Semiconductor Fab Capex Planning: The Real Numbers Nobody Tells You Upfront
Every semiconductor fab build runs over budget — because support infrastructure, consumables, and qualification costs never make it into the first spreadsheet. Here's what to actually plan for.
This guide is for: the person who just built the equipment budget for a new semiconductor fab and is feeling good about it — because you're probably missing 30–50% of what this will actually cost.
I don't say that to alarm you. I say it because I've watched it happen enough times that it's almost a ritual. Someone builds a careful spreadsheet of all the process tools. They get quotes. They find used equipment. The number looks achievable. Then 18 months later they've spent twice what they planned and are still 6 months from first qualified process.
The gap is never the tools. It's everything around the tools.
The Tool Budget vs the Real Budget
Here's the rule: take your tool budget and add 40–60% for support infrastructure, facilities prep, installation, and qualification. That's not a cushion. That's the plan.
A $10M tool budget should have a $14M–$16M total project budget. A $30M tool budget needs $42M–$48M. If your project approvals don't include this, go back to the CFO before you start buying, not after.
Facilities: The Budget That Always Grows
DI water systems for a 200mm fab: $200K–$600K for the hardware. If your building needs a new utility connection, add $50K–$200K for civil work. Budget a DI water contingency — every fab I've seen has had scope creep here.
Specialty gas distribution (process gases, purge gases, CDA): $150K–$500K for a basic system. If you're adding flammables (SiH4, H2, B2H6), the safety systems add another $100K–$300K.
HVAC/filtration for a Class 100 (ISO 5) cleanroom: $300K–$1.5M depending on size and how tight your contamination spec is. This number varies so widely that it deserves its own line item with a site survey before you commit.
Electrical capacity: most tools arrive with undocumented power requirements that are slightly different from the datasheet. Budget $50K–$150K for electrical work beyond your first estimate.
Installation: The Cost Nobody Quotes Until the Tool Is on the Floor
Tool installation for complex 200mm equipment (cluster tools, furnace systems, CVD systems) runs $10K–$50K per tool, depending on complexity and whether the OEM or a third party does it. A 4-tool etch module purchase becomes $400K in tools plus $80K–$150K in installation.
Chase the installation quotes before the tools arrive, not after. Scheduling installation contractors in a new facility with no previous relationships takes 6–10 weeks minimum.
Qualification: The Budget Nobody Puts In
Process qualification — the work that turns "installed tool" into "qualified process" — is not free labor. It requires engineering time, consumables, test wafers, and often outside expertise.
A basic etch chamber qualification (particle counts, uniformity, etch rate characterization) takes 2–6 weeks of engineer time and $5K–$20K in wafers and consumables. For a full 10-tool installation, budget $80K–$200K in qualification costs that don't appear in any equipment purchase order.
If you're moving an existing process from another fab, add 30% to the qualification budget. Process transfers are harder than process development.
Consumables: Year 1 Is Not Representative
First-year consumables on a new fab are 2–3x higher than steady-state. Why: every tool needs initial process consumables (quartz ware, chamber kits, filters), every process needs characterization runs that consume materials, and every failure mode gets discovered the first time you run the process.
Build a Year 1 consumables budget that's separate from the capital plan. For a 10-tool fab:
- Quartz ware: $30K–$80K (furnaces, etch, CVD)
- Chamber kits and O-rings: $20K–$50K
- Process gases (first fills, setup): $15K–$40K
- Wafers for qualification: $20K–$60K
This $85K–$230K shows up in your operations budget, not your capital budget. Which means nobody tracks it until the first P&L review and then someone asks why operations is over budget when the fab isn't even running production yet.
Used Equipment vs New: The Real Trade-off
Used equipment saves 40–70% on sticker price. It adds 10–20% to installation and qualification costs because you're starting with unknown history rather than a new-tool baseline.
The net is still strongly favorable for used tools on most process equipment. The exception is anything with hidden failure modes — turbopumps, RF generators, precision metrology stages. On these, the cost of a failure during qualification often exceeds the purchase savings. Budget a $20K–$40K contingency specifically for used tool surprises.
What a Realistic 200mm Research Fab Budget Looks Like
| Category | Low End | High End | |---|---|---| | Process tools (used, 10–15 tools) | $3M | $8M | | Metrology and inspection | $400K | $1.2M | | Facilities and infrastructure | $1.5M | $4M | | Installation | $200K | $600K | | Qualification and startup | $150K | $400K | | Year 1 consumables | $100K | $300K | | Contingency (15%) | $800K | $2.1M | | Total | $6.2M | $16.6M |
Teams that plan at the low end and execute toward the high end are the ones that run out of money 60% through.
FAQ
How much should I budget for semiconductor fab facilities vs equipment? Rule of thumb: facilities (DI water, gas, electrical, HVAC) typically run 30–50% of the equipment budget for a new greenfield build. If someone tells you facilities will be 15%, they're either in a pre-built cleanroom or they haven't priced the utilities.
What's usually missing from first-pass fab budgets? Installation labor, qualification wafers and engineering time, first-year consumables (especially quartz and chamber kits), and IT/data infrastructure. Also: the cost of delays, which add carrying costs on purchased equipment while installation stalls.
How do I build a contingency into a fab budget? 15–20% is standard for a new build with significant uncertainty. 10% is reasonable for an expansion into an existing facility. Never present a fab budget without a labeled contingency line — it's not pessimism, it's engineering.
Is it cheaper to buy a decommissioned turnkey fab? Sometimes. The tool cost can be 50–70% below market. But you inherit the previous fab's utility configuration, layout, and contamination history. Qualification costs often equal the savings. Do a detailed due diligence before assuming a turnkey fab is cheaper than building your own.
When should I start securing equipment vs finalizing facility design? Simultaneously, with tool delivery timed to facility readiness. The failure mode is buying tools 12 months early and storing them — which costs $500–$2,000/month per tool in storage fees and accelerates dry-out and contamination risk.
What's the biggest mistake in fab capex planning? Treating the tool budget as the project budget. The moment you separate "tools" from "everything else it takes to make the tools work," you have an honest plan.