On Moats, Perfect Competition, and Your Business's Net Profit

I recently took part in yet another discussion about moats and barriers to entry. Among startup founders, it's quite common to hear that moats are just another obsession of venture capital investors. Whenever I share my alternative perspective, I usually receive plenty of negative comments. Still, I'd like to explain once again why barriers to entry matter. You can decide for yourself whom to believe.

To understand my argument, you first need to understand the concept of perfect competition. Perfect competition is a market structure in which a large number of firms sell an identical product, and each individual firm is so small that it cannot influence the market price on its own. Moreover, if new entrants see that firms in the market are highly profitable, they begin launching similar businesses because there are no barriers to entry. Why not enter a profitable market?

Now we'll need to write down a simple formula and take a derivative. Don't worry—it's not as scary as it sounds. Let a firm's net profit be the function Pi(Q), where Q is the quantity of output produced by the firm, p is the market price, and C(Q) is the firm's total cost function, describing how total costs change with output. The firm's profit can be written as:

Pi(Q) = p × Q − C(Q)

Since the firm operates in a perfectly competitive market (remember, there are no barriers to entry), p does not depend on Q—it's simply a constant. Any firm, in the long run, wants to maximize profit in order to maximize returns to its shareholders. To find the level of output Q that maximizes profit, we take the derivative of Pi(Q) with respect to Q and set it equal to zero:

Pi′(Q) = p − C′(Q) = 0

Which implies:

p = C′(Q)

The term C′(Q) is known as the marginal cost (MC) of producing one additional unit of output. In other words, profit is maximized when the market price equals marginal cost.

But here's the key point: in the long run, a perfectly competitive business earns zero economic profit. Firms selling identical products barely break even. They cover their costs, but they do not generate excess returns.

Of course, in the real world, no two firms or products are truly identical. One company adopts new technology faster and lowers its production costs. Another has access to unique distribution channels. Someone else has built a powerful brand.

What we usually observe in reality is monopolistic competition (not to be confused with a monopoly), where many firms sell products that are highly similar but differentiated in some way. Under monopolistic competition, businesses can earn somewhat higher margins—but only until competitors copy whatever gives them an advantage.

Airlines and grocery stores are excellent examples of monopolistic competition. Take a look at their profit margins. They're typically around 1–3%.

This is why, when investors ask about your barriers to entry or your moat, they're trying to understand how long you'll be able to remain in that differentiated position—and, ideally, whether you'll eventually capture enough market share to become something closer to a monopoly.

No one expects your duct-taped MVP to be impossible to replicate. The real question is what you'll systematically do to stay ahead of your competitors over time.

If barriers to entry are low, investors should really only fund the strongest team—the one with the biggest reputation and the best network—because access to capital itself becomes an unfair competitive advantage. And as barriers to entry in an industry continue to fall, the opportunity to generate outsized profits inevitably disappears.

Twenty years ago, you could launch an online store and make a fortune. Today, many e-commerce businesses also operate on razor-thin margins.

Economic laws differ from the laws of physics in one important respect: you can violate them for a while. But you can't violate them forever. And certainly not over a ten-year horizon—which is roughly how long it takes to build a unicorn.

Incidentally, this is also why I don't believe we'll see a wave of one-person unicorns (perhaps there will be a few exceptions, but it won't become a systematic phenomenon), and why many AI agentic startups are likely to disappear.

Perhaps the most interesting implication is that today's frontier LLMs don't actually have as strong a moat as many people believe. Within the next two or three years, there will be plenty of inexpensive models that deliver perfectly adequate quality for the vast majority of use cases. Companies will deploy these models either in their own cloud infrastructure or locally on their own machines. That will leave LLM providers competing primarily in the niche where customers truly require the most capable and cutting-edge models.

Anyway, that's my perspective. You can decide for yourself whether it's worth listening to.

Subscribe to my LinkedIn here.