future risk.

The investment thesis is relatively simple from a historical track record, and current position as a dominating market leader.  However the risk of the unknown future makes the thesis slightly less clear.



Founded in 1987 in Hsinchu Science Park Taiwan, TSM pioneered the pure-play foundry business model by focusing solely on manufacturing customer products.  By choosing not to design, manufacture or market any semiconductor products under its own name the company ensures it will never compete directly with its customers.  Today TSM is the largest “pure play” semiconductor foundry in the world. They manufactured over 9,000 types of products for 465 different customers in 2017.


A brief summary of the semiconductor market verticals for the sake of definition:

  1. Integrated Device Maker:

    1. Creates their own chip designs and manufactures those chips in their own foundry

    2. Examples : Samsung and Intel

  2. Fab-less chip maker:

    1. Chip designers who do not engage in the highly capital intensive process of creating and maintaining the foundry manufacturing for chips.  They focus on the design of chip technology and outsource the manufacturing to a foundry such as TSM

    2. The rise of fabless chip designers such as ADM and NVidia over the last few years has been to the benefit of TSM

  3. Pure-play foundry:

    1. They only manufacture chips to that have been designed by others.  The focus on the technology and process of chip manufacturing and build to the design specifications of others.

    2. TSM is the leader in this space globally


TSM’s largest source of revenue is the communications/cell phone market. They are certainly exposed to the global smartphone cycle.  However, communications/smart phone is projected to drop from 60% to 45%-50% of revenue over the next 4-5 years as high performance computing (HPC), the Internet of things (IoT), and automotive applications grow at much faster rates.


•    Projected growth rate of smartphone demand next 5 years:  -2%

•    Projected growth rate of HPC/Iot/Automotive  next 5 years: 7%



The ongoing growth in mobile, rise in artificial intelligence, proliferation of the internet of things, high powered computing and automobile applications should result in sustainable upside in aggregate computing power globally.  Most analysis predicts 6% annual growth over next 5 years for the Semiconductor industry as a whole.






The numbers generally speak for themselves.  It is not all that common to find companies with this level of operating metric excellence over such a long time frame:



10 Year Average

5 Year Average


Average annual sales growth




Average annual EPS growth




Average operating margin



10 year low: 31%

Average ROE



10 year low: 19%


They clearly exhibit traits of a best of breed operator in their market.  Any of the metrics in the graph below represent an outstanding long term track record over the 17 year period.




The foundry business is highly capital intensive.  Taiwan semiconductor has built a 25% 10 year average ROE funded completely out of operating cash with a negative net debt position.   The company has no net debt, a significant cash pile, is rated A+, and has a minimal amount of intangible or goodwill value on the balance sheet.   Company balance sheet strength is important to me. I feel it gives a company optionality going forward.


A strong balance sheet will enable them to:

  1. Survive through the market down cycle (More on this below in the risk section)
  2. Leverage the balance sheet to maintain a higher ROE over time.(If done appropriately with good capex/cash flow investment)

  3. Invest in the business to adapt to change in the constantly evolving and changing world of commerce and technology.   If you can’t evolve in the modern world you eventually die. Capital access and balance sheet strength all other things being equal will give a company a better chance at doing this




As you come to the conclusion of this write up you will see part of my thesis is the medium term cash flow generating ability of the business.  One example of this is the company’s growth of shareholder return via the dividend. TSM currently pays a dividend of 3.4% that is well covered at a 50% payout of earnings and has a 21.5% 5 year CAGR.   The company is yielding more than the 30 year Treasury bond and still growing revenue, earnings and sales 10%+ year after year.




It is hard to predict the future of semiconductor technology.  But there is little debate that at the current moment Taiwan semiconductor is the market leading foundry operation in the world.


In order to understand the product cycle we need a basic understanding of semiconductor technology.


  • Moore’s Law: The number of transistors in integrated circuits doubles every year.

  • NM rating is how many nanometers wide a process size is. Its technically a measure of lambda/wavelength of light passing through the stacked layers of the chip

  • 7NM is currently the very latest technology and TSM is the only foundry in the world currently producing it.

