The Shifting Boundaries

The marvel is that in a case like that of a scarcity of one raw material, without an order being issued, without more than perhaps a handful of people knowing the cause, tens of thousands of people whose identity could not be ascertained by months of investigation, are made to use the material or its products more sparingly.

Each paradigm shift leaves behind a layer of capital that was once decisive and is now merely necessary. The canal companies did not disappear when the railways came. They became feeders to the rail network, handling short-haul bulk where water transport still made sense. The telegraph companies did not vanish when the telephone arrived. They persisted for decades in niches where written record mattered more than voice. The pattern is general: the assets of the old regime remain useful, sometimes indispensable, but they no longer command the margin. The margin migrates to whoever controls the new binding constraint.

Three thinkers from different disciplines reach the same conclusion: over long horizons, finance and institutions crystallize around the coordination demands of production at scale.

Alfred Chandler, tracing the rise of the modern corporation, showed that the managerial revolution of the late nineteenth century was an organizational response to throughput demands.(Alfred D. Chandler 1977)Alfred D. Chandler, Jr., The Visible Hand: The Managerial Revolution in American Business (Cambridge, MA: Belknap Press of Harvard University Press, 1977).View in bibliography Railroads required coordination across great distances, steel mills required continuous-flow production, and oil refineries required scheduling that matched extraction, transport, and processing. The hierarchical, multidivisional corporation emerged because the market alone could not coordinate these flows at the required speed and scale. Banks and capital markets grew to serve the needs of industrial coordination, channeling savings into the infrastructure that the new technologies demanded. Finance followed the production function.

Joel Mokyr, working from the history of technology, reached a compatible conclusion by a different route.(Mokyr 2002)Joel Mokyr, The Gifts of Athena: Historical Origins of the Knowledge Economy (Princeton, NJ: Princeton University Press, 2002).View in bibliography His concept of "useful knowledge," the stock of techniques, practices, and understanding that can be applied to production, clarifies what energy and conversion efficiency explain and what they do not. Energy provides the throughput: useful knowledge determines how efficiently that throughput is converted into output. A steam engine is useless without coal, but it is also useless without the metallurgical knowledge to cast the cylinder, the mechanical knowledge to fit the piston, and the operational knowledge to run the machine without blowing it up. The two factors are complementary.

Mancur Olson, from political economy, completed the triangle.(Olson 1982)Mancur Olson, The Rise and Decline of Nations: Economic Growth, Stagflation, and Social Rigidities (New Haven, CT: Yale University Press, 1982).View in bibliography Institutional quality matters enormously for growth, but institutions themselves are shaped by the constraints and opportunities of the era. A society with abundant coal and nascent industry will develop different institutions than a society with abundant land and dispersed agriculture. The institutions are not exogenous gifts from enlightened lawmakers: they are endogenous responses to the resource base. Good institutions can accelerate the exploitation of that base; bad institutions can squander it. The base sets the terms within which institutions operate.

Institutions and finance harden around throughput bottlenecks. When the bottleneck moves, boundaries move with it.

The conventional framing, which treats energy as a commodity input and ideas as the real driver, gets the relationship backward. Labor without dense exosomatic energy stays close to subsistence. Capital without energy is inert metal. In practice, energy and conversion efficiency set the ceiling within which the rest of the production function operates.

The current transition fits the pattern. If the binding constraint is shifting from information to computation, and computation is physically grounded in energy and hardware, then the assets that mattered in the information age—software, data, network effects—may be necessary but no longer sufficient. Physical constraints—compute allocation, power interconnection, supply-chain bottlenecks—determine who can build capacity and how fast. The margin will migrate to whoever controls that capacity.

The companies that accumulated advantages in the previous regime will not disappear, but their advantages will erode unless they can translate them into control over the new binding constraint. Software moats remain valuable, but software alone cannot clear the interconnection queue or secure transformer delivery. Network effects remain powerful, but they do not generate the gigawatts required to train and run frontier models at scale. The question is whether incumbents can convert their existing positions into the new form of capital before the constraint shifts again.

