The Simple Version
The most valuable companies in the world make most of their money from things you cannot drop on your foot. That is not a coincidence. It is the entire strategy.
What the Balance Sheets Actually Show
Compare two companies that most people would call enormous: Ford Motor Company and Meta Platforms. Ford operates more than 50 manufacturing plants globally, employs around 175,000 people, and holds tens of billions of dollars in physical property, plant, and equipment. Meta, as of recent filings, has a market capitalization that dwarfs Ford’s by a factor of roughly ten, despite owning a relatively modest collection of data centers and office buildings. The majority of Meta’s value sits in intangible assets: software, algorithms, user data, and the network effects that make its platforms more valuable as more people join them.
This pattern repeats across the most valuable companies in technology. Apple’s market capitalization has exceeded $3 trillion. Its physical assets, the buildings and equipment it actually owns, are a small fraction of that figure. Most iPhones are assembled by Foxconn in China. Apple designs the chips, writes the software, controls the retail experience, and captures the margin. The factories belong to someone else.
This is not a loophole. It is a deliberate architectural choice about where value gets created and where risk gets transferred.
Why Physical Assets Are a Trap
Owning factories sounds like strength. In many industries, it is. A steel company without a blast furnace is not a steel company. But physical assets come with fixed costs that do not shrink when revenue does. During the 2008 financial crisis, General Motors was paying to heat, staff, and maintain plants that were running at a fraction of capacity. The assets that defined the company nearly destroyed it.
Software has a fundamentally different cost structure. Writing code costs money. Deploying that code to one additional user costs nearly nothing. This property, economists call it near-zero marginal cost, means that software businesses can scale revenue without scaling expenses in proportion. A factory that produces twice as many cars needs roughly twice the raw materials and labor. A social media platform that doubles its users does not double its server costs, and it certainly does not double its engineering headcount.
This asymmetry is why investors assign high multiples to software revenue and lower ones to manufacturing revenue. The market is not being irrational. It is pricing in the difference between a business that gets more efficient as it grows and one that does not.
The Real Asset Is the Moat, Not the Building
When analysts talk about intangible assets, they mean things like brand value, patents, proprietary data, and network effects. These are genuinely difficult to value on a balance sheet under standard accounting rules, which creates a strange situation: the most important assets of the most valuable companies are largely invisible in their official financial statements.
Google’s search index represents decades of crawling, ranking, and refining. A competitor could theoretically build something equivalent, but the time and capital required create a practical barrier. Visa’s network of banks, merchants, and cardholders took fifty years to assemble. A startup with better payment technology still has to solve the problem of getting accepted everywhere, which the technology alone cannot fix.
This is what makes intangible assets so defensible. Physical assets can be replicated given enough capital. A rival can build a factory. Nobody can instantly replicate the trust, the habit, and the network density that make certain tech platforms the default choice for billions of people. The most disruptive startups solve problems nobody has officially admitted exist yet, but even the best newcomers often find themselves stalling out against incumbents whose real competitive advantage is invisible on any spreadsheet.
How This Shapes Business Strategy
Once you understand the logic, a lot of otherwise puzzling tech company behavior becomes legible.
Apple outsources manufacturing not because it cannot afford factories, but because owning factories would mean accepting the fixed-cost structure of a manufacturer. Instead, it captures the high-margin design, software, and services layers and lets suppliers absorb the capital intensity and cyclical risk. Apple’s gross margin on services consistently runs well above its hardware margin, which is why the company has spent years pushing customers toward iCloud, Apple Music, and the App Store.
Amazon Web Services follows similar logic. Amazon built the cloud infrastructure and then rented it out, turning a fixed cost into a recurring revenue stream. The physical servers exist, but the value the market assigns to AWS is based on the software abstractions built on top of them, the proprietary tools, the ecosystem of developers who have built their businesses on AWS and face real costs if they switch.
This also explains why tech companies spend so aggressively on research and development rather than capital expenditures. R&D spending builds intangible assets. It creates the next version of software, the next algorithm improvement, the next feature that extends the moat. Capital expenditure builds physical things that depreciate in straightforward, predictable ways. Tech companies lose money for years on purpose partly because they are front-loading the cost of building intangible assets that accountants cannot fully capture on a balance sheet.
What This Means for Everyone Else
The asset-light model has limits. Data centers are expensive. The compute required to train large AI models costs hundreds of millions of dollars per run. When a business reaches the scale of Google or Microsoft, physical infrastructure becomes a real and growing cost. The near-zero marginal cost story gets messier at the frontier.
There is also a concentration problem. When competitive advantage comes from data, network effects, and proprietary algorithms rather than from physical capacity anyone can build, it becomes structurally harder for new entrants to displace incumbents. A new car company can buy or lease manufacturing equipment. A new search engine cannot buy twenty years of query data and user behavior signals.
The asset-light model does not just make certain companies more valuable. It makes them more durable, and more resistant to competitive pressure, than almost any previous generation of large businesses. That is the real reason investors assign the valuations they do. The absence of physical assets is not a vulnerability. It is the point.
