January 28, 2025
There are a lot of moving variables in the investment landscape lately. We’re in a new political environment, and technology continues to blaze forward as fast as ever.
So to start the year, this newsletter issue focuses on some “North Stars”. Specifically, it covers three trends that I will make sure not to fade over the next decade from an investment perspective.
Thing 1: Energy Density
The first thing I will not fade is the importance of energy density, and the continued use of high-density sources of power.
Energy usage doesn’t just go toward electricity generation and car fuel. It goes toward excavating raw materials, manufacturing goods, transporting those goods, creating nitrogen-based fertilizers to feed people (natural gas is a major input for that), powering machines to plant and harvest crops, maintaining safe levels of heat or cold for people and businesses, purifying and pumping water to make uninhabitable places habitable, and so forth.
This high level of energy usage is what allows 8 billion people to survive on this planet, and with an increasing percentage of them in relative comfort. Humans and their livestock represent more than a 10x larger amount of biomass than all other wild mammals and birds combined, basically creating a top-heavy food chain, enabled by this energy expenditure.
It’s no accident that human lifetimes and human populations both exploded in the late 1800s and early 1900s when oil discovery and usage ramped up. And it directly contributed to the rapid growth of technology, because it’s what allowed the vast majority of people to stop working in agriculture and start working in other fields that progressed our technology, medicine, industry, and so forth.
When countries adopt cultural views or government policies against energy density for the sake of optimizing some other variable, they’ll generally diminish their share of the world’s economic output, either knowingly or unknowingly. They can make that decision, and the rest of the world will move on without them, until they choose something different.
Energy efficiency also matters of course, but that has certain limits. More energy creates a bigger foundation to grow from, and then better technology makes more efficient usage of that energy. Moving around more mass than is needed for a given purpose, and dissipating more energy in the form of ambient heat than is needed, are two common sources of energy inefficiency that better technology gradually improves over time.
That’s not to say that there’s no role for wind and solar. They are variable sources of power, and with solar power the benefit is that it can provide power in a decentralized way without being attached to the grid. But turning variable sources of power into baseload power requires massive battery arrays, which require huge amounts of battery metals. If we 10x or 20x our grid-scale battery installation to get more of our baseload power from wind and solar, we would then have to renew large portions of that entire stock of installed batteries decade after decade.
Battery technology has historically been a very hard technology to improve upon. Improving energy efficiency and energy density in batteries has been a slow process going back to the 1800s. Without stepwise improvements in battery technology (rather than just incremental ones), it will be a very long time until the majority of humanity’s power usage can come from variable sources of power.
In addition, many types of heavy industry including cement, steel, glass, and chemical manufacturing require high heat in their production process. Dense sources of energy usually produce heat which can be converted to electricity or used directly, whereas solar and wind power produce electricity which then has to be converted back into heat if it’s to be used for those industries (which is a lossy process).
As a result of all of this, I am bullish on dense sources of energy, and on companies and economies that more readily make use of those dense sources of energy as producers or consumers of it. This includes all forms of hydrocarbons, as well as nuclear power, hydro power, and geothermal power.
Thing 2: Computation
The second thing I will not fade is the quantity and quality of computation that will be coming, and the importance of being on the right side of that computation.
During much of the past century and a half, as described above, humanity’s energy usage has primarily gone to manipulating the physical environment to make it more habitable and comfortable for human life and in greater numbers.
Some of that energy has gone to computation along the way, but only a very tiny share. That’s because our best computers had computational abilities that were a negligible fraction of what the human brain can do. A basic calculator doesn’t require almost any meaningful energy consumption, and a bunch of human engineers with calculators were basically the best computation machines that we’ve had since the dawn of time. The human brain runs on about 20 watts of power.
What’s different about the 2020s it that our silicon processors are finally hitting the scale of computation comparable to what the human brain is estimated to be capable of. Forecasters have been predicting this for a while, and indeed we’re getting to somewhere around that level here in the 2020s. Here are some charts from a decade or more ago, and indeed we’re still on those trendlines:
Of course, software matters as well. Just because a computer has processing power equivalent to a mouse or a human doesn’t mean it’ll be well-programmed enough to operate as efficiently and self-sustainably as a mouse or human. The “software” that we run on has been through countless lifetimes of iteration and improvement.
And so because of that, I don’t try to make predictions regarding which year some capability will be achieved by computers. That’s very hard to do even for industry professionals.
But we’ve already seen the usefulness and importance of generative AI over the past few years. Within that rather limited period of time, large language models and other types of generative AI have impacted the way that we search for information, the way we communicate, and the way we create various forms of art. Add another 10x or 100x processing improvement and iterations of better software to this over time, and it’s not hard to see how impactful it can be.
