The Marginal Cost of the Next Product Should Fall

In a portfolio built right, the marginal cost of the next product falls toward the cost of just the part that is genuinely new. Everything else, the auth, the billing, the deploy, the governance, is already built and paid for. That is the economic engine under running many companies at once, and it is worth stating precisely, because it explains both why the strategy works and where it breaks.

The fixed cost you pay once

Every product needs the same plumbing: a way to log users in, charge them, deploy the code, log what happened, and enforce the rules. Build that once into a shared foundation and it becomes a fixed cost. The first product pays for all of it, on top of building its own actual product. That first venture is expensive, and it should be, because it is funding the foundation every later venture will inherit.

This is the part people skip when they hear the pitch. There is no discount on product one. The compounding only starts once the fixed cost is sunk, which is exactly why a single-product plan cannot justify a foundation. You need enough products for the fixed cost to amortize, which is why I build twenty companies rather than one.

The marginal cost that keeps falling

Once the foundation exists, each new product only pays the marginal cost: the differentiated logic that makes it its own thing. The plumbing is inherited. The governance is inherited. The deploy pipeline is inherited. What is left is the part a customer actually pays for, and that part is the only part you build.

So the marginal cost of product two is lower than product one. Product three is lower still, because by then the foundation has been battle-tested and the sharp edges are gone. The curve bends down toward the irreducible cost of the new idea itself. That is the whole thesis behind each next venture being cheaper to ship, stated as economics instead of a slogan.

Why the curve bends and does not just step down

It is not a flat discount. The foundation improves as more products use it, because every product stresses it in a new way and the fixes flow back to the shared core. Product four benefits from every bug products one through three already hit. The reliability compounds, not just the reuse.

The build loop compounds the same way. I turn specs into shippable increments through Bootspring, and each product refines the patterns the next one starts from. The foundation is not a frozen asset, it is a living one that gets more valuable per product, which is why the marginal cost curve keeps bending instead of leveling off after the second one.

Where the economics break

Be honest about the failure case. The marginal cost only falls if the shared part is actually shared. The moment products fork the foundation, drift onto different versions, or leak their own logic into the core, the reuse discount evaporates and you are paying near-full price per product plus the overhead of pretending to have a foundation. That is the worst outcome, and it is common.

It also breaks on volume. If you do not ship enough products, the fixed cost never amortizes, and the marginal-cost story is irrelevant because you never reach the margin. A foundation is a bet that you will ship repeatedly. If you will not, the arithmetic says copy per project and skip the foundation entirely, which I covered in when to build shared infrastructure.

What the falling curve buys you

When the marginal cost of the next product is low, you can start companies that would not survive their own full build cost. A niche product that could never justify building auth, billing, and compliance from scratch becomes viable when it inherits all of that for free. The falling marginal cost does not just make existing products cheaper, it changes which products are worth doing at all.

That is the real prize. The foundation is not a cost-savings program, it is a widening of what is possible. Get the marginal cost of the next product low enough and the portfolio can chase ideas a single-product company would have to walk past. Build the foundation once, then let the curve do the work.