Every era of capitalism develops a center of gravity for capital: the part of the economy where technological change, economic necessity, and institutional attention combine to produce outsized returns. Those centers move. They reflect the constraints of their time.

For much of the past half century, the center of gravity lay in the intangible economy. The developed world entered that period with much of its physical foundation already built. Energy was abundant. Supply chains expanded globally. Security was underwritten by a stable geopolitical order. Manufacturing could migrate toward lower costs, while software, networks, and finance scaled across an installed industrial base.

Capital responded rationally. The highest-return businesses were often those that could grow without rebuilding the physical systems beneath them. Asset-light models outperformed asset-heavy ones. Distribution mattered more than production. Software margins became the reference point against which other businesses were judged.

That era produced extraordinary companies. Its investment logic was correct for its conditions.

The conditions are changing.

Security has again become an industrial question. Electricity demand is rising after a generation of relative stability. Supply chains once optimized principally for efficiency are being reconsidered in light of strategic dependence. Defense inventories are proving easier to consume than to replenish. Artificial intelligence, despite appearing to be the ultimate intangible technology, is becoming one of the most intensive consumers of physical infrastructure in the modern economy.

Capital is beginning to follow.

I call this transition the Productive Capacity Supercycle: the multi-decade reallocation of capital toward the systems that determine how much an economy can produce, how securely it can operate, and how effectively a nation can convert technological capability into economic and strategic power.

This is not a rotation from software into hardware. It is a shift in where the world’s binding constraints now sit.

The Inheritance Is Being Depleted

The industrial systems of the twentieth century created an unusually rich inheritance. Electrical grids, ports, factories, energy systems, defense capacity, transportation networks, and large pools of technical skill already existed by the time the digital economy began its ascent.

Globalization extended the value of that inheritance. Production moved across borders. Redundancy was removed. Inventories fell. Capital intensity declined. The marginal dollar increasingly earned more by coordinating existing capacity than by constructing new capacity.

Over time, however, the distinction between a successful investment framework and a permanent law of economics began to blur.

Manufacturing complexity became something to avoid rather than something that could create scarcity. Capital intensity became a weakness rather than a variable to be evaluated against the durability of the resulting asset. Long procurement cycles were treated as structurally inferior to rapid commercial adoption even where the customer was a sovereign government with persistent strategic demand.

The underlying physical base meanwhile continued to age.

Grids depreciate. Shipyards disappear. Skilled workforces retire. Supply chains consolidate. Industrial knowledge moves with production. Defense capacity designed for limited peacetime replenishment does not automatically expand when strategic conditions change.

For decades, the developed world could treat much of this productive base as fixed infrastructure.

It was never fixed.

Constraint Has Returned

Three forces are now converging upon that inherited base.

The first is geopolitics. The postwar international system has passed from bipolarity to unipolarity and is now moving toward a more multipolar order. Power is becoming more distributed, strategic competition more persistent, and national economic capability more closely tied to geopolitical influence.

This competition is also unfolding against a radically steeper technology curve. The Cold War was a technological contest, but many of its decisive capabilities were developed through relatively concentrated state and defense-industrial systems. Today, advances in artificial intelligence, autonomy, semiconductors, cyber, space, energy, robotics, and advanced manufacturing are occurring simultaneously, often originating in commercial markets and diffusing rapidly between civilian and military use.

The result is a form of great-power competition in which technological and industrial capacity are increasingly inseparable from national power. A country’s ability to invent matters, but so does its ability to manufacture, deploy, replenish, power, and continuously improve what it invents.

Energy, semiconductors, defense production, communications infrastructure, critical minerals, compute, and advanced manufacturing are therefore increasingly treated not simply as markets, but as strategic capabilities.

That changes the economic value of resilience itself. A supply chain optimized for cost under conditions of geopolitical stability can become a liability when security of supply matters. A domestic manufacturing process that appears uneconomic in a frictionless world can acquire substantial value when availability, control, and resilience are priced explicitly.

The second force is artificial intelligence.

AI does not make the physical economy less consequential. It makes physical capacity more essential.

Computation requires semiconductors, data centers, transmission, cooling, transformers, turbines, and electricity. As intelligence moves from screens into autonomous systems and industrial processes, the physical requirements expand further into robotics, sensors, communications, manufacturing, and energy.

The ultimate expression of artificial intelligence is not intelligence confined to a screen, but intelligence acting autonomously in the physical world. A world of mass-produced, autonomous, dexterous robots would transform software intelligence into physical labor at unprecedented scale. The more capable intelligence becomes, the larger the economic opportunity to embody it in machines that can build, move, manufacture, repair, defend, and produce.

Intelligence may be expressed in software, but intelligence at scale has an industrial base — and its fullest expression is physical.

