Click here for a PDF copy of the Introduction and Act I, Pillar I on the New Order.
Click here for a PDF copy of Act I, Pillar II on Demographics.
Click here for a PDF copy of Act I, Pillar III on The Automation Imperative.
Click here for a PDF copy of Act I, Pillar IV on Weaponization.
What follows is a series of essays I am writing where I lay out my thinking on what has happened in the global economy in the last quarter century, what forces are/were responsible for this change, what the future may look like, and the implications for investment in critical materials. The model that defined much of the past 25 years, specifically low cost of capital globalization, optimized supply chains, and globally integrated trade is moving towards a model shaped by global fragmentation, the combining of econoimic and national security, economic chokepoints, and competition for strategic materials.
This shift has implications for capital allocation and investment returns, specificallt for critical materials.
The story is broken down into three Acts much like a Shakespearean play. This is an apt comparison given the triumph and tragedy we have seen and are likely to continue to experience as a host of forces collide to both create and destroy opportunities across borders.
Act I discusses the global macroeconomic forces that led to change and makes the case that these are in many ways irreversible; Act II discusses constraints to growth in this new order, and Act III discusses the implications of these changes for investors.
The economic assumptions many of us grew up on can no longer be taken for granted. Critical raw materials still form the backbone of economic growth, but these assets and projects are set to take on an elevated level of importance as governments prioritize security and resilience. The chase for the lowest cost of production is over. The “game” for investors will be figuring out which projects can be financed and operated profitably given these emerging economic and geopolitical constraints.
Each of the essays serves as a pillar which supports this new economic operating environment. This document is also a work in progress, so comments are welcome.
ACT I – FORCES – When and Why the World Changed
Pillar 1: The End of the Old Order
It’s difficult to point to a single event that has altered the course of history and brought us to where we are today – increasingly fragmented trade, aging populations, relentless technological advances, and a political class that is increasingly disconnected from the “man on the street”.
I was born in 1974 and for the almost entirety of my life, the United States has been the trusted global hegemon. Hallmarks of a hegemon include a reserve currency, deep and liquid capital markets, the rule of law, provider of a security umbrella, and patroller of the high seas. This is all designed to help global trade grow unabated.
That said, something has changed. The trust that the US had built with allies over decades has eroded. Investors, stakeholders, and even many policy makers are (quietly) questioning who is in charge and who sets the rules for the future. Can the US be trusted to provide the benefits to allies mentioned above? The rhetoric coming out of Washington DC would lead you to believe that the global operating environment is in many ways “every man for himself”. This is best demonstrated by the headlines about increased defense spending in parts of the world like the EU. Defense spending in 2025 will total EUR 418B, a 20% increase over 2024.
Multiple events have effectively shocked the old order, forcing all of us to find a new equilibrium. Seminal events including the 9/11 attacks, the 2008 Global Financial Crisis (GFC), BREXIT, the COVID pandemic and the response to it, the Trump presidencies, and the inexorable rise of China are the most obvious examples. This is not an exhaustive list, but after each event, it’s fair to say that the world looked much different after than before. Each event put a crack in the façade of a world order whose time was running out.
This old order was designed to provide economic growth and a security blanket based on the idea of free trade and a single country (the US) providing that security and underpinning the growth through reserve currency status, deep financial markets, the rule of law, economic opportunity onshore, and a power military. The result was a set of mutually self-reinforcing conditions: the US exported security and economic opportunity, consumers benefitted from lower costs and access to a wider array of goods and services. China, with a giant workforce and low labor cost, was the other side of this coin, exporting deflation.
Now, many of those forces have halted or are in reverse. The Trump Administration is (rightly) pushing other countries to spend more of their own capital on their own defense, tariffs and export restrictions are increasing the production cost of goods, the reindustrialization push in the US championed by both the Biden and Trump Administrations is highly capital intensive, countries are actively looking at US Dollar alternatives such as bitcoin, and China’s low cost growth model is on the ropes. These changes all point to a higher cost, more inflationary world going forward. There is an optimistic case centered around the deflationary push of technology such as AI and this will be discussed in a subsequent piece.
Here’s the main takeaway: A structurally higher cost of capital is colliding headlong into a higher demand for the physical raw materials (lithium, copper, rare earths) needed for the next phase of global economic growth.
