Core competency

In the early days of SpaceX, a machine shop that supplied critical parts was about to shut down. For most companies, that's a procurement headache: start the search for a new vendor, sign new contracts, lose a few months. Elon Musk's response was different. He bought the shop. He moved the machines into SpaceX, hired the machinists, and had them working that same Saturday evening, sitting them right next to the engineers who designed the parts. Bringing it in-house cut much of his manufacturing costs in half and erased the delay of waiting on someone else to deliver.

Musk later explained the logic: "To the degree that you rely on the legacy supply chain, you inherit their legacy constraints, including their speed, cost, and technology."

The reigning advice is to stay light. Find your core competency, the one thing you do better than anyone, and outsource everything else. Don't own factories if you can rent them. Don't build what you can buy. Asset-light is agile; vertical integration is a relic, expensive and slow.

And for most companies most of the time, this is genuinely right. You should not build your own data center; you should rent it from Amazon. Comparative advantage is real. Focus is real. As a default, "stick to your knitting" is sound.

The error is treating it as a law instead of a default, and missing the exception where it inverts.

When a capability is genuinely critical to your product, outsourcing it doesn't make you nimble. It chains you to the slowest, most expensive, least ambitious link in your chain.

If your supplier takes thirty days to deliver, thirty days is now your speed, no matter how fast you can think. If their costs are high, your margins are capped at whatever they leave you. If their technology stops improving, your product stops improving with it. You don't just buy a part from a vendor; you adopt their entire set of limitations as the ceiling on your own ambition. The dependency that looked like efficiency on the spreadsheet is a vulnerability: a place where someone else's priorities, deadlines, and failures get to overrule yours.

That's why, for the part of the business that actually determines whether you win, the great builders refuse to outsource. They'd rather own the constraint than inherit it.

Rockefeller's barrels

This isn't a Silicon Valley insight. John D. Rockefeller was running the same playbook in the 1860s with wooden barrels.

Refining oil meant buying barrels, and barrels were a constant problem: shortages, and a price of about $2.50 each from outside makers. Rockefeller's competitors treated this as a fixed cost of doing business and kept buying. Rockefeller brought it in-house. He bought timber, dried and built the barrels himself, and slashed the cost so far that, as he liked to point out, his competitors were paying more than double what he was for the exact same container. He did the same with everything he touched: when a plumber's bill looked padded, he put a plumber on staff by the month and kept the work inside the company.

The integration bought more than a lower cost. It bought him independence from everyone else's shortages and prices. While his rivals were at the mercy of the barrel market, Rockefeller had removed the barrel market from his life entirely. The capability that everyone else rented, he owned. And owning it became a permanent advantage they couldn't copy without rebuilding their whole operation.

From factory to face

The most complete version of this idea belongs to a man who started with almost nothing: Leonardo Del Vecchio, the founder of Luxottica. His biographer describes his entire life as "a relentless, methodical march for deeper and deeper vertical integration."

Del Vecchio began as a subcontractor, making metal parts and molds for other people's eyeglasses, doing the hard, low-margin work while the firms above him captured the value. So he climbed the chain. He started making the whole frame instead of the parts. Then he integrated forward into distribution. And then, to control how his glasses reached customers, Luxottica launched a $1.4 billion hostile takeover of US Shoe, a conglomerate Del Vecchio didn't want, except that it owned LensCrafters, the largest optical retail chain in the country. He bought the whole company, broke it up, kept the stores, and now controlled eyewear from the factory floor to the customer's face.

By owning every link (design, manufacturing, distribution, retail), Del Vecchio answered to no supplier and no middleman. He set the prices, kept the margins others would have taken, and could put his own products on the shelves while keeping competitors off them. The kid who once did everyone else's hard work for a sliver of the value ended up owning the entire chain that value flowed through.

When not to own it

Vertical integration is expensive, slow to build, and easy to overdo. Own the wrong things and you don't get Luxottica; you get a sprawling, distracted company that's mediocre at twenty businesses instead of excellent at one. The discipline is in the word critical. You integrate the parts of the chain that are strategic: the ones that determine your product's quality, your speed, your cost, your independence. You keep renting the parts that are genuine commodities, where a specialist will always beat you and the dependency costs you nothing that matters. Musk brought the machine shop in-house; he did not start mining his own iron ore. The question is never "own or outsource?" in the abstract. It's "is this capability on the critical path to the thing I'm trying to build, or isn't it?"

Get that judgment right and integration is a moat. Get it wrong and it's an anchor.

Own the critical path

"Own everything" is bloat with a noble name. Look hard at the few capabilities that genuinely determine whether your product is great, and refuse to leave those in someone else's hands.

For everything truly critical, treat dependency as a vulnerability to be removed. When a supplier's speed becomes your ceiling, build the capability yourself. When a middleman is taking the margin that should fund your next product, climb over him. When the legacy chain's constraints start defining your future, do what Musk did on that Saturday: go buy the machines. The companies that reshape industries aren't the lightest ones. They're the ones that decided which parts of the world they refused to rent, and then went and owned them.