The third force is accumulated underinvestment colliding with inelastic supply. In many strategically important systems, demand is now rising against supply that cannot respond quickly. The constraint may be permitting, skilled labor, manufacturing throughput, certification, specialized components, procurement structures, or simply the time required to construct complex assets.

This matters because scarcity changes economics.

When capacity is abundant, ownership of capacity is easily commoditized. When capacity becomes scarce, the ability to create it can become a source of pricing power, strategic relevance, and durable advantage.

The central change is therefore simple: the world’s binding constraints are moving back into productive capacity.

What Productive Capacity Means

The Productive Capacity Supercycle is the reallocation of capital toward the capabilities required to expand and secure the productive base of advanced economies.

It includes energy generation and transmission, compute infrastructure, semiconductor capacity, defense production, autonomous systems, robotics, advanced manufacturing, communications infrastructure, and space logistics.

Energy itself may become a frontier of capacity creation rather than simply a constraint to be managed. Advanced nuclear fission — and, if it crosses from scientific achievement into economically deployable infrastructure, fusion — could materially expand the amount of energy available to the economy and alter the productive ceiling of everything built above it.

These are not one sector. Their commonality is functional.

They determine the operating ceiling of the systems above them.

That distinction matters because productive capacity is not synonymous with physical hardware. Software that materially increases factory throughput expands productive capacity. Autonomy that allows machines to perform work previously constrained by human labor expands productive capacity. Communications architectures that allow distributed assets to operate reliably in contested environments expand productive capacity.

The relevant distinction is not digital versus physical.

It is whether a technology increases the capacity of the underlying system.

The cycle is therefore not simply about rebuilding what has been depleted. It is about restoring inherited capacity, expanding it where demand exceeds supply, transforming it through intelligence and autonomy, and creating entirely new forms of productive capacity as the technological frontier advances.

Seen this way, the productive economy is not replacing the digital economy. It is absorbing it.

The most consequential companies of the next era are likely to combine sophisticated intelligence with difficult physical execution. Software moves from sitting above the industrial system to becoming embedded within it: controlling factories, optimizing power systems, coordinating autonomous fleets, operating defense platforms, and managing orbital infrastructure.

Artificial intelligence represents the furthest extension of this transition. Its ultimate economic expression is not simply a better interface to information. It is intelligence acting upon the physical world.

A mass-produced autonomous, dexterous robot is perhaps the purest example: software, compute, sensing, actuation, manufacturing, and energy combined into a machine capable of performing productive labor. Intelligence ceases merely to augment human decisions and begins directly expanding the economy’s capacity to act.

The boundary between technology and industry becomes progressively less useful.

Markets Still Carry the Instincts of the Last Era

Capital markets adapt more slowly than economic regimes.

Investors develop pattern libraries from experience: which margins signal quality, which business models deserve premium valuations, which sales cycles indicate weakness, and which forms of capital intensity destroy returns. These heuristics are valuable because they compress decades of accumulated judgment.

They can also become misleading when the environment changes.

Much of the productive economy is still evaluated using assumptions formed during the era of software abstraction. Hardware is discounted for historical gross margins. Defense is discounted for procurement cycles. Energy is discounted for capital intensity. Manufacturing is discounted for operational complexity.

Those observations are not inherently wrong.

The mistake is assuming that the economics of emerging categories must resemble those of their predecessors.

Software can transform the marginal economics of hardware. Automation can alter manufacturing cost structures. New architectures can make defense systems reproducible at software-like iteration speeds. Strategic procurement can convert uncertain demand into long-duration contracted revenue. A technology that removes a severe capacity constraint can create a market substantially larger than historical comparisons imply.

This creates Structural Mispricing.

The mispricing is often temporal rather than informational. The market may understand the technology and still underestimate how quickly strategic necessity becomes economic demand. It may correctly observe present margins while failing to see the future structure of the category.

Technological truth and investment truth therefore remain separate.

An important technology can be a poor investment at the wrong price. Conversely, a company that looks conventionally unattractive can become an exceptional investment if the category around it is undergoing a structural change that the prevailing framework does not yet capture.

The task is not simply to identify important technologies.

It is to identify when necessity, technological feasibility, and company formation have converged into an investable market.

A Different Investment Discipline

That requires a different starting point.

Company-level underwriting remains essential. Technology must work. Founders must execute. Customers must buy. Capital must earn a return.

But when categories themselves are forming, bottom-up analysis alone is insufficient. The analysis must begin one level higher: with the forces determining which capabilities will become necessary.

