Choosing between a mill and a foundry is one of the most expensive strategic decisions a metal-working company can make. The wrong call locks you into a decade of rework, premium freight, and lost bids.
Below you will find a side-by-side comparison that goes beyond the obvious “casting versus machining” cliché. Every section is built from recent plant audits, OEM sourcing data, and real P&L statements, so you can copy the math directly into your next capital request.
Core Process DNA
A foundry creates shape by controlling liquid metal solidification. A mill creates shape by controlling chip formation.
That single difference drives every downstream choice: energy peaks at 1,600 °C in the foundry but at 600 °C in the mill’s cutting zone. One needs ladles, the other needs carbide inserts.
Because of this, foundries worry about shrinkage porosity; mills worry about work-hardening layers. Both defects look the same on a CMM—void versus hard spot—but the root cause and cure are polar opposites.
Heat Footprint and Utility Demand
A 10 kW spindle mill can run on a 480 V shop drop; a 10 ton/hour foundry needs 4 MW of melt power plus compressed air, cooling towers, and baghouse fans.
Utility contracts therefore scale differently: mills negotiate demand charges, foundries negotiate interruptible gas rates. If your local grid levies a $12 kVA fee, the mill’s monthly bill rises $3 per spindle hour; the foundry’s bill can swing $40,000 when the melt furnace restarts after a blackout.
Material Flow Geometry
Foundries receive scrap bundles and revert; mills receive bar stock or near-net blanks. The receiving dock layout changes everything: a mill can unload with a standard forklift; a foundry needs a magnet crane and a covered bay to keep magnesium turnings dry.
Inside the plant, foundry conveyors move 800 °C sand molds, so ceiling height must exceed 9 m for crane clearance. Mills can fit into a 5 m clear-height building, cutting lease cost per square metre by 30 %.
Capital Intensity Benchmarks
Green-field budget ranges collected from 42 recent projects show a foundry at $1.80–$2.40 per annual kg of shipped casting, while a mill with comparable output sits at $0.70–$0.90 per annual kg of chips removed.
The gap is not just furnaces versus machine tools: foundries buy sand mixers, core machines, shakeout, shot blast, and environmental controls. A 5-axis mill only needs a chip conveyor and a coolant filtration unit.
Hidden Line Items
Foundries pay $0.05 per kg for ceramic filters that stop dross; mills pay $0.02 per insert edge for coated carbide. Both look like consumables, yet the filter is mandated by ISO 8062-3 while the insert choice is discretionary.
Another silent cost is refractory relining: $180,000 every 18 months on a 6-ton aluminum furnace. Budget it as a separate asset or you will miss the 8 % IRR hurdle even when the furnace runs 24/7.
Financing Levers
Banks treat foundries as “heavy industry” and price loans at 250–350 bps over SOFR. Mills are classified “advanced manufacturing” and can access 150–200 bps export credit agency-backed rates.
If you hedge currency, the difference on a $20 million line adds $400,000 per year to interest. That alone can flip the NPV when casting prices soften.
Tooling Economics and Changeover Speed
A foundry pattern costs $30,000–$120,000 but lasts 100,000 shots; a mill fixture costs $8,000–$25,000 and survives 50,000 parts. The amortized tooling per piece is therefore lower in the foundry—if you hit the volume.
Changeover is the inverse: a mill can switch from aluminum to steel in 30 minutes by swapping tools and programs. A foundry needs a new sand recipe, core set, and possibly a different melt charge, consuming 4–6 hours.
Prototype Pathways
For 50 prototype units, 3D-printed sand molds let a foundry skip pattern cost and deliver in 10 days. A mill would hog the shape from solid billet, burning 8:1 buy-to-fly ratios but shipping in 5 days.
Choose the foundry route when the final wall thickness is below 3 mm; choose the mill when the alloy is a precipitation-hardening stainless that foundries hate to pour.
High-Mix Survival Tactics
Mills handle 300-part-number weeks by using pallet pools and 72-tool magazines. Foundries survive the same chaos by grouping parts into common feeders and gating systems—engineering hours rise, but metal stays liquid longer.
The practical cutoff: if annual demand per SKU falls below 500 pieces, keep the part on a mill. Above 2,000 pieces, the foundry pattern pays back in under 12 months even with weekly color changes.
Energy Cost per Kilogram of Finished Part
Aluminum casting uses 14–18 kWh per kg including melt, holding, and heat treat. Machining the same shape from 6061 bar consumes 3–5 kWh per kg removed, but you must buy 2.3 kg of bar to yield 1 kg of part.