This old order rested on a handful of assumptions:
· Low interest rates were a permanent feature of business. Globalization allowed for labor arbitrage and set the stage for just-in-time inventory management which allowed companies to optimize their financial performance. China’s admission into the WTO in 2001 was the matchstick, when combined with the fuel of economic activity that arguably caused the forest fire we are experiencing today: the global economy grew, but at the expense of a shrinking manufacturing base in the West. As a result, China utterly dominates many global supply chains, especially those for critical minerals. Many thought that lower interest rates would continue into perpetuity thanks to the “virtues” of capitalism and an enormous labor pool in Asia helping to keep costs down, and goods affordable to the masses. Here’s a chart of the US Treasury Ten-Year bond:
While this global benchmark is still low on a historical basis, the trend towards higher rates is undeniable.
· The US would guarantee global security. The US military, dominant since the end of World War II and especially since the end of the Cold War, was the real force that ensured seas were patrolled, security blankets were sufficient, and economic growth globally would be the beneficiary. Democracy was also hoped to be a byproduct here, but has a much spottier track record. With the Trump Administration bullying allies into spending more on their own domestic defense, the US absence on the seas, in Europe, and elsewhere removes one of the strongest pillars of global economic growth in the last 80 years. Who (or what) fills that void remains to be seen.
· Wide access to the US dollar. The US dollar has enjoyed the “exorbitant privilege” as the world’s reserve currency for decades and arguably still retains that designation despite competition from the Chinese Yuan, bitcoin, and other currencies. How much longer this privilege stays in place has been written about extensively and I won’t belabor it here. Countries wanted to and continue to hold US Treasury debt thanks to its perception as risk free. While the US has never defaulted on its debt (and I don’t think ever would), the largest economy in the world shouldering a $40 trillion debt burden where interest on the debt is now a larger budget line item than defense spending continues to give finance ministers indigestion. Everyone is looking for a US dollar alternative when none really exists. Additionally, the US dollar is now a tool of geopolitical leverage, through means such as sanctions and access to clearing networks like SWIFT. Iran, Russia, and other countries that run afoul of how the US thinks the world should run have found this out firsthand.
· Just-In-Time inventory management took precedence over Just-In-Case, or supply chain efficiency took precedence over supply chain resilience. Capital allocation under a globalized trade network was designed to minimize costs, maximize profit, and grow at all costs. Perhaps no economic shock better exposed the fallacy of this thinking than COVID when the West woke up to the fact that one country, China, has spent the previous quarter century building supply chains to dominate multiple sectors of the global economy with plenty of excess capacity to boot. Supply chain restructuring is underway in the West, but it will be years before any significant fruit is borne.
It’s worth clearly stating that this Old Order was very, very good for financial assets such as stocks and those lucky enough to hold them and not so good for physical assets or labor. Low wages wasn’t an accident, but instead was the globalized economy working as designed. It was this design that set the stage for widening income inequality and the K-Shaped economy phenomenon.
The Three Breaks
The old order didn’t end at once but was exposed by at least three events, or breaks, that have brought us to where we are today.
The first break was really one of legitimacy – the GFC in 2008. While the mechanics of the global economic system did not end in 2008, the blind belief in them did. Phenomena like Occupy Wall Street and the fact that no Wall Street executives were held accountable for bringing the global economy to the verge of a depression were cases in point. Taxpayer-funded bailouts of banks largely responsible for this global economic calamity were the salt in the wound and set the stage for today’s K-shaped economy. The medicine for this heart attack – quantitative easing (QE) was terrific for financial asset prices while globalization kept a lid on wages. Income inequality was the result.
The second break was the COVID pandemic. While the response to the GFC was focused on saving the banks, the fiscal response to COVID was to put money directly into the pockets of consumers and businesses in an effort to maintain spending in a US economy dominated by the consumer. This laid the groundwork for the end of the low interest rate era as the flood of money into the US (and global) economy inflated goods prices and wages. One of the results is sticky inflation in the US that has been above target for over five years.
The third break isn’t a single event but is the willingness to weaponize sacred cows. Specifically the weaponization of the US dollar. While not a new phenomenon, the willingness of those in Washington DC to literally kneecap other economies by restricting access to US dollars has forced a rethink in Tehran, Moscow, and Pyongyang, but also raises eyebrows in friendly capitals. If the US is willing to use the dollar as a weapon against a country that doesn’t toe the line what would stop it from making and following through on similar threats to allies? There is a reason that gold and silver touched all-time highs recently and it isn’t your typical supply and demand argument.
