How 3D Printed Metal Parts Into the Ferrari F80 Got There

• 1
  Why Ferrari Could Put 3D Printed Metal Parts Into the F80
• 2
  The Main Routes for 3D Printed Metal Parts Into a Vehicle
• 3
  Which Alloys Actually Work, and Which Ones Fight Back
• 4
  Machining Allowance: Where the Printed Part Stops and CNC Starts
• 5
  Where the Process Stops Being the Right Answer
• 6
  Matching the Process to the Part

The F80 is a limited-run car. Production starts in 2025 and runs to 2027, with 799 units built to mark the marque's 80th anniversary. Low volume plus high performance is the exact shape of problem that metal additive manufacturing fits. A casting or forging tool for a single bracket can cost more than the bracket itself and take months to cut. A laser powder bed build needs no tool at all.

3D printed metal parts into a road car still have to survive the same loads as a machined part. The difference is how the shape gets made. Powder bed fusion melts metal layer by layer, typically 20–60 μm per pass, so the part grows out of a 3D file rather than out of a die. That opens internal channels, lattice cores and organic ribs that a mold cannot pull.

The trade is surface and tolerance. An as-built laser surface usually lands around Ra 8–12 μm and a general tolerance near ±0.1 mm. That is fine for a duct or a stiffener. It is not fine for a bearing seat, a seal face or a bolted joint. Those areas still get machined, and that is where the process split actually happens.

So the F80 story is not additive versus subtractive. It is additive for the shape and subtractive for the interfaces. Read the rest of this page with that in mind, because the same logic decides whether your part should be printed at all.

Laser powder bed fusion, sold as DMLS or SLM, is the workhorse for small to medium metal parts. A fiber laser scans a fine powder layer in an inert chamber, then the build plate drops and a recoater spreads the next layer. Layer thickness runs 20–60 μm. It holds thin walls down to roughly 0.4 mm and fine lattice struts, which is why it shows up in brackets, heat shields and ducting.

Electron beam powder bed fusion uses the same powder concept with an electron beam and a heated vacuum chamber. Layer thickness is coarser, around 50–100 μm, and the residual stress is lower because the build stays hot. It suits larger, chunkier parts where cracking during a cold build would be a problem. Surface finish is rougher than laser, so expect more finishing work.

Directed energy deposition is different in kind. Powder or wire is fed into a moving melt pool, so the head is not limited by a powder bed. You can build on an existing forging or add material to a worn part. Deposition rates are higher, but feature resolution is much coarser. Use it for near-net blanks and repair, not for fine internal channels.

Binder jetting prints a binder into a powder bed, then sinters the green part in a furnace. There is no melt pool, so residual stress is low and parts can be stacked densely. Shrinkage during sintering is significant and predictable, so the CAD model gets scaled up. It is a strong fit for larger batches of small parts where per-part laser time would be the cost driver.

Metal FDM extrudes a filament loaded with metal powder and sinters it afterward. It is the cheapest entry point and the loosest on tolerance. Treat it as a prototype or fixture route, not a production path for a stressed automotive part.

Aluminium alloys are the obvious first choice for weight. AlSi10Mg and AlSi7Mg print well because the silicon gives the melt pool something to solidify against, which limits hot cracking. They machine cleanly and take anodizing. The catch is strength: printed AlSi10Mg lands well below a 7075-T6 forging, so it suits housings, brackets and ducting rather than primary structure.

Titanium Ti-6Al-4V is the flagship AM alloy. It prints dense, has excellent specific strength, and is biocompatible. It is also expensive as powder, hard to machine, and prone to distortion on thin sections. Preheating the build plate and using a stress-relief cycle before wire EDM removal is standard practice. For a supercar, titanium AM makes sense where the part is small and the stiffness-to-weight payoff is large.

Stainless steel grades 316L and 17-4PH are the practical choices for corrosion resistance and moderate strength. 316L prints very reliably. 17-4PH needs a proper solution and aging treatment to reach its properties, and that heat treatment will move the part, so machining allowance has to cover it.

Nickel alloys such as Inconel 718 hold strength at high temperature and resist oxidation, which is why they appear in exhaust and hot-side parts. They are slow to print and punishing to machine. Copper alloys are the opposite problem: high thermal conductivity pulls heat out of the melt pool, so laser power and scan strategy have to be pushed hard to reach full density.

Magnesium AZ31B and AZ91D are printable but need tight powder handling because the fines are reactive. Tool steel and 4140-type grades print to high hardness after heat treatment, which suits tooling inserts and wear parts more than body structure.

Every printed metal part that matters goes onto a machine at least once. The reason is simple. Powder bed fusion gives you a shape within about ±0.1 mm, and the top surface of a build is flatter than the sides. Nothing about that is a bearing fit.

The standard approach is to leave stock on any face that has to locate something else. On aluminium and stainless, 0.3–0.5 mm per face is enough for a clean-up pass. On titanium and Inconel, leave 0.5–0.8 mm because the material moves more during stress relief and the cut is harder to control. Datum faces get machined first, then everything else is referenced from them.

For critical features we hold ±0.005 mm on the finished cut. That covers bores, seal faces, bolt-hole patterns and any surface that mates to a machined counterface. Surface finish after a finishing pass sits in the Ra 0.2–0.8 μm band when the drawing calls for it, with Ra 0.8–1.6 μm being the common production target.

One practical trap: build orientation. If an internal channel exits on a face you plan to machine, the semi-sintered powder stuck inside has to come out first. Plan a powder evacuation hole, and put it somewhere that a machining pass will clean up. Otherwise you spend an afternoon chasing trapped powder out of a cavity.

Another trap is thread callouts directly in the printed model. Printed threads are rough and their pitch diameter wanders. Model the hole undersized and cut the thread with a tap or a thread mill after printing. It takes one extra operation and removes an entire class of assembly failures.

Printing is the wrong call when the part is a simple prismatic block. If a 3-axis mill can cut it from bar in two setups, printing adds powder cost, a heat treatment, a build plate cut and a machining pass. The printed version will cost more and take longer.

It is also the wrong call for large flat plates and long shafts. Build volume caps part size, and a 4,000 mm envelope on a CNC mill covers work that no powder bed machine in a normal shop will touch. Those parts get machined from plate or bar, full stop.

High-volume parts are a third no. At a few hundred units a year, printing wins. At tens of thousands, casting or forging amortises the tool and beats the per-part laser time. The crossover point depends on part size and geometry, but it is real and it arrives sooner than most teams expect.

Fatigue-critical parts need a hard look at surface condition. As-built laser surfaces carry partially melted powder particles that act as stress risers. If the part sees cyclic load, those surfaces need bead blasting, polishing, or a machining pass, and the drawing has to say so. A printed part with an as-built fatigue surface is a warranty claim waiting to happen.

Finally, porosity. A well-tuned laser process reaches over 99.9% density, but density is process-dependent and should be verified. For pressure-tight or structural parts, ask for a density check, CT scan on a first article, or destructive sectioning on a coupon built alongside the part.