The Economics of Scale
Why the Second Copy Costs Almost Nothing
The most consequential fact about software economics is one that most people who use software every day have never consciously noticed.
The Idea
When a factory makes a second car, it needs more steel, more labour, more time. When a software company makes a second copy of its product, it needs almost nothing. This distinction — between physical goods with high marginal costs and digital goods with near-zero marginal costs — is arguably the single most important structural fact about the tech industry. The first copy of a piece of software is extraordinarily expensive: years of engineering, design, testing, infrastructure. But the second copy, the millionth copy, the hundred-millionth copy? Essentially free to produce. Bandwidth and server costs exist, but they are trivially small compared to what it costs to build the thing in the first place. This creates a profoundly unusual economics. In most industries, growing means spending more. In software, growing means spreading a fixed cost across an ever-larger base — so the cost per user falls continuously as the user base expands. At sufficient scale, the economics become almost absurdly favourable: revenue grows, but the underlying cost of delivering the product barely moves. The implications cascade outward. It explains why tech companies can offer products for free and still build enormous businesses. It explains why they pursue scale so aggressively — not out of ambition alone, but because the underlying math rewards it. And it explains why, once a dominant platform is established, it becomes nearly impossible to dislodge: the incumbents have already paid the fixed cost that any challenger must now absorb from scratch.
In the World
In 2001, Microsoft was the clearest example of this logic working in practice. Windows had cost hundreds of millions to develop over many years. But once it existed, licensing it to the next PC manufacturer cost Microsoft almost nothing to fulfil. The same code, the same disc, the same installation. Every additional sale was nearly pure margin. The result was a company that by the late 1990s was generating profit margins that manufacturers of physical goods could barely dream of. But the more clarifying example came later, with Satya Nadella's transformation of Microsoft into a cloud company after 2014. Azure, Microsoft's cloud platform, exhibited the same near-zero marginal cost logic — but amplified. Once the data centres were built and the software written, adding a new enterprise customer largely meant switching on access. The infrastructure was already there. Revenue scaled; delivery cost barely budged. By 2023, Microsoft's cloud and software divisions were generating operating margins above 40 percent — figures that would be impossible in almost any other industry at that scale. Compare this to a supermarket chain, operating on margins of two or three percent, and the structural difference becomes viscerally clear. The supermarket must restock every shelf every day. Microsoft's cloud does not need to be rebuilt each morning. The second copy really does cost almost nothing.
Why It Matters
Understanding near-zero marginal cost changes how you read the tech industry — its strategies, its pricing, its politics. When a platform offers something for free, it is not being generous. It is pursuing scale, because scale is where the economics eventually pay off. When a tech company raises prices sharply after achieving dominance, that is the fixed cost already paid, now being monetised. When regulators struggle to apply traditional antitrust logic to software markets, part of the difficulty is that conventional economics — built around the assumption that producing more costs more — does not cleanly map onto this world. On a more personal level, it reframes how you might evaluate any technology business you encounter, invest in, or work for. The question worth asking is not just 'is this product good?' but 'what happens to the unit economics as this scales?' A company that has found genuine near-zero marginal cost dynamics is sitting on a very different kind of machine than one that must hire proportionally as it grows. That difference, compounded over years, tends to be where the most durable value — and the most concentrated power — accumulates.
A Question to Ponder
If the cost of delivering a product falls toward zero as it scales, what should determine its price — and who should get to decide?
Get a new one of these every morning.
Start learning with Thinkable