Let me start with the kind of opinion that gets you uninvited from cautious industry panels: Velo3D becoming a supplier to SpaceX and Anduril is a bigger deal than most people realize—and not for the reason you think.
It's not because the Sapphire machines are flashy. It's not because aerospace buyers have endless budgets. It's because a defense and space qualification process is the closest thing our industry has to a truth serum. Once you've been through a few audits, you start to see the difference between a machine that can print a part and a supplier that can prove the part will work every time.
I'm an application engineer who has handled metal additive manufacturing orders for eight years. I've personally made—and documented—21 significant mistakes, totaling roughly $640,000 in wasted budget. Now I maintain our team's pre-build checklist. Take it from someone who has scrapped enough titanium to know the difference between hype and qualification.
When people search for "velo3d supplier to spacex anduril" or "velo3d supplier to anduril spacex," they're usually asking one question: is this technology real? My answer: yes, but the real story is how it got real.
The Qualification Signal That Most People Skip
Publicly, Velo3D has named SpaceX and Anduril as customers in its own press releases and investor materials—those are available as of January 2025. And in this industry, having a name on a supplier list matters less than why it stays there. If the parts didn't pass inspection, the relationship would not continue. That's the part that should interest every engineer.
So when someone says "velo3d supplier to spacex anduril," they're not listing names. They're describing a qualification barrier.
I learned this the hard way. In 2018, I submitted a complex impeller to an outside metal AM shop. It looked fine on my screen. The result came back with internal porosity that no one had signed off on testing (including me). 14 pieces, $23,000, straight to scrap. The initial quote was less than half of what we spent after rework—or rather, after we admitted the rework wasn't possible. That's when I learned that the printer was never the problem. The process around it was.
How to Create 3D Files for Printing: The Question I Actually Care About
Most of my email from engineering teams boils down to how to create 3D files for printing. People expect a tutorial on CAD modeling. I usually give them a lecture on file preparation.
Here's what you need to know: the file-prep stage is where the failures hide. Orientation, support placement, thermal simulation, material calibration—any one of these can turn a beautiful CAD model into a $50,000 paperweight. If you've ever had a print fail on layer 400, you don't need me to explain the feeling.
Take a mundane example: a tool for spiral cutting carrots. It's a tiny consumer gadget. It will never need a rocket-grade metal printer. But if you design one, you still need to answer the same questions: where does the tool experience stress? What is the minimum wall thickness before the spiral snaps? Can the overhang be printed without supports? That's why the phrase "how to create 3D files for printing" is misleading. It sounds like a CAD question. It's really a manufacturing engineering question.
Velo3D's support-free capability removes one of those headaches. It's genuinely useful for channels and overhangs that would require heavy supports on other platforms. But someone still has to choose the orientation, set the simulation parameters, and judge whether the result meets the requirement. The machine is a tool. The engineer is the system.
Don't Feed the CNC-versus-Additive Panic
Let me be clear before the comments light up: I still send work to shops like Malones CNC Machining Inc. A well-run machine shop can hold tolerances that many printed parts cannot touch, and at a fraction of the cost. Anyone who tells you metal 3D printing will completely replace CNC is trying to sell you something.
But the industry is evolving. What was best practice in 2020 may not apply in 2025. The fundamentals—material property, tolerance, cost per part—haven't changed. The execution has. Part of that evolution is realizing you no longer have to choose between "machined because that's how we've always done it" and "printed because we want a showcase part."
Velo3D's path from aerospace curiosity to production supplier proves the conversation has moved from "can we print this?" to "should we print this?" That second question is harder, and it requires more experience.
But My Parts Aren't Rocket Parts
I get it. You're not making a combustion chamber for a lunar lander. Your parts are simple and your budget is tight. So why should you care about a "velo3d supplier to anduril spacex" headline?
Because the same discipline that gets a part qualified for space should be the same discipline you use on a $5,000 bracket. The paperwork might be shorter, but the questions are identical: Can the supplier prove the material matches the spec? Can they show that the build process is repeatable? What happens if a part fails on the customer's inspection? That's not aerospace bureaucracy. That's good engineering.
One of my biggest regrets is not building these questions into my early workflows. I still kick myself for accepting a material certificate that, it turned out, had no traceability. If I'd asked one more question, I'd have saved $23,000 and a week of delays. Instead, I learned the rule the expensive way: the print is only as reliable as the qualification evidence trail.
Bottom Line
I'll restate my opinion in the simplest terms: Velo3D as a supplier to SpaceX and Anduril is not trivia. It is a milestone in the industrialization of metal additive manufacturing.
The technology was already impressive. What's new is the evidence that it can survive the most demanding qualification environment in American manufacturing. That should change how every engineering team approaches metal AM—not by rushing out to buy a printer, but by raising their own standards for proof.
We don't need more printers. We need more people asking better questions. But if I can save you one mistake, this is the one: don't wait until you lose 14 pieces to start asking them.