At the business level, it can enable many types of white collar workers to each do the work of multiple workers. For example instead of programmers writing each line of code, they can increasingly oversee program creation via AI-generated code. Instead of searching manually through legal codes to win a court case, they can have AI do a lot of it for them. The better that AI gets, the more it can automate complex things for us, either entirely on its own in some cases or by extending the scale of each human worker that wields the technology.
The rise of big tech companies in the 2010s was mainly about enhancing communication and making that communication more mobile, which was not very computation intensive.
High-bandwidth information used to be broadcast in a one-to-many pattern. Big sources like book publishers and television stations would broadcast information to the masses. But now with the rise of this technology over the past couple decades, the masses broadcast high-bandwidth information to each other. This has massive impacts on society and the shape of the economy, but all of those queries and messages are relatively computationally cheap.
And by turning the personal computer into a mobile computer (smart phone) that almost everyone has, with better processing capability per unit of energy and size, it reduced our need for other things. Fewer of us need big clunky landlines, fax machines, or printers. We consume less stationary, use fewer physical calendars and envelopes and pens, and all that kind of stuff. Our little energy-efficient smartphones replace the need for dozens of other items, which saves on energy in various ways including by just cutting down on sheer manufacturing volume.
But the next wave of technology in the 2020s and 2030s is likely to be different. As our silicon computation capabilities reach and surpass the sheer processing power of the human brain, we can offload more and more white collar tasks to machines, and expand what our processors can do for us. A nontrivial amount of our energy production can go toward computation, and computation becomes an increasingly essential resource for all economies.
Much like how energy-intensive tractors increased the output of each human farmer, and energy-intensive manufacturing robots increased the output of each manufacturing worker, energy-intensive and computationally-intensive AI can increase the output of each white collar worker (accountants, programmers, designers, engineers, scientists, lawyers, doctors, analysts, and so forth) going forward.
Thing 3: Network Effects
The third thing I will not fade is the power of network effects. Winners with self-reinforcing network effects keep winning until something very disruptive stops them.
A positive network effect occurs when each additional user to a system makes the system better for all other users as well, which therefore attracts more users to the system.
An early example was the telephone. If you own the only telephone in the world, it’s useless. For every additional telephone owner, it increases the number of people that all telephone owners can call. During the age of telephone adoption, your telephone would get better every time some person or some company adopted the telephone, which attracted more people and companies to adopt telephones.
Communication protocols also benefit from network effects. Ethernet, Simple Mail Transfer Protocol, Universal Serial Bus, and other communication protocols tend to become entrenched for very long time once they become dominant. Spoken languages are also communication protocols. The dominance that English has as the global language of business is rather entrenched and self-sustaining. And for all types of spoken or written content creators, it’s very powerful for you if English is one of your fluent languages.
Social media networks are similar. The more users a given platform has, the more powerful that network is for each user.
Anything that benefits from financial liquidity is also basically a system with a positive network effect. This includes things ranging from the U.S. dollar system to the Bitcoin network. It’s why the euro and yuan have a tough time competing with the dollar, and why the 20th cryptocurrency by size is irrelevant. If you’re not first for these types of things, you might as well be last.
Network effects are not outright invincible. They’re just extremely strong. My friend Jeff Booth refers to the 10x framework: a challenger has to be 10x better than the entrenched leader to win. This applies to things ranging from startup companies to communication protocols. As he wrote in “The Greatest Game“:
When creating new technology companies, a framework I use to understand if something has an ability to win is ‘the 10x advantage’. Meaning that, unless a new company (the challenger) can deliver a 10x advantage to the market, it has no chance of creating escape velocity and becoming a new category leader. While a 10x advantage does not guarantee the success of a challenger, it greatly enhances its probability
-If someone figures out how to make a competitor to Universal Serial Bus that is 20% faster, cheaper, and more durable, it doesn’t matter. It can’t compete with the fact that billions of devices have USB ports. It starts with such a massive disadvantage such that no linear improvement can compete, and USB does update over time so it can easily close small gaps that open up.
-If someone figures out how to make something far better than USB, then it actually has a chance to catch on. It still has a big uphill battle, but in this case the challenger is so incredibly disruptive and better that it actually has a shot to start getting included by various device makers and begin overtaking USB.
Therefore, for any given network effect, I will assume continuation unless or until I see clear massive disruptors to them.
Putting This Together
Whenever I make an investment, I will use the trends of energy density, computation, and network effects as some of my north stars. I’ve already been doing that, but this piece hopefully helps clarify that further, and serve as a reminder to myself to keep leaning into that.
