Description

Taiwan Semiconductor Manufacturing Company Limited (Symbol TSM, but usually referred to as TSMC)

August 14, 2015 (Billions, except per share data) (Bloomberg Est.) 2015           2016     

Price: $20.38 (52 Wk H - $25.77, L - $19.39) EPS: $1.82         $1.93

Shares O/S: 5.2 Revenue:                $26.9        $30.1

Market Cap: $106.0 EBITDA: $17.3         $19.4

Net Debt: -7.8 B ($1.50 per share net cash)

Enterprise Value:  $98.5

Investment Thesis Summary:

TSMC currently trades for less than 10x forward earnings, net of cash, yet it should grow at twice the rate of the semiconductor industry.  Moreover, the company’s competitive position vis-à-vis Intel and Samsung continues to improve, and the company has entered a period of improving Free Cash Flow (FCF) during which the dividend should double to $1.45 per ADR by July 2018 even if earnings do not improve to the $3.00 level that I expect.  Finally, TSMC’s business model has proven very resilient.  Even if industry fundamentals deteriorate, the company’s FCF would still remain in the $1.50 - $2.00 range, as capital expenditures would likely fall 50%.   

Company Description (mostly from company material):

Established in 1987, TSMC is the world's leading independent semiconductor foundry. As the founder of the Dedicated IC Foundry segment, TSMC has built its reputation by offering advanced and "More-than-Moore" wafer production processes and unparalleled manufacturing efficiency. From its inception, TSMC has consistently offered the foundry segment's leading technologies and TSMC COMPATIBLE® design services. The company estimates that its revenue segment share among total foundries worldwide was 54% in 2014.

TSMC has consistently experienced strong growth by building solid partnerships with its customers. IC suppliers from around the world trust TSMC with their manufacturing needs, thanks to its unique integration of cutting-edge process technologies, pioneering design services, manufacturing productivity and product quality.

As of December 31, 2014, the company's monthly capacity (in 300mn equivalent wafers) totaled 736,375 wafers, compared to 644,650 wafers at the end of 2013.  TSMC operates one 150 mm wafer fab, six 200 mm wafer fabs, and three 300 mm wafer fabs.  Eight of its ten fabs are located in Taiwan, one in the United States, and one in China.  TSMC also obtains eight-inch wafer capacity from other companies in which the Company has an equity interest.

TSMC is listed on the Taiwan Stock Exchange (TWSE) under ticker number 2330, and its American Depositary Shares trade on the New York Stock Exchange (NYSE) under the symbol "TSM.”

Financial Review:    

A broad long-term review of TSMC’s performance highlights strong, steady revenue and profit growth.  Using Street estimates for 2015, revenue has grown over a ten year period from $8.3 billion in calendar 2005 to $27.1 billion this year (calendar 2015), or 12.6% annualized.  Net income increased from $2.9 billion to $10.2 billion, 12.7% annualized, over the same period.  During that period, the company experienced only one yearly decline in revenue, a 15% decline in 2009, and two yearly declines in net income, a modest 5% decrease in 2008 followed by a 15% decrease in 2009.  Importantly, FCF (OCF – Cap ex) improved some $3 billion, or 60%, in 2008, as the company ratcheted back on cap-ex and improved working capital.  Unlike most other semiconductor companies, TSMC grows organically, rarely takes a write-off, and has maintained its share count virtually flat over this ten year period without stock repurchases.

TSMC’s strategic position and financial performance are improving.  This is because product cycles continue to accelerate for the most important products (mobile device innovation dwarfs that of the PC, e.g.), and the cost of advancing to the next process node in the foundry business continues to increase; TSMC has spent approximately $28 billion on capital expenditures the last three years, and that number is only going up.  Moreover, TSMC’s strategy of pushing the technology envelope while unequivocally remaining out of the business of its customers leaves it the only clear production choice for leading semiconductor companies.  Nevertheless, Wall Street has determined that Intel’s entry into the foundry business, Samsung’s success as a foundry for Apple, progress by GlobalFoundries and other much smaller fabs, and slowing advancements in chip manufacturing threaten TSMC’s competitiveness.  The evidence strongly suggests otherwise.

