The cost curvejust moved.

The traditional curve: overhead, not volume

Every production part carries a cost curve behind it, whether or not anyone in the room has ever drawn it: injection molding, die casting, and sand casting all begin astronomically high on the first unit and fall away asymptotically as volume climbs. That shape has governed make-or-buy decisions since the Industrial Revolution, and knowing where it bends is what separates a tool that pays for itself from one that never should have been cut from metal. However, the curve describes overhead far more than it describes volume, and in this industry overhead means one thing: tooling.

The mechanism is arithmetic, and it is worth stating plainly before anyone argues about the conclusion. A die-cast tool can run $25,000 before it makes a single part, and a progressive stamping die climbs past $30,000.¹ At one unit, that unit effectively costs the entire tool. Spread the same overhead across ten thousand units and it dilutes to a few dollars a part, because the material and labor to make one more part were tiny all along. That dilution is the entire mechanism, and it only works if you make a great many of them.

Additive prints a flat line

Additive manufacturing does not behave that way, because it requires essentially no dedicated tooling: the design goes into the slicer, the slicer feeds the machine, and the part builds, with no mold, no cavity, and no fixture cut for the job. With the overhead gone, the curve stops being a curve and becomes a horizontal line. A part that costs $30 to print costs about $30 whether you make one or five hundred, because what you are buying is material, machine time, power, and a sliver of labor, none of which scales with a tool.

The volume at which the falling traditional curve finally drops below that flat line is the break-even point: call it N. Below N, additive is simply cheaper; above it, conventional pulls ahead, and it should. For complex parts at low to mid volumes, N often lands somewhere in the hundreds to low thousands, which is a wider band than most sourcing rules of thumb assume, and it is the band a surprising number of industrial parts already occupy.

One unit or five hundred, the print costs about $30 either way: with no tool left to amortize, the overhead line does not bend.

The economics of a zero-tooling process

Complexity is free

The comparison widens the moment geometry enters it. In conventional manufacturing, complexity such as undercuts, internal channels, and thin ribs drives tooling cost up and to the right, since each of those features has to be cut, cored, or slid out of the mold. Multi-slide molds, EDM work, and extra inserts can push a $25,000 tool past $100,000, and break-even moves out with it, into tens of thousands of units.

Additive carries no such penalty, because the machine lays material exactly where the model says it belongs. A lattice, an organic bracket, or a conformal cooling channel costs no more to build than a plain block of the same envelope, and frequently less, since a solid block is the least efficient thing you can print. As complexity rises, additive holds its economics at higher and higher volumes, and break-even drifts out with it. That relationship is the one worth carrying out of this page: the harder a part is to machine or mold, the better additive tends to look beside it.

Synergy, not substitution

These economics do not make additive a replacement for mass production. Volume remains a legitimate reason to cut steel, and the right-hand end of the curve says so plainly, because a run of 100,000 simple metal parts belongs to die casting or stamping, and it should. The honest framing is not substitution, but rather addition: another process on the list, selected for the parts it suits, running alongside the established ones instead of displacing them.

Additionally, the two cooperate more often than they compete, and the fastest-growing use of additive is making the tooling itself: jigs, fixtures, and even mold and form tooling. Printing that hardware pulls the traditional curve down from the very first unit, because it removes much of the up-front overhead that set the height of the curve to begin with. A near-net printed form tool, finished on a mill only where it matters, can stand in for a large slice of a $30,000 tooling bill. We work that case through in detail for stamping, where the hardened working faces stay metal and much of what surrounds them does not, in 3D-printed components in a stamping die.

Where domestic composite moves the line again

Everything above holds for metal additive, and metal additive already moves the line, with a Markforged Metal X system starting near $99,500 against $400,000 to over $1,000,000 for powder-bed fusion.² However, Key3D’s work is continuous-fiber composite rather than metal, and it moves the line twice more. The two shifts are worth separating, because they act on different parts of the number.

Near-net, not net. Like casting, additive is a near-net-shape process, and for most tooling that distinction never turns into a cost. Across jigs, fixtures, drill guides, gauges, and work-holding, the part printed to spec is the finished part. For the rare face that has to hold single-digit-micron tolerance, you print the near-net shape and post-machine only that face, which means you pay for precision exactly where the part earns it on the drawing and nowhere else.

Landed cost, not sticker price. Section 232 duties on steel and aluminum now stand at 50% of the full customs value of covered metal articles, and they can reach fabricated tooling.³ A domestic composite equivalent takes that imported, dutiable metal out of the stack, along with the freight, the brokerage, and the minimum-order quantities that travel with it. Where the part fits the process, the composite line can sit below the metal-additive line, which already sits below the machined-metal curve for complex work at low to mid volumes.

For the job these parts are asked to do, the strength is there: continuous carbon fiber laid into an Onyx matrix returns parts as strong as 6061-T6 aluminum in suitable geometries, at a fraction of the machined lead time and cost.⁴ That claim carries a fence around it, and the phrase “in suitable geometries” is doing real work; fiber has to run where the load runs, and print orientation decides whether it does. We break the material science down on its own in what actually makes a composite part strong.

The bottom line

The cost-volume curve is not an academic exercise, but rather the best single predictor of whether a part belongs in additive, and it can be run against your own part numbers in about a minute. Volume, complexity, and country of origin are the three inputs that matter, and none of them requires a study to estimate. The test reduces to one line.

Overall, the break-even volumes are moving in one direction only. As processes mature, print speeds climb, and material options widen, N keeps pushing further into territory that used to belong to conventional processes alone, and the parts worth reconsidering first are the complex, the low-volume, and the tariff-exposed. Those parts are on your floor already, likely more of them than a quick survey suggests. The question worth carrying into the next planning cycle is not whether the curve has moved, but which of your part numbers have already crossed it.