Every founder eventually runs into the same piece of arithmetic. You can sell a thing that costs almost nothing to reproduce, or you can sell a thing that costs nearly as much to make the hundredth time as it did the first. The first path is how small teams build outsized businesses. The second path is how you end up running a factory. Understanding the difference, and knowing where to look for it, is one of the more useful mental models an entrepreneur can carry around.
Marginal cost of replication is simply the cost of producing one more unit of something you’ve already created. A tailor who hand-sews a jacket pays almost the same in fabric and labor for jacket number two hundred as for jacket number one. A software developer who builds an app pays almost nothing to deliver copy number two hundred thousand, since it’s just data moving across a network. That gap is where enormous margins hide, and it’s why so much of the last thirty years of wealth creation has clustered around a small number of product categories.
Where the Cost Really Disappears
Software sits at the extreme end of this spectrum. Once the code is written, tested, and packaged, distributing it to one more customer is essentially free. Cloud hosting costs scale with usage, but they’re a rounding error next to the price customers pay, especially for anything sold as a subscription. This is why software companies can carry gross margins north of eighty percent while a retailer might celebrate hitting thirty.
Digital information products live in the same neighborhood. An ebook, an online course, a set of templates, a stock photo library, a piece of music, a paid newsletter archive: all of these are expensive or time-consuming to create once and then cost pennies to hand to the next buyer. The entrepreneur’s real expense is almost entirely front-loaded into research, writing, recording, editing, and design. After that, a payment processor and a file server do the rest of the work.
Media and entertainment follow the same logic at a larger scale. A film costs enormous sums to produce and essentially nothing to stream to an additional household. A podcast episode is expensive in time and equipment to record once, then free to serve to the next thousand listeners. This is part of why streaming platforms and podcast networks have been able to scale so aggressively without their costs scaling in step.
Then there’s a category people often overlook: pure information and data. A database, an API that answers a specific question well, a curated dataset, or a piece of proprietary research can be sold over and over to different buyers with no additional production cost per sale. The value isn’t in the atoms, it’s in the fact that someone did the work of assembling and organizing it first.Even certain physical goods can approximate this dynamic, though never quite reach it. Anything manufactured digitally, like a 3D-printed design file or a piece of generative art turned into a print, separates the expensive creative step from the cheap reproduction step. The design is the real product. The physical copy is just an afterthought in cost terms.
Why This Matters for How You Build
The practical lesson isn’t that everyone should build software. It’s that the businesses worth building are the ones where your effort compounds instead of resetting to zero with every sale. When marginal cost approaches zero, growth stops being a linear slog and starts being a matter of distribution. Your job shifts from manufacturing to marketing, from operations to audience-building, from fulfillment to positioning.
This changes what an entrepreneur should actually spend their scarce time on. If you’re building a course, the return on polishing the content one more time and then finding a channel to reach more students is far higher than the return on trying to shave costs out of delivery, because delivery costs are already close to nothing. If you’re building software, the same logic applies to features and customer acquisition rather than to manufacturing efficiency.It also changes how you should think about pricing. Because the cost of the next sale is near zero, you have enormous room to price based on the value you create for the customer rather than the cost you incurred to make the product. A course that saves someone forty hours of trial and error is worth far more than the ten hours it took you to record it, and the pricing should reflect the value delivered, not your production cost.
The Catch
None of this is free money. Low marginal cost usually means low barriers to entry too, since anyone with a laptop can theoretically build a course or a small app. The real cost shifts from production to two other places: the upfront investment required to make something good enough that people want it, and the ongoing cost of getting attention in a crowded market. Distribution, trust, and brand become the moat, because the product itself is trivially easy to copy once someone knows the formula.
The entrepreneurs who do well in these categories tend to treat the creation phase as the place to be genuinely excellent, since that’s the one-time cost that determines whether the near-zero-cost copies are worth anything at all. Then they treat distribution as an ongoing discipline rather than an afterthought, building an audience, a reputation, or a network effect that competitors can’t simply replicate along with the product itself.
The physics of the business hasn’t changed since the printing press made it cheap to copy a book. What’s changed is how many categories of product now behave like books. Recognizing which one you’re building in, and building your strategy around that reality, is often the difference between a business that scales and one that just stays busy.