There are, to be sure, other events or breaks which deserve a mention including Brexit, the wars in Ukraine and Iran, and both Trump Presidential elections. I would argue that the Trump elections are results of the breaks discussed above. The fallout from the GFC gave voters the impression that bad behavior on the part of “the elite” on Wall Street could go unpunished. After asset prices rebounded sharply thanks to the Greenspan/Bernanke Put, this distributional failure only cemented the idea that capital owners were above the law, and Donald Trump was the result – twice.
Sticky Inflation and the Rate Problem
A big challenge for governments is balancing deficit reduction while undertaking massive spending plans to reindustrialize and reshore supply chains. Reindustrialization creates policy tension and needs loose monetary policy/low rates to succeed in that a lower cost of capital boosts investment returns. As mentioned, the COVID response was aimed at keeping the US consumer spending and the economy out of a recession. This was accomplished, but at a cost of higher domestic inflation as measured by the consumer price index. Industrial policy is inflationary (especially with the US economy at full employment). While the Department of War, for example, lends and grants hundreds of billions of dollars to kickstart critical material projects, the Federal Reserve must fight these inflationary activities by raising interest rates – raising the cost of capital for these capital-intensive projects. Do you see the problem? One arm of government is adding inflationary pressure by funding infrastructure, and another arm of government is trying to suppress inflation through monetary policy adjustments. Tighter money is self-defeating in that a long-lived mine or smelter is sensitive to interest rate movements in the way a software platform isn’t. The reindustrialization buildout in the US will only happen with these mechanisms (funding, price floors, etc) which, by their nature, are inflationary. The resulting policy mismatch is the result – governments wanting productive capacity for national security reasons, while the economics of a project (a lithium converter, for example) may not generate the required return needed to initiate the buildout. A higher cost of capital is the natural consequence of the need to reshore critical mineral supply chains. Oddly, an unfixable $40 trillion deficit doesn’t seem to be the biggest issue here. The issue is the true cost of capital in this new order. Increased demand for critical materials from defense and industry, higher input costs due to choked supply chains and wars, and resource nationalism all appear to substantiate my claim.
Consequences for Asset Owners
Obviously, these forces all imply significant investment implications. While the cyclical nature of critical materials isn’t going to change, how one invests must be re-evaluated given what I’ve laid out. Higher cost of capital is not as impactful for a brownfield mine or smelter but has significant implications for a greenfield project that is dependent on sustained robust pricing. Historically, capital was raised and allocated based on a project’s potential ability to generate cash and produce goods at a globally competitive cost. In the new order, security of supply, partnerships, and government support are now the sine qua non of Western critical material project selection.
This is why we have seen so much emphasis from Western governments on funding projects – the “China price” has been the number that drives returns. The country dominates so much of the supply chain from mine to market that a higher cost of capital only impedes capital allocation here implying low (or negative!) returns. If supply capacity doesn’t get funded and built in the current environment, no additional supply materializes in the future, and China still dominates. Subsidies are an unfortunate necessity.
With apologies to Adam Smith, the Invisible Hand is irrelevant in the new order and the sooner capital allocators accept this, the better. Subsidies such as price floors, off take agreements and grants and loans determine who builds projects and where they are built. Government support doesn’t necessarily fill a structural supply deficit but does allow Western producers to earn a return above the higher cost of capital. This essay and the ones that follow aren’t designed to pick winners – either via commodity or company – as they are all subject to these same challenges in the new order. Earning an “acceptable” return requires many of the same criteria as before (top tier management, high-quality asset, structural supply deficit, etc) but must accept government as a partner.
The Bear Case
With critical materials desperately needed in the West and governments fully behind this reshoring push, one would think that there is a “critical materials put” in place in the same way many investors came to believe that the “Greenspan Put” was in place to protect equity markets from falling too far, too fast. Any investment thesis deserves an analysis of its inverse and critical materials is no exception.
