Pay attention to the months of October and April
The months of October and April are known as the most dangerous months in Taiwan. These months are when the waters are calm and best suitable for an amphibious assault from invaders.

US Speaker Nancy Pelosi will arrive in Taiwan within a few hours from now (as of time of writing). Scrolling through Twitter, I've noticed that over the past few days alone, there's a surge in the number of videos of Chinese military vehicles entering the Fujian province, the Chinese province that is directly across from Taiwan. These videos reminded me of the months of videos of Russian tanks and military vehicles going to the Ukrainian border before the invasion.

While the media focuses on Ukraine, few people are paying more attention to Taiwan. If war happens, the world will be impacted in a detrimental way. Taiwan has 61% of the world's capacity to produce chips that are 16mm and smaller. Having the majority of chip production offline will make it difficult for the world to produce more electronics. And with tech being the main driver of global economic growth, seeing the sector grind to a halt could bring the world into the Great Depression.

$TSM is going to be the biggest business casualty as nearly all of its foundries could get destroyed if an invasion happens. $UMC will be another casualty.

As for $INTC $GFS and other foundry companies that don't have factories in Taiwan, they will see a surge in business from $NVDA $AMD $AAPL etc. But because their capacity is a lot lower than TSMC, these fabless chip companies will either pony up and pay a lot more to have their chips produced or risk not having any foundry willing to produce their chips when they can gain more profit producing chips for another firm.

$ASML $LRCX and other semiconductor equipment companies will see surging demand as foundry companies aggressively ramp up investment in building more factories.

While Pelosi's visit won't trigger an immediate invasion, it could inspire China to ramp up its military capacity in the Fujian province to the point where they are ready to launch an invasion in October.
Dylan Patel's avatar
$104.3m follower assets
Intel Is Throwing The Kitchen Sink, But Is The Turn Around
Plan Reasonable?
Deep dive on Tower Semiconductor Fabs/IP
Intel Culture Shift
Future product and roadmap competitiveness by business unit

I mapped out Tower Semiconductors capacity. I also wrote a
lot about their specific differentiated technologies.
In short, Intel acquired ~2 million wafers per year of a lot
of niche technologies and great people to help accelerate their foundry push.

I also wrote a lot about the culture shift at Intel, including a really great story about Intel and how they treated various semicap firms (poorly).
This specific story is Applied Materials, but tons of other horror stories I've heard the supply chain.

Something that shouldn't be underestimated is accountability throughout the organization.
OKRs were invented by Andy Grove, Brian Krzanich removed them, and Pat Gelsinger has brought them back!

I also discussed a lot about Intel's capital allocation strategy, spending plans, and fabs.
There's a particularly interesting potential partnership and MOR signed with Brookfield Asset Management, a real estate, private equity, and alternative investment fund.

Not all chip foundry stocks should be treated the same
The media focuses a lot on Russia and Ukraine and less on China and Taiwan.

Taiwan is the big hub for chipmaking. Companies like $TSM and $UMC produce the majority of their chips in Taiwan. While these companies contribute the majority of chip manufacturing to the world, if ever China decides to invade Taiwan, these companies will have to halt production. Then, the rest of the world have to do business with foundry businesses outside of Taiwan.

$GFS, the second-largest chip foundry company (TSMC is #1), has its foundry factories in the US, Germany, and Singapore. They are nowhere to be found in Taiwan.

$INTC, another major foundry company, has many locations in the US, Latin America, Malaysia, Vietnam, and Israel. Unlike GlobalFoundries, Intel also has foundry factories in China, which also puts them at risk if ever China decides to enact policies against the exporting of chips to the US.

With tensions in Taiwan rising, I'm becoming less comfortable with investing in Taiwanese chip makers. I prefer a chipmaker like GlobalFoundries or Intel, which are more diversified geographically or have many factories in the US.

Thoughts?
Dylan Patel's avatar
$104.3m follower assets
Great article. I bought more $TXN today. Great earnings and I especially appreciate that they do spend more on Capex and expect to do so in the future. Reinvestment into the business is more imporant than payouts to shareholders/buybacks.
View 5 more comments
Dylan Patel's avatar
$104.3m follower assets
GlobalFoundries $GFS IPO – Mubadala Lost Over $22.4B, Now They Are Hoping Public Markets Bail Them Out
Low utilization rates, countless divested assets, cut R&D/capex, and still doesn't make $ in the greatest semiconductor boom of all time.
SemiAnalysis believes Mubadala should've rolled the regulatory dice in trying to let Intel acquire the entire firm for a price that prevents them from losing a lot of money. Many investors will puke on this IPO.
i started covering $AMD when it spun off $GFS. it was clear then, and still is, that $AMD made the right move.

it took them a few years to extricate themselves from the purchase commitments they made as part of the deal, but it was worth the pain. even when $AMD was trading at ~$2.10, and they were kind of on the ropes in terms of delivering their roadmap, Mubadala was one of their biggest shareholders.

