The Balance Sheet That Broke Economics
In 2006, a researcher at the Brookings Institution named Margaret Blair documented something that had been quietly bothering accountants for years: the market value of American corporations had become almost completely disconnected from their book value. The physical stuff on their balance sheets, the equipment, the real estate, the inventory, explained less and less of what investors were actually paying for.
That gap has only widened since. Apple, at various points the most valuable company on earth, operates with a relatively small amount of physical assets relative to its market capitalization. Google’s parent Alphabet sits on massive server infrastructure, but that infrastructure is not the primary reason the company is worth what it is. The assets that matter most at these companies do not show up on any depreciation schedule.
This is not an accident of accounting. It is the central design principle of the most successful businesses the technology industry has ever produced, and it has profound consequences for how competition works, how wealth accumulates, and why building a moat around a software business looks nothing like building a moat around a factory.
What It Means to Own Nothing Worth Taxing
Start with the basics of why physical assets are a problem, not an advantage, for most businesses.
A factory depreciates. Trucks wear out. Inventory spoils or becomes obsolete. Every physical asset requires maintenance, insurance, and eventually replacement. More importantly, physical assets create fixed costs, which means you are bleeding money whether your production line is running at full capacity or sitting idle. This is the trap that destroyed the American auto industry’s profit margins for decades.
Software has none of these properties. A piece of code written in 2010 does not wear out by 2024. Distributing it to a million users costs almost nothing more than distributing it to ten. The marginal cost of serving an additional customer approaches zero in a way that is simply impossible for any business that moves physical objects around the world.
This is the structural advantage that every serious technology investor understands: software businesses can grow revenue without proportionally growing costs. The economics of scale in physical businesses eventually plateau and often reverse. The economics of scale in software businesses can compound almost indefinitely, which is why the biggest software companies have operating margins that would make a manufacturing executive weep.
The Hidden Assets That Actually Drive Value
So if it is not physical assets, what are investors paying for?
Three things, mostly: network effects, proprietary data, and distribution. None of these appear on a balance sheet in any meaningful way.
Network effects are the phenomenon where a product becomes more valuable as more people use it. The phone network is the classic example, but the modern versions are more interesting. When Uber entered a new city, it needed drivers to attract riders and riders to attract drivers. Once it crossed a critical threshold, the network essentially defended itself. Competitors had to solve the same cold-start problem from scratch, while Uber’s existing network made every new user incrementally more valuable.
Proprietary data is subtler but arguably more durable. Google processes billions of searches per day. Every search that results in a click is a data point about what people actually want versus what they typed. A company that started a search engine today would have better algorithms and cheaper servers than Google did in 1998. They would not have two decades of behavioral data. That data is the moat, and it is invisible on any financial statement.
Distribution is the least glamorous of the three but often the most decisive. Microsoft’s dominance in enterprise software is not primarily about product quality. It is about the fact that IT departments at large organizations have built workflows, integrations, and institutional knowledge around Microsoft products. Switching costs are a form of asset that belongs entirely to the company, appears nowhere in its filings, and is extraordinarily difficult for competitors to erode. Enterprise software contracts are designed to be impossible to cancel precisely because the switching cost asset only works if leaving is painful.
How You Build Something That Cannot Be Copied
The standard startup advice about building defensible businesses misses the point in an interesting way. Most founders think about defensibility in terms of features, which is wrong. Features can be copied. The companies that built lasting advantages built something more like gravity.
Facebook in 2008 had no technology advantage over a well-funded competitor. Its code was not particularly elegant. Its servers were not magical. What it had was everyone you knew already on the platform. The network itself was the product, and the network could not be replicated by writing better code.
This is why tech giants buy competitors they could build instead of building them. When Facebook acquired Instagram for one billion dollars in 2012, Instagram had thirteen employees. Facebook was not buying their engineering talent or their servers. It was buying their user network before it could grow large enough to compete directly. The physical and technical assets were almost irrelevant to the price.
The same logic applies to acquisitions of companies with valuable data sets, distribution relationships, or proprietary algorithms. The acquirer could often build a technical equivalent from scratch. They cannot replicate the intangible asset that took years to accumulate.