One of the core arguments is that interest rates stay structurally higher, which raises the cost of capital and lowers returns on greenfield projects. While this appears to be true based on five years of CPI data, interest rates in the 1970s were much higher than today and preceded a 40-year bond bull market, so regime changes aren’t necessarily a given. Additionally, the impact of AI (to be discussed in a subsequent essay) could eventually deliver the promised productivity gains to the global economy, keeping a lid on rates and raising wages without stoking inflation.
Second, demographics (covered in the next essay) would seem to favor secular stagnation and lower rates in the future as major economies continue to slow their population growth though, as I’ll demonstrate, this is a debatable premise.
Third, China continues to manipulate and in some cases flood critical metals markets, depressing prices and contributing to oversupply, decimating Western mine and refinery economics.
Fourth, the bond vigilantes appear in the form of much higher government bond yields, forcing governments to prioritize budget deficit management over fiscal expansion to the detriment of the reindustrialization thesis.
Fifth, trying to fund and manage projects with twenty year lives against a four-year election cycle is difficult regardless of the macro backdrop. The possibility that a post-Trump presidential administration changes or undoes some of the policies in place does exist, however unlikely. Combatting Chinese dominance in critical metal supply chains seems to be one of the few bipartisan issues in Washington DC these days.
The next essays in Act I will dive deeper into why the world has changed. The next essay will look at the effects of demographics.
Act 1, Pillar 2: Demographics as a Demand Driver
The grand bargain that underpinned economic growth in the West is breaking. The promise was that if you worked hard, you could own a home, raise a family, lead a comfortable quality of life and retire and leave something to your heirs. This model rested, in part, on favorable demographics: a society continuing to grow both organically (more babies being born), an expanding labor force, and immigration.
We are now learning what happens when the demographic tailwind disappears and the grand bargain reverses.
Demographics in most of the West (and China and Japan) has gone in reverse. The consequences are wide as demographics directly influences economic growth rates, labor availability, inflation, and even demand for critical materials.
A central tension is the result. Dependency ratios are rising, implying that the demographic dividend which underpins economic growth is in danger. An aging population can lower economic growth and consumption which can lead to lower prices and interest rates (classic deflation). On the other hand, a smaller labor pool means fewer workers which enhances worker bargaining power (classic inflation). To complicate matters slightly, more worker leverage can accelerate investment into automation, possibly reducing the need for human labor. An aging population doesn’t necessarily mean fewer workers in absolute terms but does introduce fiscal pressures (funding Medicare and Social Security) which have no easy or apparent remedies. Again, this is not a phenomenon unique to one country.
For investors, the second point is important. If fewer workers are needed to support the economy, productivity gains are essential. While automation is one solution, this requires copious amounts of capital, new infrastructure, and raw materials – typically critical raw materials. Critical materials underpin this automation and so access to them is not optional.
The Rate of Change
To start, what is important to remember is that the age level isn’t the most critical piece of data, but instead the rate of change of age. In other words, how quickly or slowly is a society aging? While the charts linked here and in the Appendix show relatively uniform aging across many societies, it is China that is notable. With the working age population shrinking, this undercuts one of the country’s competitive advantages, namely it’s deep well of labor. So as China’s labor pool shrinks, the single most powerful reason for low-cost goods in the West (offshoring and labor arbitrage) looks to be ending. While moving manufacturing operations to other low-cost destinations such as Vietnam is a possible solution, no country can match China’s scale.
This introduces a fascinating debate around what happens on a macro scale when powerful inflationary forces (such as decreasing labor availability) crash headlong into powerful deflationary forces (such as automation). Overall, the deflationary case vis-à-vis demographics is currently the most widely believed. With an older population altering consumption patterns, this can keep a lid on inflation as lower consumption leads to lower demand, higher savings, and lower interest rates. Japanese economic performance in recent decades is an example.
The inflationary case is more interesting as it’s not as intuitive. The best primer on why inflation rather than deflation is the direction of travel was written by Charles Goodhart and Manoj Pradhan, titled ‘The Great Demographic Reversal’. The central claim of the book is that the forces that led to decades of disinflation (China’s entry into the global economy and Baby Boomers in the workforce producing a labor glut) could have only happened once and is now in reverse as China ages and 10,000 Baby Boomers reach retirement age each day. It was this labor glut that kept inflation at bay and delivered the “goldilocks economy”.
