Greenfield Fab
Greenfield and brownfield are the two basic site strategies for building semiconductor manufacturing capacity: a greenfield fab is built from undeveloped land with no existing infrastructure, while a brownfield fab adds or retrofits capacity at a site that already has power, water, permits, and often a workforce in place.
The core trade is time versus control. A greenfield site gives a company a clean layout optimized for the newest tool generation and room to add modules for decades, but it means securing land, water rights, and a power allocation from zero and typically runs 4 to 5 years from groundbreaking to first wafers; a brownfield expansion inherits an existing cleanroom shell, substation, and often a trained workforce, which can pull that timeline down to roughly 2 to 3 years at the cost of working around an existing building's layout and load limits.
Gigafab-scale campuses are usually greenfield first, brownfield after. TSMC's Fab 18 in Tainan and Samsung's Pyeongtaek site both started as greenfield builds and then grew into multi-module gigafabs through a series of brownfield-style additions on the same land — the expensive, slow greenfield step happens once, and every module added afterward behaves like a brownfield project reusing that site's substation and water plant.
| Factor | Greenfield fab | Brownfield fab |
|---|---|---|
| Typical time to first wafers | 4 to 5 years | 2 to 3 years |
| Power, water, gas infrastructure | built from zero | often already in place |
| Layout and tool floor plan | fully custom | constrained by existing structure |
| Permitting and zoning | new environmental review | often an amendment to an existing permit |
| Typical use case | gigafab-scale new campuses | adding a module or new node to a running site |
Mature-node and specialty fabs lean brownfield far more often than leading-edge sites. GlobalFoundries, UMC, and Texas Instruments have expanded existing 200mm and 300mm sites repeatedly rather than building new campuses, because mature-node customers care more about cost and lead time than the newest building, and a brownfield expansion amortizes an already-depreciated site's fixed infrastructure the same way a second gigafab module does.
The choice also shows up in incentive negotiations. Greenfield projects attract larger headline government subsidies, such as CHIPS Act grants tied to TSMC Arizona and Samsung Taylor, because they create new jobs and new regional infrastructure a state or country can point to; brownfield expansions are cheaper to subsidize per job created but make a weaker political announcement, which is part of why greenfield megaprojects dominate the press even though brownfield expansions deliver far more of the industry's actual new capacity.
Read a fab announcement through a *site-history* lens rather than a *size* lens: a brownfield expansion reaching full production in two years is not a smaller achievement than a greenfield megaproject taking five, it is simply reusing capital another project already sank into power, water, and permits — the same shared-infrastructure logic that makes a second gigafab module cheaper than the first, applied one level up to the choice of site itself.