In 2014, Gartner reported that TSMC grew 25.1% versus overall market growth of 16.1%.  TSMC’s foundry business remained 5x larger than UMC and GlobalFoundries and almost 9x Samsung’s foundry business based on Gartner’s numbers.  Through June 2015, TSMC increased revenue 29.1% y/y.  The company also reported that July revenues improved 24.7% y/y.  Clearly, the company continues to take share through July.  The bugaboo for the street is that Apple has purportedly selected Samsung and its partner GlobalFoundries to produce the lion’s share of the Apple A9 processor for the iPhone 6S, expected to ship late September.  Qualcomm has also reportedly selected Samsung to produce its latest generation Snapdragon processor.  Although not confirmed, these strategic losses would lead to slowing revenue for TSMC in the second half of 2015.  The street projects $6.7 billion in revenue and $.44 eps for September, down from $7 billion in revenue and $.49 eps in 2014 Q3, and a further decline to $6.6 billion in revenue (down from $7.2 billion in 2014) and $.41 (down from $.50 in 2014) in December.  Although I am skeptical that Apple gave this much business to Samsung, this would likely be the worst case and still implies $1.82 in eps in 2015, up from $1.69 in 2014.  

Notwithstanding the potential second half slowdown, TSMC’s management maintains that it can grow revenue by at least a 10% CAGR from 2014-2019 and improve cash flow at a faster rate.   I project a FCF CAGR of greater than 20% from 2014-2019.  Management has observed that balance sheet cash levels are sufficient and that any improvement in FCF will be paid out in dividends.  Taiwanese companies historically are less concerned about steady increases in dividends and more focused on returning excess cash when it is earned to shareholders.   TSMC’s board historically has proposed a dividend once a year, usually in February.  Shareholders need to approve the dividend, which usually occurs in June, leading to a July/August distribution.  The company paid a $.72 dividend in July 2014, the first meaningful increase since 2006 and 44% above the July 2013 dividend of $.50.  I anticipate that the board will propose a dividend in the $.90-.95 range in February 2016 even if the second half slowdown occurs.                          

Cap-Ex Drives Growth and Massive Cash Flows

With approximately 65% EBITDA margins, 10% operating expenses, and a 12% tax rate, additional revenue clearly generates significant cash for TSMC.  A leading outsourcing chip manufacturer of course requires superior R&D, responsive customer service, and precise quality control to secure business, but to deliver the huge scale required for top OEMS, TSMC most importantly must spend on capacity.  The company has spent some $60B on cap-ex the last ten years.  The table below highlights that it takes cap-ex to grow in this business, but with an approximate two-year payback that is not such a bad thing.  Periods of significant growth usually see capital intensity (cap-ex/revenue) levels of around 50%. During periods of modest growth, this falls to 30% or less.  In the 2005-2009 period, capital intensity approximated 25% while growth was modest.  FCF over that period exceeded operating income.  The massive investments made in 2010-2013 (capital intensity averaged 47.9%) allowed the company to almost triple revenues.  As the semiconductor industry has matured, an almost ideal situation faces TSMC.  Semiconductors should continue to grow 4-6% with foundry growing somewhat faster.  This will allow capital intensity to decline to 30% or less and FCF to approach and possibly exceed earnings.  Yet, TSMC will continue to spend over $10 billion a year in cap-ex.  This will provide an almost insurmountable hurdle for new entrants or aspiring competitors.  Moreover, leading OEMS, as well as emerging technology companies will have to look to TSMC for foundry services, as it will possess superior process technology and service, as well as strategic neutrality.

TSMC 2005-2015 Capital Expenditures versus Operating Metrics

Strategic Analysis:

In light of projected y/y earnings declines the next two quarters at TSMC, why do we have such confidence the company will continue to grow 2016-2019?

First, the projected declines for the next two quarters partially stem from extraordinarily hard compares and horrendous PC chip demand.  TSMC grew revenue 37% 2H 2014, as it gained meaningful market share at Apple.  Prior to 2014, Apple was not a foundry customer of TSMC, relying on Samsung for mobile processors and Intel for Mac processors.   However, as Apple became a greater share of the chip market, it had no choice but to look to TSMC for foundry capacity.  Apple clearly needs to guarantee availability of chip manufacturing at the leading edge, and Samsung fiercely competes with Apple in the mobile market.  Strategically, Apple must provide TSMC adequate business to keep honest Samsung and Intel, who competes with Apple, as well.  Moreover, Apple has a stated policy of bringing more strategic chip design in-house.  TSMC will clearly spend to capture these new opportunities, but also has a record of ensuring that it earns a superior ROIC.   