The elderly then become consumers rather than producers and labor becomes scarce which enhances bargaining power, pushing up wages and fomenting inflation in the economy. Healthcare begins to account for a larger portion of the economy, is labor-intensive, and resistant to productivity gains. While I have always been in the deflation camp, I find this line of argument around inflation compelling and difficult to refute.
There is a deflation argument here with respect to aging societies and it rests with automation and technology. Those goods that can be manufactured by leveraging robotics hold the promise to counteract the possible inflationary forces unleashed by healthcare costs in an aging society. Which force “wins” is fodder for a robust long-term debate.
This is not a new phenomenon and historically when labor is scarce or expensive, capital comes to the rescue to build automation. Perhaps a positive takeaway for resource investors here is that automation is resource intensive. Enormous amounts of copper, lithium, rare earth magnets, and other materials are needed to bridge the gap between scarce labor and automation. If labor remains scarce, capital must come to the rescue to automate activities that go unfilled.
So, the “flow” looks something like this:
Aging society –> labor scarcity –> automation –> materials and energy demand
Viewed through this lens, demographic changes look to be an accelerant to critical materials demand. It is important to remember that the economic forces behind inflation and the economic forces that determine demand for critical materials are not the same. While automation could be deflationary as it increases output at lower cost, significant investment is needed to build the infrastructure supporting automation which would be inflationary in the near-term. Investment demand, productivity, and substitution all impact this debate in different ways.
A Brief Note on Politics
Immigration is a force that can help defray the impact of an aging society, but its impact depends on how much immigration occurs and how quickly. Immigration in the US has been the rocket fuel for economic growth in the country for decades and so the current political climate around this issue is disheartening. Talk about cutting off your nose to spite your face. Without adequate immigration, labor availability remains in question, and the inflationary winds continue to blow.
The Bear Case
There are some reasons to be wary of the inflationary consequences of deteriorating demographics.
First, should automation and AI permeate through society faster than the labor pool shrinks, the aging threat could be deflationary after all as the much hoped-for productivity miracle arrives as promised. Stable economic growth underpinned by healthy wage growth and widespread access to transformational technology is a policymaker’s dream scenario.
Second, there isn’t a good case study of an aging society producing inflation that I could find. Japan has, by most accounts, failed to ignite any inflation in the domestic economy despite decades of trying.
Third, the politics of immigration could reverse course and increased flow of labor across borders keeps a lid on any inflationary impulses. How likely any (or all of these) are is up for a rigorous debate, but in the interest of a balanced look at outcomes, they each need to be considered and handicapped.
Fourth, there is a possibility that inflation and deflation co-exist. As mentioned earlier, automation can be deflationary in an economy over the long-term through increased productivity while the buildout of the infrastructure needed to support the automation is inflationary due to substantial investment in raw materials, etc.
Given that these essays all involve AI in some way, harnessing the oncoming freight train of this technology could be viewed as a necessity rather than a “nice to have”. If populations shrink and grow old, productivity growth is an important solution to growing an economy with less available labor. Thus, building AI infrastructure and harnessing the technology becomes vital to maintaining a viable standard of living. The thesis here on demography is compelling, but the true implications for critical materials demand will depend on the rate of change of aging as well as how quickly automation technologies are adopted.
The next essay in this series will focus on the automation imperative.
Act I, Pillar III – When Asset-Lite Gets Heavy
The continued rollout of AI into our lives is likely to introduce a new technological paradigm. The long-term benefits are still to-be-determined and despite some of the legitimate concerns about the technology and the economic benefits, AI continues to improve. One thing is certain: the AI buildout will require an enormous amount of resources.
AI is a story about access to power and compute. The constraints we hear about are not technology-based, they are physical. Building the necessary power and compute infrastructure is both capital and resource intensive. Electricity, grid infrastructure, semiconductors, and the raw materials all get you to the starting line. No copper, no power. No lithium, no grid-scale storage. No scandium, no solid oxide fuel cells, and so on. Multi-year waits for gas turbines and backlogs for semiconductors are but two examples of near-term constraints on the growth rate of AI.
In tech cycles of the past, but of this new economy was built around low-marginal cost software. The AI buildout is almost the opposite. As mentioned above, access to the physical tools gets you to the starting line.
