Casting and fabrication are usually compared on unit price, and that comparison is misleading because the two routes spend money at different times. A fabrication pays for labour and material on every part. A casting pays for tooling once and then pays less per part, but only after the tool exists and only if the same shape is wanted many times over.
So the real question is not which process is cheaper. It is how many times you are going to pay for the shape, and how much of the shape's cost is fixed rather than variable. A part made once is almost always cheaper fabricated. A part made in a long series is often cheaper cast, provided the geometry suits casting and the buyer can absorb the tooling at the front.
What makes the decision difficult is that the two routes are not interchangeable at the level of detail. A casting and a fabrication of the same nominal part are different objects with different surfaces, different wall behaviour and different failure modes.
A casting buys geometry freedom. Shapes that would need several fabricated pieces and several seams can be produced as one continuous form with material where the load is and thin sections where it is not. Bosses, ribs, internal passages and compound curves that would be expensive or impossible to weld together come out of a mould as a single part.
It also buys repetition. Once the tool is right, every part is the same part, and the labour content per unit collapses. That consistency is worth a great deal in an assembly where a bracket has to meet a machined face on every unit, because the variation between parts is small and predictable rather than dependent on who welded it.
The third thing it buys is a settled surface. A cast part has no seam lines to finish, no weld to dress, and no heat-affected zone along a joint. For a visible component, that is worth more than the unit price difference suggests.
It costs a tool, and the tool costs money before a single part exists. That tool also has to be right, because correcting it means cutting metal rather than adjusting a drawing. The first part is therefore a long way from the decision, and the whole route rewards a buyer who is certain about the geometry rather than one who is still exploring it.
It costs flexibility. Once the tool exists, changing the part means changing the tool, which means the design freezes at the point the tool is ordered. A fabrication can absorb a change after the first article with a cut and a weld. A casting cannot absorb it without a new tool or a modification that is often more expensive than the original.
And it costs a longer first cycle. The pattern or die has to be made, tried, and corrected, and only then does production start.
A casting has to come out of its tool, which means every surface parallel to the draw direction needs draft. That draft is not visible in a model and it changes the part: a wall that was drawn vertical becomes slightly tapered, and a feature that relied on a truly vertical face no longer has one. Any dimension that matters has to be reconciled with the draft rather than drawn independently of it.
Wall thickness is the other constraint. Castings want reasonably uniform walls, because thick sections cool more slowly than thin ones and the difference produces shrinkage and internal stress. A design that concentrates material in one area to solve a stiffness problem creates a cooling problem in the same place. Ribs exist precisely to add stiffness without adding a thick section.
Small features are the ones that quietly disappear. Sharp internal corners, deep narrow slots, fine threads and thin upstanding walls all sit at the edge of what a mould can fill and release.
Cast material is not fully dense. Gas and shrinkage leave voids, and the voids are not always where the inspection finds them. A part can pass a check on its accessible surfaces and still have porosity in a section that matters. That is the reason castings are often specified with a soundness requirement rather than just a shape.
Porosity matters most where the part has to be sealed or welded. A casting that must hold fluid, or that has to be joined to a fabricated section by welding, is sensitive to what is inside the wall. Welding over porosity produces a joint that looks complete and leaks, or a repair that chases a void deeper into the part.
The practical consequence is that repairability has to be decided in advance. Some castings are routinely impregnated or welded to close voids, and some are rejected. Which one applies depends on the service, and it should be written into the requirement rather than discovered at the end.
Fabrication wins whenever the geometry is basically prismatic. A bracket made from plate, tube and a couple of bends does not benefit much from being cast, because a casting would reproduce a shape that was already cheap to make. The tooling cost buys nothing when the fabricated version has few operations.
It also wins where the section has to be thin and uniform over a large area. Sheet and tube give a consistent wall by their nature, and a casting of the same form has to fight shrinkage to achieve it. For panels, shields, ducting and tube runs, fabrication is not a compromise. It is the process the geometry was always asking for.
And it wins on change. In development work, where the shape is being refined and the first parts are instruments for learning rather than products, a route that can be modified between builds is worth more than a route with a lower unit price. A fabricated prototype that is slightly worse than the eventual casting is still a better first step than a tool cut for a shape nobody has tested.
Many real assemblies are neither one nor the other. A cast body with fabricated connections, a fabricated shell with cast inserts at the mounting points, or a casting that is machined after casting to establish its functional faces, all combine the two. The reason is usually that each process is being used for the thing it does best: casting for the complex form, fabrication for the simple joint, machining for the tolerance that neither can hold.
The hybrid needs its interfaces defined carefully. A cast insert welded into a fabricated shell has a joint between two materials with different behaviour, and that joint is where the problems concentrate. Specifying the insert's material, its surface condition before welding, and which faces are machined afterwards is the difference between a hybrid that works and one that cracks at the boundary.
The route decision is easier when these points are settled explicitly rather than by default.
None of that is a costing exercise. It is a description of the part, and the part usually makes the route obvious once it is described properly.
Quantity matters, but shape decides first. A complex form with internal features and compound curves is a casting candidate at almost any volume above one, because fabricating it means many pieces and many seams. A simple formed bracket stays a fabrication at any volume, because the tooling buys nothing that the sheet did not already provide.
Between those extremes the decision is about how much of the part is functional geometry and how much is joining. Every seam in a fabrication is a joint to be made and inspected. Every undercut in a casting is a feature to be tooled and released. Comparing the two means comparing those counts rather than comparing a price per unit.
Our own work sits mainly on the fabrication side, which is why the TIG aluminum welding pages show what welded aluminium sections can achieve without a tool, including shapes that would otherwise need a casting. The titanium fabrication components and TIG titanium welding parts ranges show the same logic in a material that is difficult to cast thinly.
Comparisons between materials and their practical limits sit under shop by material, and the way a route is agreed with a customer is set out under customization. Smaller fabricated components that would rarely justify a tool are collected under other parts. The right question is not which process is better. It is which one the geometry has been asking for all along.
