American Enterprise After Abstraction

The difficult object

A large power transformer is engineered for a particular place in the electric grid. Its construction depends on specialized electrical steel, copper conductors, insulation, bushings, cooling systems, and a testing process inseparable from the machine itself. It is expensive, difficult to transport, and slow to replace. A 2022 Department of Energy assessment estimated that more than 90 percent of the electricity consumed in the United States passes through a large power transformer. The object is physically imposing, yet its visible mass represents only the final stage of a much larger production system.

Materials must meet exact specifications, and suppliers carry technical constraints of their own. Engineers account for electromagnetic behavior, thermal performance, mechanical stress, and the conditions of the grid where the machine will operate. Utilities have to forecast need far enough ahead to order something that may take years to arrive. Capital remains committed while steel, components, labor, transport, installation, and testing converge on one piece of equipment. Once the transformer enters service, that coordination disappears into an ordinary fact: electricity continues to move. The finished machine is compressed evidence of institutions, standards, software, equipment, knowledge, and time working together. Its difficulty lies less in any single ingredient than in the accumulated ability to make the ingredients answer to one another.

The achievement of distance

Abstraction is sometimes described as a retreat from the real. The description misses what abstraction makes possible. A corporation allows capital and work to continue beyond the reach of one person. Accounting turns dispersed activity into something that can be compared. Standards permit unfamiliar parties to coordinate. Financial markets connect savings to uses that would otherwise remain beyond personal knowledge. Software carries judgment through repeatable processes and interfaces. Each mechanism creates distance from the original act, and that distance allows capability to travel.

The United States has built formidable strength in these forms. In 2024, knowledge- and technology-intensive industries produced $3.3 trillion in U.S. value added, equal to 11 percent of gross domestic product. The country accounted for 43 percent of global value added in knowledge- and technology-intensive services and 75 percent in software publishing. Code, research, information, finance, and organizational systems belong to the productive core of the economy even when no heavy object marks their output. The scale of the service economy is not proof that productive life has become unreal. It reflects, in part, the extent to which knowledge, coordination, and technical methods have become inputs into nearly every other form of work.

The supposed boundary between an abstract economy and a physical one also dissolves inside the value chain. A recent Bureau of Economic Analysis working paper traces research, software, labor, capital, and imported content through upstream production rather than assigning every input to the industry that performs the final stage. A farm product, for example, may contain little direct research capital in its own accounts while relying on chemical inputs produced through research-intensive processes. Hospital services may embody more upstream software than the hospital industry’s own software intensity suggests. Looking only at the final producer can therefore make a physical product appear less technical, or a service less industrially dependent, than the system required to support it.

Useful distance becomes dangerous when it permits ignorance. A company can retain a supplier contract while losing the competence to judge the supplier’s process. It can own a design while becoming unable to qualify the object made from it. Capital can value a patent without understanding the sequence required to turn the patent into a reliable product. A country may preserve access to a component without knowing how long replacement would take if that access failed. Access, capacity, and capability describe different conditions: the ability to obtain something, the presence of assets that can produce it, and the accumulated ability to produce it repeatedly under relevant conditions. One can remain visible on paper while the others disappear.

The problem is not distance itself. Abstraction becomes fragile when a claim on capability is mistaken for capability, when a contract is treated as a substitute for technical understanding, or when an apparently diversified supply network relies on the same hidden source. The more productive question is whether the abstract layer still conveys the condition of the work beneath it.

Capability and dependency

Semiconductors make the architecture visible. Design, fabrication, equipment, materials, packaging, testing, and software are distributed across firms and geographies. The specialization has produced remarkable technical progress, allowing each layer to deepen its expertise and serve a global market. It has also made clear that strength in design and innovation can coexist with a thinner position elsewhere in the production system. A 2025 Government Accountability Office report described the scale of expanding U.S. fabrication: a new facility can cost several billion dollars and require two to five years to become operational after permitting, construction, utility connections, and equipment installation. The building is one milestone inside a longer process of qualification and repeated production.

Capacity can be financed and installed. Capability has to be accumulated. Suppliers must meet specifications consistently; utilities must deliver power and water at the required quality; technicians must recognize when a process begins to drift; managers must move from qualification to volume without losing yield; customers must commit enough demand to sustain the system. These forms of knowledge sit across companies and institutions rather than inside one asset. A machine on a floor can be counted immediately. The ability to use it reliably, economically, and at the required quality emerges through repetition.