This raises a question for investors: which companies are best positioned to capture the growth, and which materials and infrastructure are requirements rather than “nice to haves”? Each question can run amok and turn into its own bubble – a significant concern in the current AI buildout where hundreds of billions of dollars are planned to be spent in the coming years. Complicating these questions slightly is also the fact that AI profitability and returns on AI infrastructure are different.
The basic premise of this Pillar is that the AI buildout and subsequent proliferation throughout the economy could backstop significant demand for resources. Whether or not that demand results in solid investment returns depends on a host of factors including capital intensity, cost of capital, and structural supply constraints.
The US-based large language model developers (LLMs) such as OpenAI, Google, or Anthropic aren’t the only companies racing to build AI infrastructure at scale and the rise of China-based open-source models has led some to view AI as an arms race. It isn’t likely that one of these companies can outspend the other to win in the same way as the US outspent the USSR during the Cold War in a different type of arms race.
Overspending on power or compute infrastructure could lead to ruin as spending gets too far ahead of sustainable revenue. Though some of the LLMs are not public, they are believed to be highly unprofitable during this stage of their growth.
So, there are likely two trades here for investors: a tech trade to bet on which LLM will emerge (relatively) unscathed from the inevitable boom and bust cycle and a resource trade to bet on which raw materials and infrastructure will be most needed as the AI buildout continues.
Historically, the tech trade and resource trade were opposites – tech was a great deflationary impulse and a reason why hard assets underperformed. Hard assets typically thrive in an inflationary environment, and the proliferation of tech put this trade to bed more than once. This relationship has now inverted, and it is resource demand taking precedence over tech. The previous tech investment cycles were capital-light. The AI buildout is anything but with it running on copper, steel, natural gas, and a host of other materials. Access to power and raw materials increasingly determines the pace of growth.
Access to compute is a story about power which is physical. Nvidia’s market capitalization of $5.5 trillion is the market betting that this company will continue to grow through the cycle based on its technology which is in part underpinned by a resilient supply chain. Resilience will also be discussed in Act II as a constraint.
From the 1990s onward, the appeal of tech investing has rested on the fact that software was zero marginal cost and Moore’s Law would continue (though not indefinitely). This investment model was capital-light, low-commodity intensity, and high return on invested capital (ROIC). The era of digitization and investment in bits took precedence over investment in atoms.
What Happened?
As AI has expanded its reach, computing economics has changed. Investment in cutting edge technologies depends upon a number of issues including speed and scale of deployment once the tech is developed. As Moore’s Law continued to slow down, this necessitated either new technologies or rethinking tech architecture to “squeeze more juice out of the lemon”. This can increase the bill of materials as a change in workflow or supply chain dynamics can cost both time and money.
A data center is more than four walls and a roof. It is an enormously complex ecosystem of power, cooling, and energy distribution. Semiconductors, the most discussed piece of the datacenter architecture are a minimal part of the footprint. The need for the raw materials to complete a datacenter buildout is what matters. While chip performance may continue to improve, implying a reduction in the resources needed to produce an equivalent unit of compute, overall resource demand can still increase – Jevons Paradox for all to see.
Many refer to the Microsoft study from 2009 that stated a data center requires 27t of copper per MW of capacity for power distribution. Beyond copper, electrical steel, gas turbines, gallium, germanium, silicon carbide, gallium nitride, lithium, nickel, rare earth magnets, and water are an incomplete list of resources and infrastructure to turn a data center from a plan into a working, scaled operation.
The scale of this buildout is in some ways not unique in history with many comparing it to the railroad boom in the 19th century or internet boom in 1999-2001. These booms were capital intensive, put huge pressure on resource demand, and laid the groundwork for productivity booms, though only after incinerating much of the capital that went into creating the boom in the first place. Will the AI datacenter investment binge play out the same way? The market consensus is yes. Therefore, the main risk for investors is thinking about the speed of deployment here and how resource-intensive each piece of infrastructure is. Less can be more, but you better have your analytical insights trained on the right part of this supply chain.
The previous essay concerned demographics and its impact on growth. If multiple companies and countries around the world are devoting greater and greater sums of CAPEX to try and win the data center race and ignite a productivity boom, the resources which underpin this investment move become both an investment and policy priority. This is one of the reasons why the US government has been so aggressive in funding critical materials projects. This support in the form of off-takes, grants, loans, and equity investments can both reduce financing risk, crowding “in” private capital and help with chart a path to profitability.
