A metal foundry is not one standardized business. The economics change sharply depending on alloy, process, casting size, batch length, quality requirements, and how much finishing is done in-house. The official U.S. classification under NAICS industry group 3315 covers establishments that pour molten metal into molds or dies, including iron, steel, investment-casting, die-casting, and nonferrous operations. A founder must narrow that broad category before estimating equipment, staffing, permits, or sales.
The practical divide is between a jobbing foundry that handles many customer parts in smaller batches and a production foundry built around repeat programs. Jobbing work can command a higher price per pound because engineering, setup, pattern changes, and inspection are spread across fewer units. Repeat production can offer steadier furnace loading and lower unit cost, but it usually requires more tooling, automation, customer approvals, and working capital.
Green-sand casting
No-bake molding
Investment casting
Permanent mold
Die casting
Ferrous alloys
Nonferrous alloys
Jobbing work
Initial planning snapshot
The project is capital-intensive, slow to commission, and highly sensitive to quality performance.
These figures are planning assumptions, not published industry averages. The American Foundry Society describes metalcasting as a large U.S. manufacturing sector serving infrastructure, transportation, defense, medical, and other markets. That breadth is a useful reminder that customer requirements vary as much as the castings themselves.
A useful financial model starts with physical drivers and ends with cash. It should not begin with a guessed annual revenue number. Start with available furnace hours, melt rate, planned utilization, metal yield, internal scrap, shipped weight, price per pound or piece, tooling revenue, and secondary-operation revenue. Then calculate metal, consumables, direct labor, energy, freight, and outside processing.
Financial model flow
Physical production assumptions should roll through profit, cash flow, owner earnings, and payback.
The balance sheet matters as much as the income statement. Metal inventory rises before shipments. Work in process sits in molding, heat treatment, finishing, machining, or inspection. Customers may pay 45-75 days after invoice, while payroll and utilities are due sooner. A growing foundry can therefore require more cash even while EBITDA improves.
Startup assets also affect depreciation and taxes. The IRS explains MACRS treatment and conventions in Publication 946. Tax depreciation can improve after-tax cash flow, but it does not create operating cash if the plant is underutilized. The model should show book depreciation, tax depreciation assumptions, debt principal, interest, maintenance capex, and replacement reserves separately.
Payback should use cash available after the plant is stable, not peak EBITDA from a fully utilized spreadsheet. The numerator must include equipment, build-out, permitting, launch tooling, contingency, and working capital that will remain tied up. The denominator should be free cash after taxes, debt service, maintenance capex, and the cash needed to support growth.
Payback sensitivity
The difference between an attractive and weak project is usually utilization, yield, and cash tied up during ramp.
A payback under four years is possible for a well-positioned acquisition, a captive supplier with committed volume, or a specialty foundry with scarce capability. It should not be the default assumption for a greenfield operation. The risk-adjusted base case should include commissioning delays, customer approvals, learning-curve scrap, furnace or dust-collector downtime, insurance deductibles, and at least one major maintenance event.
The investment case becomes stronger when the foundry has contracted or highly visible demand, customer-funded tooling, transferable environmental permits, redundant critical equipment, a trained management team, and pricing mechanisms for metal and energy inflation. It becomes weaker when one customer controls more than 35%-40% of revenue, the process depends on one furnace, the site has unresolved environmental history, or projected profit depends on scrap and labor performance that the team has never achieved.