Regarding… Vol 3 no 3 – Single point of failure

Dear Fellow Investors and Friends,

Welcome to my investment (and other) musings. For those of you who are new, this letter is where I share my efforts to make sense of the world around me.

I do appreciate you taking the time to read this.

Today is Thursday, January 30th, the 30th day of the year (fortunately, it’s the last time I will use that particular redundancy this year). There are 335 days left until the end of the year. Yes, the first month of the year is just about over. I hope you’ve managed to keep your New Year’s resolutions going!

Thirty-nine years ago, on January 26 1986, the Space Shuttle Challenger exploded shortly after launch.

I clearly remember watching the inferno on TV. I had just started working that year, and when I came home, exhausted, I slumped on the couch in front of the TV to watch the 6 o’clock news, as one did in those long-gone days.

The images of the explosion were the first thing I saw on the news.

But the news didn’t tell us the back story of why it exploded shortly after take-off.

It was a cold morning on the day of the launch, and the engineers at Morton Thiokol, the contractor responsible for the shuttle’s solid rocket boosters, had serious concerns about the low temperatures affecting the O-rings. These were critical components that sealed joints in the boosters. They were designed to prevent hot gases from escaping during launch but had never been tested in such cold weather.

Despite these concerns, NASA officials decided to proceed, believing that the existing safety measures were sufficient. The decision was influenced by a culture that prioritised schedule adherence over safety concerns.

Seventy-three seconds after liftoff, the O-rings failed, hot gas escaped, and the shuttle exploded.

This week’s news about DeepSeek got me thinking about the risks faced by the semiconductor (chip) industry. The biggest risk is not a cheaper, better AI algorithm but a lack of redundancy.

Redundancy is an integral part of engineering system design. It means including extra components or systems that serve the same function as others so that the system can operate without interruption if one part fails.

A lack of redundancy is a design fault you want to avoid.

In systems, centralisation is efficient but can have poor redundancy. Decentralised, distributed systems are less efficient but have much greater redundancy.

In industries with significant benefits to scale, incumbents can do one of two things.

• Internalise scale benefits, use their super-profits to buy up the competition, and lobby politicians to create a virtual monopoly. Growth and profitability are high. Google/Alphabet is the prime example here.
• They can share these benefits with their clients, forgoing additional profits per unit, but they can scale by taking greater market share by underpricing their competition. Costco is the poster child here. Shoprite in South Africa is also following this strategy.

Technology is an eminently scalable industry, so companies often face this choice. Most choose the first route, leading to a winner-takes-all outcome, which is why almost everything in the tech industry is centralised. A single dominant network can benefit all participants enormously if pricing is kept in check.

How many search engines do you use? Telecoms providers? Online movie channels? Mobility apps?

In centralised tech, the product is often so good that it is indistinguishable from magic. Can you imagine life without your smartphone and its apps? Remember when you had to drive to the video shop to rent a movie to watch on a Sunday night? Remember when you had to drive to the grocer to do your monthly shop?

Increased customer satisfaction is reflected in the share prices of the companies that bring us this fantastic tech: Meta, Apple, Alphabet, Nvidia, etc. The Magnificent 7 have been the main drivers of the bull market in US stocks – and the world – over the past 15 years.

The great centralisation of the tech industry has benefitted consumers and investors.

But, as they say, that’s not all, folks…

Not only has the tech industry centralised into silos, each serving an individual need to their customers,  but the underlying computer chips or semiconductors have also been centralised into a single provider.

TSMC is the most mission-critical company on Earth. As the largest pure-play foundry, it exclusively manufactures semiconductors for others without designing or selling chips under its own brand. By transforming innovative designs into physical chips, TSMC plays an indispensable role in the supply chain.

While TSMC has a 60% market share in foundries, its share in advanced chips is closer to 90%. Only TSMC and Samsung currently manufacture these leading chips. If we imagine an upside-down pyramid with TSMC at the bottom, a considerable amount of value rests on it.

If TSMC were to get damaged or disappear, suffering would be widespread.

The semiconductor industry has a dirty little secret – it uses lots of energy. According to this paper, semiconductor fabrication facilities are among the most energy-intensive manufacturing plants globally. In 2021, the industry consumed approximately 149 billion kWh, enough to power a city of over 25 million people for a year.

This energy consumption is projected to increase dramatically. Estimates suggest that by 2030, the industry could consume around 237 terawatt-hours (TWh) of electricity – equivalent to Australia’s total electricity consumption in 2021.

According to data from the Statistical Review of World Energy, Taiwan – where TSMC is based – relies on fossil fuels for 90% of its energy. Virtually all the oil, natural gas, and coal it consumes is imported.

What appears to be a dominant, highly competitive industry from the outside lacks redundancy, leaving it highly vulnerable to a risk almost no one ever discusses.