Governments have been instrumental in helping the private sector develop technologies that have benefitted citizens as a whole including GPS and the internet. So, governments funding critical material projects, semiconductor fabs, and other infrastructure projects is déjà vu all over again to quote Yogi Berra. Whether or not AI works as advertised isn’t entirely what matters. Producing and supplying the raw materials needed to make AI generate a productivity miracle is what generates above average returns. Success in this mission will come through public/private partnerships where capital allocation and risk are shared across the capital stack.
Second Order Demand Drivers
Despite the attention the datacenter buildout receives, there is no shortage of additional demand drivers underpinned by technology that bode well for critical material demand. These include automation and robotics, defense and autonomous systems, and all sorts of AI applied to mining tech such as predictive discovery. While demand on a per unit basis for magnets, for example, in a humanoid robot may not be significant, collectively these totals can add up quickly. Exogenous factors such as Chinese export controls on materials and magnets has artificially pushed up ex-China pricing, forcing supply chain managers and government officials to try to rapidly fix a tenuous situation that can only be fixed in a timeframe measured in years.
The Bear Case
The basic pitch here is that the AI data center build out is particularly resource and energy intensive. What could derail this thesis?
First, could more efficient compute or algorithms reduce infrastructure demand? Doing more with less has been a hallmark of technology. More efficient chips, better architecture and smaller models could help build a more efficient, and smaller AI ecosystem. Another way to think about this is that Jevon’s Paradox could fail to deliver. For those unaware, Jevon’s Paradox states that as a technology improves and increases the efficiency of use, the total consumption of that resource rises, rather than falls.
Second, the overbuild thesis comes true. Think of the 19th Century railroad boom or the dot com boom and bust in 1999-2001. Capacity simply runs too far ahead of demand.
Third, circular financing. Do vendor financing and related-party transactions imply demand that simply doesn’t exist?
Fourth, strong demand may not result in above-average returns for investors. Fierce competition, excess capacity, and falling prices could lower returns across the supply chain. With respect to raw materials, supply expanding faster than demand, an unfortunately common occurrence, could push resource prices down. This may be good for end users, but force producers to make difficult decisions as operating at a loss is not sustainable.
The AI buildout underway has few parallels in economic history. It is likely to play out as a traditional boom and bust cycle and the companies and infrastructure that emerge from the ashes will be particularly well-placed to succeed.
For critical materials investors, you must gauge where the structural constraints on growth exist and find those companies that can earn returns in a highly cyclical environment.
The next essay in this series will explore the Weaponization of Everything.
Act I, Pillar IV – Weaponization
In 1996, New York Times Columnist and foreign policy thinker Thomas Friedman made the claim that no two countries with a McDonald’s had ever gone to war. He also claimed that ‘the world is flat’. Both claims have proven to be false, however, there is nuance here.
While we have learned a great deal about how the global economy functions since those claims, perhaps the biggest lesson is just how interdependent and intertwined the global economy really is thanks to decades of globalization and supply chain optimization. Perhaps the most unpleasant surprise to policymakers is that just as it took decades to make the global economy so intertwined, it may take just as long to “untangle” trade links – supposing this is a good idea, which I don’t think it is. Disrupting this means taking a huge political risk and no policymaker wants to seriously jeopardize their career by getting in front of this freight train.
The pursuit of the lowest cost of production led to concentration of supply chains in many cases. While low-cost goods for consumers were a byproduct of this phenomenon, supply chain optimization has led to more, not less geopolitical competition. Initially, economic policy tools such as tariffs or sanctions were leveraged and this has now expanded to actual goods such as magnets or semiconductors becoming weapons to be wielded by the country that controls that specific supply chain.
These are the building blocks of future economic growth and social stability, so it’s not a surprise. What is surprising and somewhat confounding is the end game which, at this point, is anyone’s guess. Once a vulnerability has been discovered and a chokepoint has been leveraged, it may be impossible to put the genie back in the bottle.
Framing interdependence
Perhaps the best frame for this argument is Henry Farrell and Abraham Newman’s Weaponized Interdependence. In this seminal piece, the authors state the opposite of Thomas Friedman’s argument. The world is not flat. It is, in fact, a series of hubs and spokes with the hubs being crucial to economic leverage. Owning a hub means knowing what flows through it (magnets, raw materials, etc) and where it ends up. It also means you have the leverage to stop the flow of a good to those in need. If what I’m describing sounds like the rare earth magnet supply chain, then you can clearly see why the theory is so powerful.