  • As chips get smaller:

    • More energy efficient

    • Cheaper to make. (Can fit more dies on each wafer) but equipment and technology is more expensive.  (Economies of scale become even more important)

  • They are completing depreciation of old technology as they are investing in newer technology.

    • Expected to complete depreciation of their 28nm technology in 2019, and 16nm in 2021

    • As you can see in the charts below they are slowly manufacturing less of older technology and replace that with new technology over time.

  • As they are building the latest technology, they are improving processing and integration of  older technology

    • Better peripheral functions and integrated capabilities



  1. They lead the market in size and scale

    1. 56% of global pure-play foundry market

    2. Account for 90% of the industries net-profit

  2. They currently hold the dominant position as the technology platform leader.  They are the most advanced foundry operation in the world.

    1. 18-21 months ahead of the competition in technology development.  No one else in the world can manufacture 7nm chips right now.

    2. Consistently first in next generation technology (First in 7nm technology, already in development of 5nm and 3nm)

    3. Leadership in mature technologies (80-25nm) with more enriched and optimized processes.  In older technologies such as 28nm, TSM’s products show higher PPA (performance, power, area density) and best of breed technology suite

  3. Many analyst predict that over the next 3-4 years TSM’s scale and technology superiority will allow them to

    1. Grow market share from 56% to 66% of the outsourced foundry market as a whole

    2. Dominate 90% of the advanced node market enabling them to capture the secular growth of IoT, high powered computing and automotive market growth.

  4. Focused market leader in foundry capabilities.

    1. Has never owned or marketed a semiconductor product.  Foundation of trust with wide range of industry customers.

    2. Growth of fab-less(no foundry) companies chip companies has been beneficial. (NVidia, AMD etc.)

    3. Capital need for foundry business have become even more intense creating a wider moat.   

    4. Can take best practices learned across wide customer base and continually improve as market leader.


I personally am a believer in the long term continued secular growth of the semiconductor market. The cell phone market remains the largest driver and there will certainly be up and down cycles in the global smartphone market.  I believe the long term secular trend is the continued digitization of the world around us. The internet of things, artificial intelligence, high power computing, crypto-currencies, and the automation of the automobile will continue to drive the need for semiconductor volume growth and technology advancement.  For example, the US national Highway Traffic Safety Administration announced that it would require emergency automatic braking as a standard feature on all new cars by 2022. That is going to require fairly advanced chip design to be deployed to all new cars over the next 5 years. This by itself is in no way a massive catalyst, but just another small example of the continuing secular trend of increasing digital complexity that will drive the semiconductor market.  



Reminding ourselves of Bernstein’s balance make us consider why the historical analysis thesis alone is by itself not enough to invest capital.  The company’s futures prospects must be further examined.




The semiconductor industry can be highly cyclical based on end user demand for products such as cellular phones, TV’s, Computers etc.  The 2000 and 2008 recessions caused over 20% drops in demand for semiconductor chips. Combine a cyclical end market with a high fixed cost business and any downturn which decreases utilization can pose significant margin risk.

While I acknowledge the risk of the current cyclical trend in semiconductors can break.  I am comfortable with the cycle risk for the following reasons:

  1. I believe in long term secular growth of the market

  2. TSM is a market leading operator with a fortress balance sheet that will enable them to survive and thrive through the down cycle

  3. Even in past recession such as 2000 and 2008, the company has maintained positive net income and cash from operations.  They have shown their ability to manage through the cycle.

  4. I generally try to avoid market cycle timing, but rather focus on companies who I believe can manage through them.




  • TSM’s top customer is currently Apple and it represents a significant 17% of their revenue.   Their top two customers typically represent 25%-30% of their revenue stream.

  • Applied materials is 27% of their capex spend and there are 3 companies which represent 10% or greater.

  • COGS is less concentrated.  They have multiple silicon, chemical, and gas providers.  However access to top quality silicon etc. is a requirement of the business


I must acknowledge the concentration risk and dependence on a few key customers is a concern.  It is unknown how this could change in the future. On a positive note, I believe their business is becoming more diversified from an end market and customer base prospective over the next 5 years as the semiconductor market evolves.   I also believe their technology advantage gives them moat like attributes. Apple continues to move a higher % of their semiconductor needs to TSM because they have the best chip making capabilities.