According to CNBC: “There were three major power outages in Taiwan in the past seven years, and the island has experienced a slew of smaller disruptions in the past year. According to local reports, as recently as April 2024, in Northern Taiwan alone, multiple power shortages were recorded over three days. In 2022, there were 313 power outage incidents. A big power outage that year affected more than 5 million households, while another massive blackout in 2017 hit almost 7 million households.”

Trillions and trillions of dollars of business value rests on one company. And this company is predominantly reliant on imported coal.

What could go wrong?

Deep what-what, you may ask. And I would answer, who knows? Not me. I don’t know my Dorito from my Blackwell. So I’m afraid I can’t add anything to the hoo-hah over “China-AI” except to speculate that it is quite possibly very similar to the “China virus” – in effect, if not substance.

I can point you in the direction of people who know far more about such things than I do.

Ben Thompson’s weekly column, Stratechery, is my go-to source for insights on the tech world. This week, he released a FAQ on the implications of DeepSeek. Yes, I know; even after reading the column twice, I still don’t understand most of it. But his conclusions were understandable (and interesting):

• Nvidia’s moats have been breached, and the stock faces much more uncertainty than it did before.
• The biggest winners are consumers and businesses who can anticipate a future of effectively free AI products.
• China is also a big winner, as their technical prowess is further established.
• Another set of winners are the big consumer tech companies. In a world of free AI, it’s product and distribution that matter most, and those companies have won that game.

Apropos that last point, @bespokeinvest on X today pointed out that the loss in Nvidia’s market cap was almost equal to the gain by Microsoft, Amazon, Meta, Alphabet and Apple:

Unsurprisingly, the consumer-facing tech companies are doubling down on their capex.

From Reuters:

“Days after Chinese upstart DeepSeek revealed a breakthrough in cheap AI computing that shook the US technology industry, the chief executives of Microsoft and Meta defended massive spending that they said was key to staying competitive in the new field.”

Of course, Warren Buffett and Charlie Munger have some common sense for us about the tech sector.

“The great lesson in microeconomics is to discriminate when the technology will help you and when it will kill you. And most people do not get this straight in their heads. There are all kinds of wonderful new inventions that give you nothing as owners except to spend a lot more money in business that’s still going to be lousy. The money still won’t come to you. The advantages that stem from great improvements will flow through to the customers.”
– Charlie Munger

“Technology is clearly a boost to business productivity and a driver of better consumer products and the like, so as an individual, I have a high appreciation for the power of technology. I have avoided the technology sector as an investor because I generally don’t have a solid grasp of what differentiates many technology companies. I don’t know how to spot a durable advantage in technology. To get rich, you find businesses with a durable competitive advantage and don’t overpay for them. Technology is based on change, and change is really the enemy of the investor. Change is more rapid and unpredictable in technology than in the rest of the economy.”
– Warren Buffett

Nvidia shareholders shouldn’t feel too bad – the share price is still at pretty elevated levels:

I did come across an Nvidia investor comparing notes with a value investor yesterday:

My take: I can’t even pretend to understand the tech behind AI, but I do understand competitive economic environments. The profit pool generated by chip designers like Nvidia is just too large not to attract competition. Also, historically, the marginal pricing of everything China competes in devolves to the functional equivalent of 0.

I have no dog in this race; watching it play out from the sidelines is much less stressful.

One of the main reasons I am not changing my asset allocation based on forecasts about market crashes or booms is that projections are pretty useless. They are either too bland to contain any actionable information or just plain wrong.

Here’s a list of predictions made 100 years ago about 2025. One or two were quite prescient, but most were just plain wrong. People will live to 150? I wish. There will be nothing left to laugh at? Tell that to Dave Chappelle. We will run out of oil? They’ve been saying that for decades.

Here is a list of predictions for 2025. These are what the “experts” see coming. In my opinion, they are not particularly actionable.

We all love more certainty rather than less, so there will always be a market for predictions. In fact, the stock market is built around forecasting, and many investors pay top dollar for the expert’s views on the future.

Here’s Charlie Munger on the topic:

“People have always had this craving to know the future. The king used to hire the magician or the forecaster, and he’d look at sheep guts or something for an answer on how to handle the next war. There’s always been a market for people who purport to know the future based on their expertise, and a lot of that is still going on. People have an economic incentive to sell some nostrum. It can be sold over and over and over again… I think it’s disgusting. It’s much better to make a living by being part of a system that delivers value to the people who are buying the product.”

My take: Charlie didn’t become one of the wealthiest men in the world by accident. The first step to financial freedom is to stop listening to (and paying for) predictions and forecasts and instead play the cards as they are dealt. The MWI Worldwide Flexible fund is structured to survive and thrive regardless of the future, which is why it is nicknamed “the cockroach.”

For the past five years, there has been an ongoing war called “the streaming wars.” In this war, legacy media has been battling new media. As wars go, it’s been pretty ugly. Lots of money has been wasted on attempts to overcome the opposition, and many executives have gone MIA. Some companies have perished.

