The Semiconductor Edition
On chips, automotive, and the future
Ryan Anderson (RJA) is making his fourth appearance in WITI after previously writing about Video Assisted Review, The Richest Game, and Counties. He is a marketing director based in Atlanta, GA, and has previously lived in Maine, Bermuda, Texas, LA, and NYC.
Ryan here. I have been deaf in my right ear since I was eight years old. This was the result of a benign growth called a cholesteatoma that destroyed my eardrum and inner ear bones. In high school, I decided I wanted to be an electrical engineer so I could help make speakers and hearing aids for people with similar issues. While that only lasted 3 semesters (a co-op job quickly dissuaded me from pursuing that line of work), I did get to spend hours learning how to design integrated circuits that could end up on semiconductors.
Articles have started appearing in the press recently about a semiconductor shortage, and what that may mean for the economy. “Semiconductor” is a broad term for a miniature electronic device also known as an integrated circuit. They are essentially small, purpose-built computing engines, very good at handling specific commands. There are 4 major types of semiconductors: standard chips (few tasks done well, like a checkout scanner at the grocery store), memory chips (RAM and ROM), microprocessors (CPUs, the brain in your computers), and the much more complex “system-on-a-chip” of SOCs. Within that SOC category are microcontrollers, which are like microprocessors but add their own dedicated memory and peripherals.
Credit: Photographs in the Carol M. Highsmith Archive, Library of Congress, Prints and Photographs Division.
The industry hardest hit by the shortage right now is automobile manufacturing. Volkswagen, Nissan, GM, and Ford have all announced significant production cuts. Ford has projected as much as a 20% drop in production this quarter due to the shortage. As a comparison, automobile industry sales only fell by an estimated 15% in 2020 as a result of the worldwide pandemic.
The leading cause of this shortage is that nearly 70% of microcontroller units that end up in cars rely, at least in part, on semiconductors from a single Taiwanese company: a company so in demand that they can make car manufacturers, a multi-trillion dollar industry, wait.
Why is this interesting?
Taiwan Semiconductor Manufacturing Company (TSMC) was founded in 1987 by Morris Chang, a former TI executive. TSMC’s big innovation was decoupling the manufacturing process from the design process. Historically, semiconductor companies like Intel, TI, and AMD designed their chips in-house and then built the manufacturing processes to produce them at scale. By being vertically integrated (owning and controlling multiple parts of the value chain), companies were able to leverage their expertise in a new market that wasn’t yet built for specialization.
TSMC’s bet was that, as the market for semiconductors grew, there would be more companies who needed custom chips than could afford to build a foundry. Foundries cost billions of dollars and years to construct, making them far outside the budget of all but the largest companies. One key reason for this is the level of control and precision required. Go to the 47:45 mark of this video to see a TI foundry at work.
"A speck of household dust landing on one of these delicate silicon wafers would be comparable to Mount Everest landing in the streets of Manhattan"
TSMC would fill that gap with what is known as a fabless foundry, or a manufacturing plant that can take another company’s designs and turn out millions of chips quickly and cheaply. By being totally focused on manufacturing expertise, TSMC and other fabless foundries can invest billions of dollars every year into R&D and manufacturing processes that the thousands of companies who rely on semiconductors can benefit from. It’s been so successful that TSMC now manufactures chips for major tech companies like Nvidia, AMD, and even Apple’s new custom A14 (mobile) and M1 (laptop) chips.
Chips from those 3 companies are incredibly valuable. A top-of-the-line Nvidia graphics card costs almost $2,000 and AMD CPUs can run over $600. Apple’s costs aren’t as clear since they don’t sell the component, but those chips enable them to sell $1,000 phones and $3,000 laptops.
The automobile industry operates differently. The initial buyers of the semiconductors are companies like Renesas and NXP who design the microcontrollers a car relies on. Automobile manufacturers then buy the microcontrollers from these suppliers based on factors like supply chain, performance, and price. Naturally this additional step in the chain has the effect of pushing down TSMC’s margins on these semiconductors versus selling products directly to a higher margin buyer like Nvidia or Apple. Because of this value chain, TSMC may end up charging $1 for a semiconductor that a $40,000 car relies on. As a manufacturer who has more demand than they can fill, TSMC is going to prioritize the high-value chips. Meaning these car parts, which only account for 3% of TSMC’s revenue, go to the back of the line. (RJA)
__
Partner Post: WITI x The Browser
With the flood of newsletters, there’s a lot of similar looking links floating around. Enter The Browser. It is a beautiful synthesis of articles you won’t find elsewhere, which opens your brain and your references. Like what we aim to do with WITI, the Browser team are obsessed with quality first and the finest rather than the latest. So there’s no peg to trending topics, but rather high signal, good links. We highly recommend subscribing. Sign up with 20 percent off with a special WITI code by clicking here and using WITI20 at checkout. (CJN)
__
International Radio Station of the Day:
Every green dot you see is a radio station that you can stream right within the browser. Escape your apartment with a little international flare at radio.garden(RJA)
Quick Links:
There is another important wrinkle to the semiconductor story and the geopolitical risks associated with so much critical manufacturing happening on a small island just off the coast of mainland China. (RJA)
Related, a blast-from-the-past NYT article from 1993 titled “Turning the Desktop PC Into a Talk Radio Medium” (RJA)
Thanks for reading,
Noah (NRB) & Colin (CJN) & Ryan (RJA)
Why is this interesting? is a daily email from Noah Brier & Colin Nagy (and friends!) about interesting things. If you’ve enjoyed this edition, please consider forwarding it to a friend. If you’re reading it for the first time, consider subscribing (it’s free!).