Make or buy: deciding which parts of your business to own and which to buy

The plant Boeing sold, and bought back

In 2005 Boeing sold its fuselage plant in Wichita, Kansas, for about $900 million in cash. The idea was to stop building the big pieces of its airplanes and become a lean systems integrator: design the plane, buy the parts, put them together.

The division had cost about $2.2 billion a year to run, and selling it turned that cost into a supplier's invoice. Nineteen years later Boeing agreed to buy the business back. Explaining why, it said that bringing the work home would

“strengthen aviation safety, improve quality and serve the interests of our customers, employees, and shareholders”Boeing, on reacquiring Spirit AeroSystems, as recorded in the Boeing case

The price, agreed in July 2024, was a total enterprise value of about $8.3 billion, and the deal brought about 15,000 employees back inside. The retreat had started much earlier: in March 2008, Boeing moved to buy out a partner's stake in a 787 assembly plant to regain direct control of its supply chain.

That is the whole decision in one story. Selling the plant looked like a saving because the saving was easy to count. What it cost to depend on someone else for the fuselage showed up later, and it showed up all at once. The rest of this guide is about the companies that faced a version of that choice, which way they went, and which of them had to go back.

What make or buy is

The choice is never all or nothing. Every company buys something, so the real question is which layer to own: the part customers pay for, the part whose failure stops the line, or neither. It runs in both directions. A company can reach back toward its suppliers, as Nucor did when it built its own steel mill, or forward toward its customers, as BYD did when a battery maker bought a carmaker.

Why it is so hard

On paper this is a cost comparison: what it costs to make the thing, against what it costs to buy it. In a real company, three things make that comparison wrong.

The ways companies did it

The decisions on record take five shapes. Some move work inside, some move it out, and one keeps it outside but ties it down.

What 34 companies did

Here is every decision on record, placed by the year it was made and marked by how it turned out. Hover a mark, or open the table, for the company and the outcome.

Of the 32 decisions that can be judged, 22 were never undone, 6 were walked back and 4 were kept only in part.

The walk-backs

Put the reversals and the partial retreats side by side and three groups show. One came to depend on a supplier and found out what that dependence cost. Another brought work home to make a point, and closed the plant when the point stopped paying. The third bought its way into a layer next to its own business, studios or warehouses, and sold or folded it when it did not pay.

The Ericsson row has a twin, though it is filed as a crisis decision rather than a sourcing one. The same March 2000 fire at the same Philips chip plant hit Nokia too. Nokia read Philips's reassuring first report as a reason to check, escalated within days and lined up other chip capacity. Ericsson took the report at face value. Nokia took the market; Ericsson's shortfall was estimated at 7 million phones.

The reshorings that failed each brought home one product or a slice of output: a water heater, part of a lock range, one family of hand tools. The ones that held were built for demand that was already there. Caterpillar's $200 million plant in Victoria, Texas took over the excavators bound for U.S. customers, survived a trade-war layoff in 2019, and was adding 200 positions in 2025.

Owning the work is cheap to give up and expensive to get back.

What the share price did

For many of the decisions there is a clean market record: the company's total return over the three years after the decision, against the S&P 500.

Read it slowly. Boeing's sale of Wichita beat the index by 46.6 points over the next three years, and was undone nineteen years later at a total enterprise value of about $8.3 billion. Intel's 2021 decision to build chips for others still stands; its shares trailed the index by 62.6 points over three years, and its foundry lost $13 billion in 2024. Three years is too short to see what a sourcing decision costs.

How to run this decision

If you are weighing a version of this choice, here is what the record suggests doing, in order.

1. Name the layer customers pay for

   Before you compare costs, write down the one part of the product that makes customers choose you, and the one part whose failure stops the line. Those are the candidates to own or to tie down. Everything else can be bought. Uniqlo decided in 1987 to control everything from design to the price tag, and it still leaves the fiber mills to Toray.

   Seen in: Uniqlo, 1987 · BYD, 2003
2. Don't let a single supplier be why it was cheaper

   If a sole supplier is the cheaper option, say what a second one would cost and treat the difference as insurance you are choosing not to buy. Ericsson made Philips its only source of radio chips to simplify its supply chain, so when the plant burned there was no qualified alternative.

   Seen in: Ericsson, 2000
3. Price the buy-back before you sell

   Assume you are wrong and must bring the work home in fifteen years, at a price the other side sets. Put that price next to the yearly saving. Boeing sold Wichita for about $900 million and bought it back at a total enterprise value of about $8.3 billion.

   Seen in: Boeing, 2005 and 2024
4. Treat a supplier's reassurance as a question

   When a key supplier says a problem will be fixed soon, check the claim and start looking for other capacity the same week. Nokia escalated within days of Philips's first report; Ericsson waited, and its phone business was badly damaged.

   Seen in: Nokia and Ericsson, 2000
5. Prove the process small before you build the flagship

   A new plant that depends on a process nobody has run at scale is two bets, not one. Stanley Black & Decker built its Fort Worth plant around a first-of-its-kind automated process, never fully ramped it, and announced the closure less than three years after the plant opened.

   Seen in: Stanley Black & Decker, 2019
6. Bring work home for demand, not for a headline

   Reshoring one product to show that it can be done leaves a plant with no reason to exist when that product's sales slip. Build where enough demand already sits to keep the plant busy in a bad year, and give it more than one thing to make.

   Seen in: GE Appliances, 2012 · Caterpillar, 2010
7. Try a contract before a deed

   An exclusive partner who builds only for you gives you much of the control of ownership without the capital. Seven-Eleven Japan and Uniqlo both chose that route, and ASML took stakes only in suppliers it could not replace.

   Seen in: 7-Eleven, 1979 · Uniqlo, 2006 · ASML, 2012

To work the numbers for your own case, read Make or buy: how to decide what your company must own.

Warning signs

What it has looked like when this decision was going wrong.

• The only reason given for a sole supplier is that one is simpler than two.Ericsson, 2000
• The case for selling a plant has a line for the saving and none for buying it back.Boeing, 2005
• The new plant exists to prove a point, and the point is the headline.GE Appliances, 2012 · Stanley Black & Decker, 2019
• The investment is a pledge with a big number and nothing binding behind it.Foxconn, 2017
• A supplier's reassuring report is accepted without a check.Ericsson, 2000

All 34 cases

Each case is one company at one decision: what was on the table, what it chose, why, and what followed, with every fact sourced.

Questions

What is a make-or-buy decision?

The choice between doing a piece of work inside your company and paying another company to do it. Choosing to make more of the chain yourself is called vertical integration; choosing to buy more of it is called outsourcing.

What is the difference between vertical integration and outsourcing?

They are the two directions of the same decision. Vertical integration brings work inside, by building a plant or buying a supplier. Outsourcing sends it out, by selling a plant or signing a supply contract. Many companies do both at once, for different parts of what they sell.

What are examples of vertical integration?

Nucor building its own steel mill, BYD buying a carmaker to use its batteries, Apple designing its own chips after buying P.A. Semi, and Amazon building its own delivery network. Boeing's sale and later repurchase of its Wichita fuselage plant is the clearest example in this guide of an outsourcing decision being undone.

Why do reshoring projects fail?

In the cases on record, the reshorings that were undone each brought home one product or a slice of output: a water heater, part of a lock range, a family of hand tools. When that product's sales or process faltered, the plant had nothing else to make.