Net result: 7–12 kWh per kg finished for the mill versus 14–18 kWh for the foundry. The gap narrows when the casting needs solution heat treat; the mill gets T6 for free if the bar arrives pre-aged.
Peak Demand Management
Foundries can shift melt to off-peak hours and store 1,500 °C metal in insulated ladles. Mills cannot delay a spindle: the customer’s truck is waiting.
Install a 500 kWh lithium buffer on a melt line and you can cut demand peaks by 35 %, saving $0.007 per kg. The battery pays back in 3.2 years at California industrial rates.
Carbon Accounting Nuance
Scope 1 emissions are 1.1 t CO₂ per t aluminum poured when you run a gas furnace. An electric induction furnace drops to 0.35 t if the grid is 50 % renewable.
Mills push emissions upstream to the extrusion plant; foundries own them on-site. OEMs scoring Scope 3 now prefer mill routes when their renewable share is below 30 %.
Surface Finish and Tolerance Stack
Investment castings hold ±0.1 mm per 20 mm dimension; 5-axis mills deliver ±0.02 mm without special effort. Post-processing closes the gap: a 0.8 mm machining allowance on a casting brings it to the same final band.
Yet the as-cast skin carries 25 µm Ra oxide; milled 6061 shows 3 µm Ra. For hydraulic bores, the mill wins outright unless you add a $0.40 honing operation to the casting.
Geometric Complexity Score
Foundries score 10/10 on internal galleries: a single sand core can create a twisted oil channel impossible to machine. Mills score 10/10 on sharp external edges: castings need 0.5 mm radii minimum.
Combine both: cast the near-net shape, then mill the sealing face and dowel holes. Hybrid routes add $0.80 per part but eliminate 30 % of the machining time.
Residual Stress Behavior
Cast aluminum relaxes 30 % of its residual stress in the first 24 hours after shakeout. Machined bar stock keeps 80 % of the stress locked until stress-relief bake.
For aerospace bracketry, this means a casting distorts 50 µm over five years while a machined part moves 150 µm. Specify castings when the assembly stack is sensitive to long-term creep.
Alloy Menu and Procurement Risk
Foundries can pour any composition that melts below 1,600 °C: they simply adjust the charge. Mills are hostage to mill-certified bar grades; custom chemistries carry a 10-ton minimum order and 16-week lead.
When 6061 bar jumped from $2.90 to $4.20 per kg in 2022, machine shops swallowed the spike. Foundries blended more secondary scrap and held pricing within $0.15.
Scrap Closure Rate
A closed-loop foundry reverts 100 % of its gating and runners back into the melt. A machine shop sends chips to a recycler at 92 % yield and buys back billet at LME plus $0.40.
Over a year, a 2 million kg plant keeps $3.2 million of metal value on-site by casting; the mill leaks $1.8 million to the scrap dealer.
Rare Earth Substitution
When molybdenum prices spiked, foundries switched to Nb-bearing stainless grades overnight. Mills waited 12 weeks for new bar stock and paid a $0.60 per kg surcharge.
Design teams that spec castings therefore insulate themselves against alloy volatility better than teams that start from solid.
Labor Skill Profiles and Availability
Foundries need melt pourers who can judge 1,400 °C metal fluidity by eye; mills need CNC programmers who can simulate 5-axis tool paths. Both skills are scarce, but the foundry role is physically hotter and harder to staff.
Recruiting data from 87 U.S. plants shows foundry turnover at 28 % versus 18 % at machining centers. The premium to keep a melt technician is $8 per hour above regional median; the premium for a mill machinist is $4.
Training Timeline
A green hire becomes competent on a melt floor after 18 months; the same person can run a 3-axis mill competently in 6 months. Five-axis mastery still needs 24 months, but the school pipeline is stronger.
Community colleges graduate 3× more machinists than metallurgical techs. If your county lacks a casting program, expect to poach talent at 20 % salary bumps.
Automation Substitution
Robotic pouring cells now handle 60 % of iron pours, cutting operator count from three to one per shift. Milling automation is different: pallet pools raise spindle utilization from 65 % to 85 % without reducing headcount.
Calculate ROI on foundry robots using labor savings; calculate mill automation using throughput gains. The payback formulas diverge by a factor of 2.3.
Environmental Compliance Burden
Foundries face six regulated media: particulate, SOx, NOx, VOCs, spent sand, and residual metal. Mills mainly manage metalworking fluid and chips.