The following is a quote directly from the risk section of company’s most recent 10k that caught my eye:


“Chinese companies are expected to be key players for new semiconductor fab development and fab equipment spending through 2020. There are over twenty new semiconductor fab projects that have been announced or are being developed within China in part due to various incentives provided by the Chinese government. Furthermore, the Company’s competitors may, from time to time, also decide to undertake aggressive pricing initiatives in one or several technology nodes.”

I view this as a concerned and decided to look into it further.

  • China’s chip manufacturing revenue was up 29% in 2017

    • Need for automation is expanding due to labor shortages

  • China is expcted to be fastest growing region for chip manufacturing over the next 5 years

  • The Chinese government is promoting its “made in China 2025” initiative and investing in various technologies including semiconductors.

    • Capital from the state backed China IC fund is supporting growth in the country’s foundry industry

  • TSM is also expanding in China and it should help them gain domestic chip orders in that fast growing market.

    • Maintaining focus on technology leadership and operational excellence not volume.

    • Is migrating existing equipment from Taiwan to China which lowers initial capital outlay

    • Expect China capacity load to go from 6.7% to 8.6% by end of 2019


The risk of Chinese supply “flooding” the market and putting pressure on TSM could be significant.  (Look at what the Chinese did to steel markets in 2014-2016). This will need to be going forward. I believe TSM’s chip technology lead will help them maintain market share but there is no guarantee it would not hurt a large percentage of their more “commoditized” product suite.




While TSM’s historical track record and current position as market leader are enticing, their stock price has appreciated in step.   An additional risk item for TSM is their valuation shows little margin for safety. I generally use two models in valuation. An ROE based dividend discount model and a free cash flow growth model.   In both cases I use a 9.5% discount rate to account for cyclicality of the business, my concerns about porter’s concentration and the risk of potential Chinese oversupply.

Model #1: ROE DDM

I assume a 23% ROE which is lower than their 10, 5 and 3 year averages around 25%  but assumes a conservative approach coming off a strong cyclical run in semiconductors and is closer to their current run rate.  I use their current run rate dividend payout ratio of 50% and a terminal multiple of 14x given their capital intensive business.  The model values the company at.96x intrinsic values right now.

Model #2 Free Cash Flow:

In this model, I assume a long term free cash flow growth of 5%.  5% is well below their current free cash flow growth rates but I am taking into account the highly capital intensive nature of the business and potential cash flow leakage from any slowdown in the cycle.  One factor that could cause concern is that over their history only about 70% of their earnings has actually come through to free cash flow. This is explainable by the constant investment cycle into the technology platform combined with a high growth rate.  You need to believe that: 1. The company can keep generating ROIC on their investments and 2. In a down market they turn down the capex in order to maintain margins and cash flow. Using a 5% free cash flow growth rate I come to price of .98x intrinsic value.


I do not hold a position with the issuer such as employment, directorship, or consultancy.
I and/or others I advise do not hold a material investment in the issuer's securities.



Taiwan Semiconductor’s historical success and market leading position are enticing for an investment.   It has some elevated risks from a concentrated customer base, supply dynamics with the Chinese government backing new foundries, and little margin for safety in their valuation.   I am still an investor here. I believe the capex investments of the last few years are good money investments and set to provide a runway for the next 4-5 years of company growth.  They hold a clear technology advantage over the competition that will enable them to continue their recent trend of successful sales, earnings and cash flow growth. This will support continued dividend growth and from a yield perspective valuation will move in line. I believe the company is a best of breed operator who has a track record of success and the balance sheet to survive and thrive out of any downturn. I am a believer in the long term secular trend of increasing need for digital computing power globally.  Companies with this level of historical success and market dominance are generally not going to be bought as deep value stocks. This has more of a GARP (growth at relative price) feel to me and while the margin of safety is small, I used very conservative numbers in my IV modeling and I don’t believe you are overpaying for a company that is set to continue its success for the next 3-5 years.

    show   sort by    
      Back to top