$AMD managed probably one of the most amazing turnarounds in semiconductor history with the hiring of Jim Keller (legendary chip designer) to rebuild the design pipeline, and Lisa Su as CEO to steer the company into a new era.

still, to your point about not making $ in the global semi boom, why $GFS failed to eek out some success is mind-boggling. the need for more semi capacity, more competition (it's really only TSMC), and more foundry players, could not be more apparent today.
View 1 more comment
$TSM - Company Overview
$TSM: Fueling the Future of Computing

Taiwan Semiconductor Mfg. Co. (TSMC) has become the leading manufacturer (fab) of semiconductor chips by capitalizing on a series of egregious missteps at $INTC

This is not a buy rec, but rather info for when it's actionable

A thread👇

1/ The Basics

The first "pure-play" foundry -> enabled the rise of the "fabless" (i.e. design-only) industry

A foundry ("Fab") manufactures microchips (integrated circuits "IC") for customers such as fabless chip companies.

Fab only = sole focus on better process technologies
Image

2/ First Mover

The leader in chip mfg. is the fab that can get to the next “process node” first

Process Node (“Tech”): Used to indicate feature size of transistor - > now marketing tool for chip generation

As @gavinsbaker wrote in Jan '19, TSMC passed INTC at the 7nm node

3/ Market Overview

All electronics = powered by semis

3 Types of Companies:

Integrated Device Mfgs. (“IDM”): $INTC, KOSE: A005930, $TXN, $MU

Fabless: $NVDA, $AMD, $QCOM, $AVGO

Fabs: $TSM, $UMC, $SMICY, GF

Market Size: Semis power all electronics…

4/ Technology Advantage

Tech dominance while other foundries can’t keep up (GF dropped out at 7nm)

Tech advantage -> customer additions ($AMD from GF)

1) 5nm already in production
2) Specialty tech (5G)
3) Advanced Packaging

TSM: 52% market share

Manu leader + Ecosystem

5/ Technology Platforms

Automotive Electronics: IP ecosystem & RF tech driven by autonomous & EV

High Performance Computing: Cloud datacenters and communication infrastructures

IoT: Fueled by wearables, smart homes / cities / industries

Smartphone: Mobile proliferation

6/ Q4 2020 Financial Results

Accelerating Rev growth YoY: 16%, 22%, 24.2%

Expanding EBIT %: 35%, 37.7%, 40%, 41.3%

2x FCF YoY

Accelerating EPS Growth YoY: 24.2%, 43.5%, 49.2%

ROA: 10.7%, 12.1%, 13.1%, 14%

ROE: 20.9%, 23.4%, 27.8%, 29.1%

7/ Capital Expenditures

2021 Guidance: $25B-$28B vs. 7B in 2020 (47% - 65% YoY Increase)

LT Capital Intensity: mid-30%
Accelerates CapEx spend to prep for higher growth

2021 Capital Intensity est.: ~48.5%

On conf. call: to support expected higher Revenue CAGR next 5 years

8/ Revenue Composition

5nm shipments accelerated in Q4 – now 20% of Rev

Advanced Tech (< 16nm tech) = 62% vs. 56% Q4 2019

Smartphone continues to dominate

Main HPB growth areas: CPU, networking & AI accelerator
  • Could CPU growth be from $INTC?

HPC - main growth driver

9/ Next Step: N3

2021 Rev Driver = HPC & Auto

HPC: “We see a stronger innovation is coming our way on N3 as well as on N5.”

N3: Volume production 2H 2022

N3: 70% logic density gain, up to 15% performance gain & up to 30% power reduction vs. N5 (175MM transistors / mm^2)

10/ Future R&D Leadership

3nm logic tech platform & apps (2021): 6th gen 3D platform

Beyond 3nm by 2023 – 2nm in pipeline

Projects: Beyond-2nm node, 3D transistors, new memory, etc.

Focus: novel materials, processes, devices, nanowires & memories (+8-10 yrs away)

11/ Conclusion

$TSM is fueling the world of ubiquitous computing and interconnected devices

Has overtaken $INTC in mfg. leadership -> economic benefits

Positioned to benefit and capitalize on hyper-growth trends: AI / autonomous driving, 5G, IoT

12/ Resources


Anything written by @foolallthetime (Twitter)


TSMC 2020 Q4 Quarterly Results -
post mediapost media
Love the write-up + super informative.

Is it unlikely that fabless companies eventually move into developing wafer foundries of their own? I get that one advantage to being Fabless is you can spend more on R&D + maintain high production volumes but is there not enough advantages to be vertically integrated for them to push their suppliers out?

(note I've done only a bit of research here so coming from a beginner's POV)
View 4 more comments
Next
Commonstock is a social network that amplifies the knowledge of the best investors, verified by actual track records for signal over noise. Community members can link their existing brokerage accounts and share their real time portfolio, performance and trades (by percent only, dollar amounts never shared). Commonstock is not a brokerage, but a social layer on top of existing brokerages helping to create more engaged and informed investors.