Self-sufficiency is not a sensible response. Specialization works. Trade and multinational production can lower costs, create scale, deepen expertise, and spread the gains from innovation. Research on the interaction of innovation and global production has found aggregate gains from lower barriers to multinational production and has shown that production workers can benefit even in countries that specialize more heavily in innovation. Those results do not resolve every distributional, security, or resilience question. They do defeat the assumption that geographic separation is inherently a loss of competence.

The stronger objective is legibility. An enterprise should know which dependencies are diversified, which can be substituted quickly, and which contain technical knowledge no longer held inside the firm. It should understand the difference between a supplier that provides an ordinary input and one that preserves an ability the buyer could not readily recreate. Public institutions face the same problem at a different scale. Not every imported component is strategic, and domestic location does not guarantee competence. The relevant facts include concentration, substitutability, time to recover, supplier health, operating knowledge, and the consequences of interruption. Geography is one of those facts, not the whole analysis.

For much of the recent era, capital and talent favored businesses that could grow without matching increases in physical infrastructure. The economics were compelling: rapid iteration, global distribution, attractive margins, and feedback measured in days rather than years. Exceptional companies emerged from that model, and their success changed what looked sophisticated. Businesses exposed to long qualification cycles, safety, regulation, procurement, hardware, or heavy working capital often appeared awkward beside software that could be deployed almost instantly.

The renewed interest in difficult enterprise is best understood as a return of consequence rather than a repudiation of software. Energy, defense, biotechnology, manufacturing, logistics, and healthcare still require direct contact with constraints that cannot be abstracted away. Difficulty carries no virtue by itself. A company may be hard to build because the science is immature, the market is weak, or the economics never improve. Difficulty matters when a constraint stands between society and an ability worth possessing, and when credible evidence shows that the company is learning how to remove it.

When abstraction returns to the work

The next era of productive capability will be saturated with software. Simulation can narrow a design space before material is consumed. Machine vision can identify patterns in quality data that an unaided inspection would miss. Artificial intelligence can help engineers search operating histories, compare alternatives, and interpret complex signals. The value of these tools depends on the quality of their connection to the underlying domain. A digital interface is easy to admire because it is visible. The harder work is preserving meaning as information moves across design, production, quality, supply networks, and service.

A 2024 NIST roadmap for digital-thread technology identified recurring barriers in standards, semantics, integration architecture, data ownership, intellectual property, trust, and cost. Data copied from one system to another can lose context even while becoming easier to access. A model does not repeal material limits. A dashboard cannot stabilize a measurement process. An optimization engine can reduce inventory while increasing exposure to a component whose replacement time it does not understand. Technical intelligence appears in the connection between computation and consequence, not in the presence of computation alone.

Services belong inside the same production system. Research, finance, logistics, testing, certification, insurance, training, and distribution often determine whether a technical object can be produced and used at scale. An advanced machine remains unproductive without the institutions capable of financing, qualifying, deploying, and supporting it. A medical technology may embody difficult science and still fail to become useful if reimbursement, clinical practice, training, or regulation prevents adoption. The service intensity of the American economy is compatible with deep productive strength. Its value depends on whether those services extend capability or add further distance from it.

Capital performs a similar act of compression. It compares unlike opportunities through expected returns, risk categories, and time horizons so resources can move beyond what one person already knows. That comparison is necessary. Trouble begins when the model becomes more legible than the business. Construction, prototype, qualification, repeatable production, and customer adoption are different states. Progress in one can expose weakness in another. Technical performance may improve while cost remains uneconomic; production yield may rise while customer integration stalls; a successful pilot may reveal an operating burden that prevents broad use. Serious ownership depends on understanding the sequence by which the relevant risks are expected to decline.

Longer duration can give difficult work time to mature, but time does not create technical competence. Patient capital can support disciplined iteration through uncertainty. The same patience can preserve a weak thesis after the evidence has changed. Capital must be close enough to the work to distinguish the slow accumulation of capability from delay that has acquired a more flattering name. No source of funding can substitute for operating competence. Public institutions and private markets can each support useful pieces of the system; either can finance assets without producing capability if the people, demand, process knowledge, and operating evidence do not develop with them.

Companies are among the places where capability is preserved. They hold methods, supplier relationships, technical standards, customer knowledge, decision rights, and the habit of solving a particular class of problem repeatedly. Their value is not limited to current output or the contracts visible on a balance sheet. They maintain an organized capacity to learn, adapt, and perform work that would otherwise have to be reconstructed from fragments.

The United States already possesses exceptional abstractions in research, software, services, capital formation, and organizational design. The work ahead is not to choose between an abstract economy and a real one. There has never been a clean separation. American enterprise will remain consequential where capital, software, services, and technical knowledge remain answerable to the world they are meant to change.