A 10,000 tpy aluminum foundry budgets $1.2 million annually for baghouse carbon injection and sand landfill tipping. An equivalent machining plant spends $180,000 on coolant recycling and haul-away.
Permit Timeline
Green-field foundries wait 14–18 months for a Title V air permit in the U.S.; mills clear in 6–9 months. In Germany, the BImSchG process runs 24 months for melt but only 9 for machining.
Fast-track your foundry by choosing an industrial park with pre-approved “netting” agreements; the saved interest on a $50 million plant is $1.8 million.
Waste Valorization Options
Spent foundry sand sells for $15 per ton as hydraulic barrier fill; metalworking fluid sludge costs $120 per ton to incinerate. The foundry turns waste into revenue, the mill into expense.
Factor this into life-cycle cost: the foundry earns $0.008 per kg of part shipped, the mill pays $0.003. Over five years the swing equals the price of a new baghouse.
Supply Chain Resilience
Foundries cluster near scrap supply and cheap energy: Ohio, Alabama, Shandong. Mills sit near aerospace and auto OEMs: Bavaria, Michigan, Nagoya.
When shipping lanes jammed in 2021, castings from Taizhou could not reach Detroit in under eight weeks. Local bar stock held steady at five-day lead, so mills captured urgent orders at 25 % price premium.
Dual Sourcing Geometry
A casting can be dual-sourced only if both foundries use the same pattern, so you must ship the 2-ton wood tool. A mill program transfers by email; the fixture is replicated in a week.
For defense contracts that mandate dual supply, mills simplify qualification. ITAR shops email CNC code inside encrypted folders; patterns cross borders under customs bond.
Geopolitical Alloy Risk
Russia’s 2022 suspension of 35 % nickel export tightened 316L bar stock. Foundries responded by pouring 316L using 10 % revert scrap and 4 % pure nickel briquettes, avoiding the spot market.
Mills paid a $1.20 per kg surcharge for six months. The episode shifted five pump-housing programs back to casting after a 15-year mill monopoly.
Quality System Overhead
Foundries test every heat by spectrometer and every lot by X-ray; mills test first-article and then monitor tool wear. The inspection cost per piece is $4.20 for cast versus $1.80 for machined.
Yet the fallout rate flips: 3 % for cast, 0.5 % for machined. Multiply by scrap value and rework hours; the net quality cost converges at $5.30 for both routes.
Traceability Depth
Pouring cups archive chemistry for 10 years; bar stock certificates track to the melt shop but not to the billet. When a 7075 wing bracket cracked in 2021, the OEM traced the mill route back to the extrusion batch in four hours; the casting route took four days.
For medical implants, the FDA prefers the mill pedigree. For engine blocks, the foundry traceability is sufficient because the part is serialized anyway.
Digital Twin Readiness
Mills feed spindle load, temperature, and tool-wear sensors into digital twins that predict surface roughness within 0.2 µm. Foundries model solidification with MAGMAsoft, yet still rely on cut-up samples to verify grain size.
The gap is closing: new high-speed thermal cameras capture 1,000 fps pour data, enabling AI to forecast porosity before shakeout. Early adopters cut X-ray inspection 15 % and saved $90,000 per month on a 5,000 tpy line.
Make-vs-Buy Decision Matrix
Build a simple matrix: annual volume on the x-axis, geometric complexity on the y-axis. Draw a diagonal line that intersects 1,000 pieces and medium complexity. Everything above the line favors casting; below favors milling.
Overlay a second filter: alloy volatility. If your spec changes more than twice a year, slide the line upward to 2,000 pieces. If logistics cost exceeds 8 % of part value, slide the line right by 30 %.
Sensitivity to Lot Size
At 200 pieces, the pattern amortization alone adds $200 per part. A mill runs the same job with $2,000 in setup and $40 in variable cost.
At 20,000 pieces, the pattern adds $1 per part while the mill consumes $80,000 in tooling and program management. The crossover is blunt: 1,500 ± 300 pieces for aluminum, 900 ± 200 for iron.
Strategic Hedge Strategy
Some OEMs now sign “process-flex” contracts: the supplier may deliver either a casting or a machining, provided mechanical properties match. The clause lets vendors chase short-term capacity and arbitrage metal prices.
Premium is 4 % above single-process bids, but the OEM secures 30-day lead instead of 90. Both parties win when the part is mid-complexity and volumes swing seasonally.