In practice, this means that when evaluating an investment, I have to ask myself:
-Is this competing with denser sources of energy?
-Is this on the wrong side of the ongoing rapid growth in computation?
-Is this competing with network effects larger and more powerful than itself?
If a potential investment checks “yes” for any one of those three questions, then I’d have to be very critical of investing in it. It’s not to say that no potential investment can touch on any of those, but the hurdle rate is much higher for those that do and they should be considered more speculative. It’s much easier and safer to have those three trends as tailwinds than headwinds.
Portfolio Updates
I have several investment accounts, and I provide updates on my asset allocation and investment selections for some of the portfolios in each newsletter issue every six weeks.
These portfolios include the model portfolio account specifically for this newsletter and my relatively passive indexed retirement account. Members of my premium research service also have access to three additional model portfolios and my other holdings, with more frequent updates.
M1 Finance Newsletter Portfolio
I started this account in September 2018 with $10k of new capital, and I dollar-cost average in over time by about $8k per year.
It’s one of my smallest accounts, but the goal is for the portfolio to be accessible and to show newsletter readers my best representation of where I think value is in the market. It’s a low-turnover multi-asset globally diversified portfolio that focuses on liquid investments and is scalable to virtually any size.
And here’s the breakdown of the holdings in those slices:
Changes since the previous issue:
- Sold BMY.
Returns Update
The portfolio has generated more than $46k in gains over principal. If I had instead dollar-cost averaged into a similar globally-diversified portfolio of 80% stocks and 20% bonds with the same amounts on the same dates, using the iShares Core Aggression Allocation ETF (AOA) as a benchmark, it would only have generated a bit over $25k in gains over principal.
Most of the excess returns came from overall macro asset allocation decisions, such as replacing part of the bond side with precious metals, keeping the remaining bond duration short, and replacing a small slice of the equity section with bitcoin exposure.
In other words, the portfolio was more diversified than the average type, and the extra diversification mainly leaned toward hard assets since we’re in an era of fiscal dominance.
Other Model Portfolios and Accounts
I have three other real-money model portfolios that I share within my premium research service, including:
- Fortress Income Portfolio
- ETF-Only Portfolio
- No Limits Portfolio
Plus, I have personal accounts at Fidelity and Schwab, and I share those within the service as well.
Final Thoughts: Trends Can Get Ahead of Themselves
Although I will not fade any of these three trends structurally, it doesn’t mean that a trend can’t become overdone for a couple years. Most of them have some degree of cyclicality.
Bitcoin, for example, has had 16 years of structural growth but rather clear and measurable boom-bust cycles along the way when the market capitalization spikes way above the on-chain total cost basis (a.k.a. realized capitalization):
In fact, bitcoin is the only asset I’m aware of that has had four separate instances of drawing down over 75% and bouncing back to new all-time-highs each time.
The semiconductor/hardware industry broadly is infamously cyclical as well, despite being so important within the modern economy. Several companies like AMD, Apple, and Micron have all seen three drawdowns of over 75% followed by new all-time highs each time.
The recent popularity of DeepSeek, a Chinese artificial intelligence company, potentially showcases an example of that in action. With China under various chip sanctions, DeepSeek developed AI models that rival some of the AI leaders at a fraction of the cost and processing power, and then open-sourced much of what they did.
Many AI companies are not yet profitable, despite spending a ton of money on capital expenditures. It’s possible that they have overspent, and their equity valuations are high. Even strong trends go through periods of contraction and expansion, and this trend is probably no different even if it is bigger than prior ones.
But even when a given trend goes through a contraction, we shouldn’t lose sight of what it can disrupt. Just because a given trend is overpriced or overbought from an investment perspective, doesn’t mean that it won’t continue to suppress the margins of other industries or otherwise disrupt and displace them.
From the second law of thermodynamics, we know that entropy (i.e. disorder) of an isolated system always increases over time. Life in many ways represents the little pockets of pushback to that idea; it finds a way to organize things toward order in their little areas of the isolated system, at the cost of increasing disorder elsewhere. And humans are a particularly massive form of life, due to the scale in which we can harness other sources of energy to create massive order in some locations, at the cost of creating massive disorder elsewhere.
Computation is one of the highest forms of order. Humans started with fire and spears, and after thousands of years we’ve collectively iterated toward supercomputers. Putting together countless atoms to make a mathematical machine, and even to some extent a reasoning machine, is an incredible achievement. As the computational density of this process surpasses even our own processing ability, that’s a first for our planet, and is the essence of what science fiction is made from.
Best regards,
