Free tools Windows power users keep installed
One-click scans. No signup required.
iTechGuides is reader-supported. When you buy through links on our site, we may earn an affiliate commission. As an Amazon Associate I earn from qualifying purchases. Learn more
“Real Men Own Fabs” was Jerry Sanders III’s famous defense of the integrated-device-manufacturer (IDM) model, in which a serious chip company designed processors and operated its own wafer fabs. The model lost its economic advantage as fabrication plants, process development and upgrades became multi-billion-dollar commitments. AMD separated its manufacturing operation into GlobalFoundries in late 2008 and thereafter relied on commercial foundries, including TSMC. Fabless design paired with specialist foundries became the industry’s prevailing model.
What “Real Men Own Fabs” meant
Jerry Sanders III, AMD’s co-founder and former chief executive, used the phrase as a chest-thumping summary of the IDM philosophy. Owning a fab promised direct control over process technology, production schedules, yields and the connection between engineering and manufacturing.
That approach fit an earlier semiconductor economy. A company that could fund a plant and keep it busy might differentiate its products through manufacturing know-how as well as circuit design. Fab ownership was therefore presented as evidence that a chipmaker was committed to competing at the highest level, not merely designing parts for someone else to build.
The slogan was never a technical rule that every successful chip company had to follow. It was a statement about where competitive advantage was believed to reside in the IDM era.
#1 Best Overall
Why owning fabs became dangerous for AMD
Costs rose faster than most designers could absorb
Modern fabs require enormous upfront investment, continual equipment replacement and expensive process-development teams. Each new manufacturing generation brings another cycle of research, tooling, qualification and ramp-up. A company must pay those costs whether demand is strong or weak, while also keeping the plant sufficiently utilized to spread fixed costs across wafers.
EE Times reported in 2013 that AMD’s decision to keep building and operating its own fabs as costs rose nearly bankrupted the company. The problem was not simply that AMD owned buildings; it carried the financial and operational risk of maintaining a leading-edge manufacturing capability while competing against companies with far larger production volumes.
Manufacturing stopped being a capability every designer needed in-house
Commercial foundries changed the strategic calculation. A foundry can serve many chip designers and distribute plant, equipment and process-development costs across their combined demand. Designers can concentrate capital and staff on architecture, verification, software support and product launches rather than duplicating a full manufacturing organization.
Do these 3 things before closing this tab:
1Clear out junk files and repair common Windows errors2Fix the driver behind crashes, sound loss and screen glitches3Repair Windows errors before they cause bigger problemsAMD executive W.J. Rhines described the shift in a 2002 Los Angeles Times interview: “Well, what we now see is a different scenario. AMD will have to partner with people who view manufacturing as their core competency.” That statement captured a move away from Sanders’s earlier assumption that ownership itself was central to competitiveness.
When AMD spun out its fabs
In late 2008, AMD divested its fab network into a separate company that became GlobalFoundries. This was a structural separation of manufacturing from AMD’s chip-design business, not a rebranding of the old IDM model.
After the separation, AMD became dependent on external foundries for processors, chipsets, graphics processors and other products. EDN identifies TSMC as AMD’s primary foundry partner for leading-edge devices during the period it describes. AMD could still specify products, collaborate on process requirements and manage supply contracts, but it no longer had its own in-house wafer-fabrication network.
How the fabless model took over
The commercial-foundry model gained momentum in the mid-1990s. TSMC and other foundries demonstrated that manufacturing could be a standalone business serving many design companies. That allowed “fabless” firms to compete without owning a fab, provided they could secure capacity and deliver designs that customers wanted.
CEVA’s 2024 explainer describes fabless production as the standard, or de facto, approach. Outsourcing specialized manufacturing often makes better economic sense than performing every step internally, especially when process technology is advancing rapidly and each generation demands a fresh investment.
The change also reflected specialization. Foundries focused on yields, process integration, equipment utilization and manufacturing scale. Fabless companies focused on architecture, chip design, product differentiation and market timing. The result was a broader ecosystem in which one foundry’s capacity could support many competing designs.
Independent reader supportYour contribution helps us test, update, and keep practical guides available for everyone.IDM versus fabless: what changed?
| Consideration | IDM with owned fabs | Fabless design with commercial foundries |
|---|---|---|
| Capital intensity | The chip company funds plants, equipment, process development and upgrades. | The designer avoids owning fabs and pays a foundry for manufacturing capacity and services. |
| Control | Direct control of manufacturing decisions, process integration and scheduling. | Less direct control; process choices and execution depend on the foundry relationship. |
| Time to market | New products may depend on the company’s own process-development schedule and ramp. | Designers can access an established commercial node, subject to foundry availability and qualification. |
| Scale and utilization | One company must keep its plants productive enough to justify fixed costs. | A foundry spreads capacity and process costs across many customers. |
| Strategic exposure | High fixed-asset and under-utilization risk, but fewer external manufacturing dependencies. | Lower fixed-asset risk, but exposure to capacity allocation, supply constraints and partner decisions. |
Did owning fabs become irrelevant?
No. The slogan stopped describing the prevailing business model, but manufacturing ownership can still matter where process control, supply assurance, specialized technologies or national-security considerations justify the expense. Some companies continue to operate fabs, and foundries themselves must own and run them.
What changed is the default assumption for a chip designer. A successful company no longer has to own a fab to be a serious semiconductor competitor. It must instead decide whether the benefits of direct manufacturing control outweigh the cost, complexity and risk of maintaining that capability.
The lasting lesson from AMD’s reversal
AMD’s path from proud IDM to foundry customer illustrates a broader industry economics lesson: vertical integration is advantageous only when its control benefits exceed its fixed costs and operational burden. As fabs became more expensive and technically specialized, commercial foundries offered a way to buy manufacturing capability as a service.
Bolaji Ojo summarized the new reality in an EE Times analysis in 2013: “Fabless is the norm today; real men don’t think about fabs.” The line deliberately echoes Sanders, but the underlying point is practical rather than ideological. In the modern semiconductor industry, design excellence and access to manufacturing are often separate capabilities, supplied by different companies.
Quick Recap
Product prices and availability are accurate as of the date/time indicated and are subject to change. Any price and availability information displayed on Amazon at the time of purchase will apply.