Financial metrics can be actively misleading during transitions, because they measure performance against assumptions that the new regime will invalidate.

The Pennsylvania Railroad was among the largest corporations in the world by market capitalization by 1900—worth more than the U.S. Treasury. Its network was unrivaled; its operational discipline was the template for industrial management; its bonds were held by the most conservative institutions in America. By 1970, it was bankrupt, merged in desperation with its longtime rival, the New York Central, into Penn Central, which collapsed within two years. The transition took ninety years. This one will not. The assets remained; the network still existed; the trains still ran. What had changed was the constraint. Trucks and automobiles offered flexibility that fixed routes could not match; highways reached places that rails did not; the logistics geometry of the postwar economy no longer centered on the corridors the railroads had built. The Pennsylvania Railroad's market capitalization in 1900 was measuring dominance of a constraint that was about to shift.

The market capitalizations of the great oil majors looked similarly impressive in 1970. ExxonMobil, Shell, and BP were among the largest companies on earth, and the logic of their position seemed unassailable: the world ran on oil, demand was growing, and they controlled the infrastructure to extract, refine, and distribute it. By 2020, the industry was facing questions about long-term demand that would have seemed absurd a generation earlier. The constraint had not yet fully shifted, but the anticipation of the shift was already visible in relative valuations. Technology companies with a fraction of the oil majors' tangible assets commanded multiples of their market capitalizations.

The pattern is not that incumbents always lose: the transitions take decades to unfold. The pattern is that the metrics of the old regime do not reliably predict success in the new one. A high stock price and strong network effects may appear dominant, but if the business model depends on assumptions that the new regime will invalidate, the valuation is a lagging indicator dressed as a leading one.

Finance amplifies whatever the production function rewards: this is its nature and its limit. When the production function rewards control over coal and steel, capital flows to coal and steel. When it rewards control over software and networks, capital flows to software and networks. When it rewards control over energy-structured-into-computation, capital flows there. It already is, visible in the scale of investment pouring into data center construction, chip fabrication, and power infrastructure. Capital reliably migrates to the new constraint. Ownership and terms determine who captures the rents.

When the binding constraint moves, the locus of durable advantage moves with it. This is exactly what Braudel's third floor records.

The market economy will continue to function as it always has, with prices coordinating decentralized decisions and competition disciplining inefficiency. But the third floor, the zone of large-scale position and durable advantage, will be occupied by a different set of players than it was in the previous era. The third floor in the information age was occupied by platform owners: those who controlled the network effects, the data flywheels, the app stores, the advertising marketplaces. The third floor in the emerging regime will be occupied by those who control the physical infrastructure that makes large-scale computation possible: the fabs, the data centers, the power purchase agreements, the grid interconnection rights, the cooling capacity, the supply chains for high-bandwidth memory and advanced packaging. These are heavy, expensive, slow to build, and subject to constraints that software cannot route around. The new operators of the third floor will be those who can deploy capital at the intersection of energy and computation, at a scale that forecloses competition before the next re-ranking arrives.

A structural question shadows this transition. The third floor has historically been occupied by entities with legal standing: corporations that can sue and be sued, institutions that can accumulate reputation across decades, principals who can be held accountable for contractual breach. Autonomous computational agents, lacking persistent identity and legal personality, cannot occupy this floor through the mechanisms that served prior occupants. If the production function increasingly routes through such agents, either the agents must acquire standing through new legal categories, or coordination must occur through mechanisms that substitute for legal recourse. The settlement infrastructure that emerges will determine which path prevails.

Once energy becomes abundant, the battle is no longer over fuel but over permission: who may turn throughput into property, and under what terms. But throughput itself obeys laws older than any charter. Intelligence has a thermodynamic cost, and that cost sets limits no institution can repeal.