The Liability Side of Owning Nothing
There is a version of this story that gets told in business schools as pure triumph, and it leaves out something important.
When a company’s assets are intangible, they are also fragile in ways that physical assets are not. A factory that burns down can be rebuilt with insurance money. A reputation destroyed by a data breach cannot be. Trust, once lost, does not depreciate on a predictable schedule, it can simply disappear.
More pointedly: companies that own almost nothing physical often depend entirely on platforms or infrastructure they do not control. An app business built on Apple’s App Store discovered this in painful clarity when Apple changed its review policies or its fee structures. A media company built on Facebook’s distribution learned it when Facebook’s algorithm changed. These businesses had real intangible assets, audiences and content libraries, but their access to those assets depended on relationships with platform owners who had very different interests.
The WeWork story is instructive in a different direction. WeWork tried to claim it was a technology company with intangible assets rather than a real estate company with enormous physical liabilities. The market was not fooled for long. The actual asset being sold was leases, the most physical and fixed cost imaginable. The technology framing was mythology layered on top of a balance sheet that looked nothing like the companies it was trying to emulate. The SaaS version of this story is quieter and harder to see, but the underlying confusion between real intangible assets and narrative intangible assets shows up everywhere.
Why This Changes How Monopoly Works
Traditional antitrust thinking developed in a world of physical assets. Standard Oil was powerful because it controlled the physical infrastructure of oil refining and distribution. Breaking it up meant breaking up actual pipes and facilities. The power was visible and divisible.
Intangible asset monopolies are different. You cannot break up a network effect by dividing a company’s servers. If you forced Facebook to spin off Instagram, you would have two smaller networks, each less valuable than the combined one, but the users would not redistribute evenly to competitors. They might just consolidate onto whichever half kept more of their friends.
You cannot break up a data advantage by deleting half the data. The regulatory frameworks built around physical-asset monopolies are genuinely poorly suited to companies whose power lives in accumulated behavioral data and entrenched network effects. This is not a defense of those companies. It is an observation that the tools built for one kind of market power work badly on another kind.
Governments are starting to recognize this, but the solutions remain poorly defined. The European Union’s Digital Markets Act attempts to address it through interoperability requirements, essentially forcing dominant platforms to share access to the network in ways that reduce switching costs. Whether that works in practice is an open question.
The Copycat Problem and Why It Matters Less Than It Should
The obvious counter-argument to everything above is: if your advantage is intangible, why can’t a competitor just build the same intangible thing?
Sometimes they can. But time is the hidden variable. Google took years to accumulate the search query data that makes its results better than a cold-start alternative. Amazon spent a decade building the logistics relationships and seller trust that makes its marketplace work. These advantages compounded slowly and then became almost impenetrable.
The companies that win with intangible assets usually win by getting somewhere first and then growing the advantage faster than competitors can close the gap. The window where you can actually catch up is real but often shorter than it looks from outside. By the time a competitor is obviously dangerous, the incumbent has usually been compounding their advantage for long enough that catching up requires more than matching their current state, you have to match where they will be in three years, because that is when your product will actually reach market scale.
What This Means
The shift from physical to intangible assets is not a feature of tech companies specifically. It is the direction the entire economy has been moving for decades, with technology companies simply at the leading edge. Understanding it matters for several reasons.
For founders: the question worth asking about any business is not what you own but what would be hard to replicate in three years if a well-funded competitor started today. Physical assets almost never answer that question satisfactorily. Network effects, proprietary data, and genuine switching costs sometimes do.
For investors: book value has become nearly meaningless as a valuation input for software companies. The assets that matter are exactly the ones that are hardest to measure, which creates both the opportunity (mispricings when intangible assets are underappreciated) and the risk (mispricings when intangible assets are invented whole cloth, as with WeWork).
For regulators: the tools built to address market concentration in physical industries need significant updating. A company can have enormous market power without owning much of anything that can be seized, divided, or sold at auction. The power is real. It just lives somewhere harder to point at.
The most valuable companies on earth got that way by building things that cannot be moved, taxed, or burned down. That is either genius or terrifying depending on which side of the balance sheet you are on.