Second, TSMC’s strategic position is improving relative to competitors, not declining.  On its second quarter earnings conference call, Intel indicated much lower cap-ex plans and a slower progression, 2.5 years instead of 2 years, to the next process node (10 nanometers (nm)).  Intel also recently agreed to acquire Altera, its only potential non-X86 architecture foundry customer; Altera still utilizes TSMC for all production chips.  Reportedly, one of the reasons for Intel’s acquisition is that Altera had decided to retain TSMC as its foundry.  Intel’s competitive position relative to TSMC’s continues to deteriorate.  Historically, Intel would lead TSMC by 1.5 to 2 years to the next process node.  By 2017, TSMC is expected to be ahead of Intel and even with Samsung.  As highlighted above, success in this business is determined by proper strategic cap-ex spending.  Intel lowering its cap-ex to $8 billion in 2015 while TSMC continues to spend over $10 billion speaks volumes.  In addition, GlobalFoundries has reportedly experienced yield difficulties ramping Apple’s A9 processor despite its strategic alignment with Samsung.  This is not surprising, as GlobalFoundries, the company created via AMD’s manufacturing spinoff, has consistently experienced difficulties gaining scale and delivering services.  It has tried to accomplish this with acquisitions of Charter Semiconductor in 2010 and recently IBM’s foundry business in July.  Struggling financially, It is hard to imagine that over time GlobalFoundries can compete with a 50% plus market share player spending in excess of $10 billion a year.  Finally, Samsung has signaled some moderation in total cap-ex, as well, plus must support memory, Flat Panels, and internal consumption in addition to external foundry customers with its approximate $14 B in annual cap-ex.  

TSMC has maintained its $10.5-$11B target for 2015, despite only spending $3.1 B in the first half.  Thus, TSMC now intends to spend $7.3-$7.8 B in H2 2015.  For a company that has a strong track record of driving revenue and cash flow growth with cap-ex, this is an insightful indicator that the company has visibility into meaningful growth opportunities.  The new business likely stems from additional wins from Apple, emerging Chinese mobile OEMS, and IoT players.  (TSMC has stated IoT is growing 50% plus.)  TSMC has highlighted that it must work with customers for 18-24 months before implementing a new node; thus the company has signaled stronger growth with this planned cap-ex spending and its guidance for revenue growth from 2015-2019. Most keen observers (and TSMC management) of the foundry industry acknowledge that Samsung and Intel have advanced to 14 nm ahead of TSMC.  Yet, TSMC continues to deliver a very competitive 16 nm offering and gain share at the profitable 28 and 20 nm nodes (still the largest revenue generators) plus is poised to become the leader in manufacturing at the 10 and 7 nm nodes. This is a remarkable accomplishment since the company has historically pursued a fast follower strategy with superior customer service.  Now as the foundry industry matures, TSM should lead technologically, possess the most scale, and offer strategic independence to the largest OEMs as well as leading emerging technology companies.  Highlighting the company’s financial strength, TSMC should more than cover the massive 2H spend out of Earnings + depreciation ($4.4B + $3.75B).   Moreover, the approximate $5B of FCF of 1H comfortably covers my projected 2015 dividend.  Finally, in a move that further highlights the company’s savvy capital allocation, TSMC should realize approximately $750 million in equity gains in 2015 from investments in suppliers (ASML + Vanguard).  

Conclusion:

It is not often that you have the opportunity to purchase at 10x enterprise value to earnings the dominant player with a stellar track record and improving strategic position in a critical industry that possesses no visible secular challenges.   TSMC offers that opportunity because, notwithstanding superb performance the last several years, the company faces two quarters of tough compares during an inventory correction for the semiconductor end market and a period of temporary lost market share at Apple.  For TSMC, a superior technology roadmap and strategic position concurrent with moderating industry capital expenditure intensity (but still daunting dollar wise for competitors) should support several years of strong revenue and cash flow growth.  Earnings should grow from $1.80 this year to the $3.00 range the next several years.  Over that period, FCF should approach or exceed earnings despite continued revenue growth.  As a result, the dividend should double over the same period to approximately $1.50.  This performance should bring about a rerating of the stock and cause it to trade to at least $40.00 per share.       

Catalyst

  Dividend increase, bottoming of semi cycle, clarity on TSM's positioning vs. Samsung re: Apple