That means understanding not only present demand, but the conditions likely to create future demand. What are adversaries building, and what vulnerabilities will those capabilities expose? Which technologies emerging at the scientific and engineering frontier could alter the strategic calculus of nations? Where are critical supply chains concentrated or fragile enough to become strategic dependencies? And which new threat vectors are visible in research today before they become operational realities?

The same analysis must consider the other side of technological change: what has newly become possible. Scientific advances, falling component costs, new manufacturing techniques, improvements in autonomy, or changes in computing architectures can make feasible an approach that would have been uneconomic or technically impossible only a few years earlier.

The categories visible today are unlikely to exhaust the cycle. Some of its most consequential expressions may still exist principally in laboratories. As technologies cross thresholds of scientific feasibility, cost, manufacturability, and deployment, entirely new forms of productive capacity can emerge. Brain-computer interfaces, new biological manufacturing systems, novel materials, or technologies not yet commercially legible may eventually belong to the same framework if they materially expand what individuals, machines, or economies are capable of producing.

The intersection of these forces matters. New vulnerabilities create requirements. New technologies create solutions. Strategic necessity creates demand.

The investment question is whether these changes are temporary or structural, how large a market becomes when a capability moves from optional to essential, whether incumbents can satisfy the requirement, and which new company is positioned to define the resulting category.

This is Category Foresight: identifying economically consequential categories before consensus has fully named them.

The relevant sources therefore extend beyond conventional market analysis. Defense strategy, adversary research, scientific literature, energy systems, industrial bottlenecks, supply-chain concentration, procurement plans, technical roadmaps, government budgets, and geopolitical developments all become inputs into investment judgment.

These signals often appear before conventional market data does.

An adversary’s research program can reveal a future vulnerability years before procurement responds to it. A breakthrough at the technological frontier can create a capability for which no established market yet exists. A concentrated supply chain can remain economically efficient for decades and then become strategically unacceptable almost overnight.

The objective is not simply to predict which existing markets will grow. It is to understand how technological and strategic change alters what economies and nations must be capable of doing — and therefore where new markets must form.

This is not thematic investing. A theme identifies something that may become important.

Category Foresight seeks to understand what becomes possible, what becomes vulnerable, and what therefore becomes necessary — and to identify the company capable of translating that necessity into a durable economic position.

In the last era, investors could often begin by asking whether a company could capture an existing market more efficiently.

In the emerging one, the more consequential question is increasingly whether a new market is about to become unavoidable.

We call the investable expression of this shift Global Resilience. It is not a sector, and it is not synonymous with defense. It encompasses the technologies and companies that expand, secure, automate, or unlock the productive capacity on which economic and strategic power increasingly depend.

Energy, compute, manufacturing, autonomy, defense, space, and the categories still emerging from the technological frontier are connected by the same underlying question: which capabilities become indispensable as technological possibility expands and strategic constraints tighten?

The Return of the Productive Frontier

Capitalism has seen this pattern before.

Railroads, electrification, automobiles, telecommunications, aviation, and the postwar industrial buildout each required enormous amounts of capital before their full economic consequences were understood. Each generated speculation, overbuilding, and failure. Each also produced extraordinary fortunes and helped create the institutions that defined the era that followed.

The lesson is not that industrial buildouts are inherently attractive.

It is that when the productive frontier moves, the locus of wealth creation moves with it.

That frontier is moving again.

The physical systems inherited from the twentieth century are now being asked to support an economic and strategic environment for which they were not designed. The energy system must power a computational economy of unprecedented scale. Manufacturing must become more automated, distributed, and resilient. Defense production must evolve from exquisite platforms produced in small quantities toward autonomous, software-defined systems capable of being manufactured at volume. Compute itself is becoming infrastructure on the scale of energy. Space is becoming an operating layer of the terrestrial economy. And artificial intelligence is moving beyond digital applications into machines capable of perceiving, deciding, and acting directly upon the physical world.

These developments are often treated as separate industries.

They are better understood as expressions of the same underlying capital cycle.

The productive economy now being built will not resemble the industrial economy of the past. It will be more computational, more autonomous, more software-defined, and more deeply intertwined with national strategy. Intelligence will be embedded throughout the physical systems on which economic and strategic power depend.

For several decades, the productive frontier lay principally in abstraction: software, networks, information, and financialization. The highest-value technologies were often those that could scale across a physical base that had already been built.

The frontier is now moving toward the systems that make that abstraction possible — and toward the machines through which intelligence acquires agency in the physical world.

Energy. Compute. Manufacturing. Autonomy. Defense. Space.

Together, they form the infrastructure through which technological power becomes productive capacity, and productive capacity becomes economic and strategic power.

This is the capital cycle now taking shape: the Productive Capacity Supercycle.

The supercycle is the direction of history’s capital. We intend to finance it.