Select Page

Announcing a new feature–my stock pick of the day. With Taiwan Semiconductor as my first pick.

The hoo ha at the end of last week over restrictions on the sale of U.S. semiconductors to China’s Huawei Technologies, high-lighted the central role that Taiwan Semiconductor Manufacturing (TSM) plays in the global technology sector. The stock jumped on news that the company would build a new foundry in Arizona on the assumption that the new plant would let the company avoid U.S. restrictions on technology sales. That assumption got a little shakier later in the day when officials at the Commerce Department clarified that the restrictions covered the sale of any chips to Huawei that had been manufactured using chip making equipment from U.S. companies such as Applied Materials (AMAT.)

The sudden attention paid to Taiwan Semiconductor reminded every one that there are, for all intents and purposes only three chip manufacturers in the world, Intel (INTC), Samsung Electronics, and Taiwan Semiconductor. And of these Taiwan Semiconductor Manufacturing is the only major independent chip foundry. If you don’t want to have your chips manufactured by a potential competitor such as Intel or Samsung, then Taiwan Semiconductor is your default choice. Which is why the company makes chips for everyone from Qualcomm (QCOM) to Apple (AAPL) to Nvidia (NVDA) to Advanced Micro Devices (AMD), to, yes, Intel.

In crucial ways Taiwan Semiconductor has acquired the chip making leadership position that used to belong to Intel.

That’s because chip foundries are incredibly expensive propositions. The cost of Taiwan Semiconductor’s Arizona foundry is estimated at around $12 billion from 2021 to 2029 (with first production projected for 2024.) Unless you make a lot of chips that kind of capital spending level is prohibitive.

And building the foundry itself isn’t the end of spending. Foundries are under constant pressure from their customers–or from in-house chip makers–to drive chip manufacturing processes to smaller and smaller scales that allow chip companies to pack more and more transistors–and therefore processing power–onto single chips. Nvidia’s recently announced Ampere A100 chip packs 54 billion transistors onto a chip with a die size of just 826 square millimeters. That’s way more than the 21.1 billion transistors that Nvidia’s GV100 chip packs onto a 815 square millimeter piece of silicon. (For technical reasons Nvidia couldn’t make the chip much larger since current chip etching lithography equipment can’t handle a size bigger than 850 square millimeters.)

How do you pack 2.5 times more transistors on a chip that’s only 1.3% larger? You build a production line that’s able to etch the transistors much closer to together. The current technology frontier is now 7 nanometers. Samsung has it. And so does Taiwan Semidconductor where customers such as Advanced Micro Devices, Apple, and now Nvidia use it to make their most cutting edge chips. The next step is 5 nanometer manufacturing. If you’re a chip company like Nvidia or Apple, you want to do business with the foundry that has the newest, smallest technology. Being on the cutting edge of power and speed is what lets a company like Apple charge a premium for its products.

And if you’re a company such as Taiwan Semiconductor you want to use the revenue from all those chip makers to fund the investment in those new production lines that will keep those customers sending you orders for their newest chips.

And you can afford to fund that capital spending because you have all those customers. It’s a very clear virtuous cycle. But it’s only available to a few companies with massive scale. Intel has struggled to gain share in the independent foundry market since so many potential customers see it as a competitor. New foundry startup GlobalFoundries is still working to build competitive scale. Customers have no compunction in switching foundries if the price is right–if the manufacturing process is comparable. Right now at the manufacturing end it’s a two-horse race between Samsung and Taiwan Semiconductor.

The next leap is the move from the 7 nanometer manufacturing process now on line at Samsung and Taiwan Semiconductor to 5 nanometer manufacturing at both Samsung and Taiwan Semiconductor.

Looking out over the 7 and 5 nanometer horizon it looks like Taiwan Semiconductor will see strong demand for those leading-edge manufacturing technologies from the smartphone companies as they begin 5G deployment, and for cloud server and artificial intelligence chips from Nvidia, Advanced Micro Devices, and Xilinx. Morningstar projects that trend will result in 9% annual sales growth over the next 5 years. Gross margins, Morningstar calculates, will remain in the high 40% range but the ramp to more complex 5 nanometer chips and the continued expense of research and development, and capital spending will keep operating margins in the high 30% range.

Taiwan Semiconductor has been a member of my long-term 50 Stocks Portfolio since October 7, 2019. The position is up 5.96% since that initial pick. (You can see all my picks in the 50 Stocks Portfolio on If you haven’t already created an account, you’ll have to do that by giving us your email. But no credit card. Access is free and you’ll also get my daily email alerts at 8 p.m. each day in your email box. By the way, we never sell or trade your email address.)

The stock closed down 2.04% yesterday May 19 at $50.81. The February 12 2020 high was $59.63. The March 11, 2020 low was $43.89.

All support and resistance is in a narrow range at the moment. The 200-day moving average is at $51.24 with the 50-day at $50.24 and the 20-day at $52.18.

About 14% of Taiwan Semiconductors 2019 revenue came from Huawei and the latest restrictions on sales to the Chinese company could send a ripple or two through Taiwan Semiconductor’s stock price. I think the company has a broad enough stable of customers to replace that revenue, but the transition might still send the stock down for a bit (regardless of what might happen in the wider market and growth in the global economy.) Any Chinese retaliation for the Huawei restrictions could slow growth at Taiwan Semiconductor’s customers. If I could get the stock on weakness near $46 I’d be  buyer for long and medium term portfolios. (I’d call my 12-18 month Jubak Picks Portfolio medium term.)

Please note: As in this case a Stock Pick of the Day is not always a recommendation to buy on the day of the pick. Check market conditions and your assessment of market trends. The Pick of the Day is a stock that I find attractive on the fundamentals, but not one that I would buy at any price. In other words, if you agree with my assessment of the day’s stock, use your judgement on when to buy. I will give you my best assessment of timing for a buy at the end of each post.