In the same way the US effectively “controls” access to the US dollar and clearing networks such as SWIFT, China’s control of much of the midstream piece of critical material supply chains is the mirror image. This is the economic game of chicken we find ourselves in today.
Escalation Visualized
Before looking further into why this weaponization of everything may be irreversible, a timeline, or ladder of events that brought us to this point is useful.
Here’s a link showing how more and more goods became weapons, how, and against whom.
What is notable here is that the menu of what can and is ultimately weaponized continues to expand. Also, this isn’t just a “good guy versus bad guy” tit-for-tat. The Trump Administration has been more than willing to threaten allies with the same weapons used on adversaries. Weaponization was once a last resort. Now it’s an opening act in any negotiations and why it is a new phase in the new order introduced in Act I.
What Does a Chokepoint Look Like?
A chokepoint is like pornography – you know it when you see it. But perhaps defining the term is required since chokepoints can take so many different forms. There are at least five factors to consider when assessing a chokepoint:
1. Location – Where on the supply chain does the chokepoint sit?
2. Substitution – Can the chokepoint be substituted for a more readily available good at a suitable price?
3. Time to replace – Supposing substitution is a possibility, how long will this take? Months? Years?
4. Concentration – How much of a supply chain is controlled by a single country or company?
5. Cost – What costs (economic, political, social) must the recipient of the chokepoint bear?
Chokepoints have evolved and financial chokepoints and physical chokepoints behave differently. A financial chokepoint (US dollar access) can be implemented and restored faster than a raw material chokepoint (rare earth permanent magnets, refineries). So, any leverage exercised over raw material supply chains is necessarily longer-lasting – but not indefinite.
Why is it so difficult to untie this knot?
Countries that demonstrate chokepoints are like magicians that show the audience how they do their tricks. Once you’ve seen it, you can’t unsee it. While policymakers in 2020 may not have been aware of Chinese dominance in rare earth permanent magnet production, they absolutely are now.
The natural solution is to raise and deploy capital to mitigate any dependencies. While this makes sense, most supply chain managers are loath to go this route given unfavorable economics and the sheer amount of time it takes to rebuild a supply chain. Multi-year capital commitments for long-lived assets like mines or refineries are difficult to justify when most Western companies are fixated on next quarter’s results. Hence the need for Federal involvement in any supply chain rebuild.
So, there are costs associated with both action and inaction. Weaponizing chokepoints invites retaliation and escalates both the implicit and explicit costs of doing business. Weaponization is not irreversible as political winds can shift rather suddenly. What doesn’t fade is the threat. This could explain why investment ramifications could outlast the possibility of suffering the threat of a chokepoint. Building and operating long-lived assets such as mines or refineries takes time and capital and in order to generate an acceptable return, must run at capacity. Any interruption here and economic value is destroyed. Large infrastructure assets can’t simply be shut off in the same way a tariff can be removed. The investment and political cycles aren’t synched and this can create uncertainty in investment targeting.
The Bear Case
There are at least four reasons the bear case needs to be considered.
First, trade fostered by globalization has proven to be resilient. Change is hard (and expensive), so companies adapt rather quickly. Interestingly, despite a multi-year trade war on a global basis, trade values continue to increase. In 2024, the WTO reported that merchandise trade volume increased by 2.9% and the dollar value of that trade rose to $24.4 trillion. Trade can continue to grow while those trade links and patterns are rearranged.
Second, the effectiveness of export controls is doubtful. Export controls of goods like advanced semiconductors are designed to keep this cutting-edge technology out of the hands of adversaries. While it looks like this policy is negotiable on the part of the Trump administration, these controls often accelerate indigenous innovation of the same good.
Third, as we have seen in the US, policies can reverse within a political cycle and with political cycles shorter than mine or refinery lives, it is difficult to make a huge financial bet when policies are deemed to be fickle. Investment horizon is critical.
Fourth, the old investing adage applies here: nothing cures high prices like high prices. Scarcity premia may sow the seeds of their own destruction unless the economics of a project can be supported at lower-for-longer prices.
We now turn to Act II and look at constraints.