Here is a chart of subscribers of some of the participants in the war:

One way to look at Netflix’s dominance is to compare its market cap to that of Disney over the past two decades:

• January 2005: Disney ($46B) was ~15x larger than Netflix ($3B)
• January 2015: Disney ($175B) was ~5x larger than Netflix ($33B)
• January 2025: Netflix ($415B) is ~2x larger than Disney ($203B)

Bearing in mind that Disney also owns entertainment parks and other physical assets!

I’m happy to say the war is over, and Netflix has won. The market concurs. Here’s Netflix’s share price, which reached a new high last week:

I wrote about Netflix in Vol 2 No. 4. In it, I suggested that Netflix would be one of the winners, but I didn’t own it due to the war. In wars, you always have unpredictable collateral damage.

My take: If the war is over, capital discipline should return to the sector, making it investable again. For the equity portion of the MWI Worldwide Flexible Fund, in which I am building the “forever portfolio” of stocks, I’ve chosen Disney because it is more “Lindy”, and its business model has a self-reinforcing flywheel effect, not solely reliant on streaming.

The Daily Upside recently published an article titled “The State of the Billionaire Space Race.” It describes the fascinating race between Musk (SpaceX) and Bezos (Blue Origin) while ignoring developments in China and India, which are also engaged in space programs.

It’s fun to speculate about why this “race” exists. SpaceX says its central goal is to establish a self-sustaining city on Mars, which Musk envisions as crucial for humanity’s future. Blue Origin’s goal is to make space accessible for all.

My take: These tech giants are taking “prepping” to new levels. On the positive side, there will be positive technological spin-offs, just as the Apollo program had massive benefits for technological development in the USA.

Here’s Monty Python’s take on people who forecast the end of the world. It manages to poke fun at experts, forecasts, religion and speech-impaired people, all in the space of 5 minutes. It was made 50 years ago in less “politically correct” times.

My take: End-of-the-world forecasts are ten a penny. I would ignore them; whatever is in your portfolio will not matter if it happens.

I’ve ordered a CGM, or “Continuous Glucose Monitor.” This simple medical device tracks blood glucose levels continuously, providing insights into how various factors – food, exercise, stress, and medication – affect blood sugar levels.

Stable blood sugar levels – as measured by glucose – are vital for several factors:

• Reduces the risk of chronic diseases
• Increases energy levels
• Improved mood and mental well-being
• Better cognitive function
• Improved weight management

Well, that’s the theory.

Here’s a short video by the “Glucose Goddess,” Jessie Inchauspe. In it, she discusses the advantages of maintaining steady glucose levels and offers some helpful hacks.

My take: I’m on a mission to extend my lifespan and my healthspan – it’s not worth living a long life in which you are sick and weak in the last 20 or 30 years. As Dr. Peter Attia says, you need to be able to live an active and productive life in your marginal (last) decade. I have an idea that monitoring glucose levels might help me with this. I will report back.

Last year, I completed 1983, 1984, 1985, 1986, and 1987 on my musical journey to document the best music from my birth year. I recently finished compiling the best music of 1988.

In 1988, I was in my second year of conscription. I finished an officer’s course at Personnel Services School in Pretoria, avoiding combat. My first job was as a platoon commander for the first basic training intake of 1988. Being the Personnel Services school, it contained mainly people like me who had first studied to avoid conscription for as long as possible – also, lots of  “G4K4s” or conscripts certified as unfit for heavy duty. Of course, being the smart guys they were and having the advantage of being well-connected in medical circles, there were many Jewish guys in this cohort.

I had a lot of fun in those first 6 months.

For the second six months, I wangled my way into an even more cushy job – the second in command of a “company”, which consisted of 10 or so platoons. A “company” is headed by a Permanent Force captain, with me being his skivvy. With the explicit aim of doing as little work as humanly possible.

My grand plan hit a snag the day before the second intake when the PF captain was involved in a serious accident. I was made company commander, which resulted in me working harder than I had ever worked before.

Managing 300 conscripts – none of whom wanted to be there, including myself – was challenging. However, I learned a lot about people and their incentives, which has benefited me throughout life. Eventually, I also married a Jew – a practical illustration of how much I learned from my troops, for which I will be forever grateful.

So, for me, the music of 1988 is the music of a year in which I expected to kick my heels back but ended up doing much more. I also ran my first Two Oceans marathon, the highlight of which was sharing a cigarette with a spectator at the top of Constantia Neck. Stupidly, I only quit 20 years later! (This is, of course, what we in Afrikaans call “fynbrag”. Well done on spotting it!)

Here’s a playlist of the best albums

On Apple Music

And on Spotify

Here’s a playlist of the top 20 songs

On Apple Music

On Spotify

And here is a long list of the year’s best songs, only on Apple Music.

I hope you find something in there that brings back good memories!

And remember, even if it isn’t the end of the world, you still need to be careful out there.

Piet Viljoen
RECM