Last October, I was sitting in a windowless conference room at 6:40 PM, staring at a spreadsheet that refused to add up. Every quarter, our team chases the same two numbers: cost per part and lead time. And every quarter, someone asks me why the second one keeps slipping.
I'm a procurement manager at a 40-person aerospace contract manufacturer. I've managed our outsourced fabrication budget—roughly $1.1 million a year—for six years. I've negotiated with more vendors than I can count, and I've logged every order in a cost tracking system that my team jokes is older than our ERP. I'm not a metallurgist. I'm the person who asks, "What does it actually cost to get this part on a truck?"
So when engineering told me we needed to qualify new suppliers for metal additive manufacturing, I didn't get excited. I got suspicious.
Velo3D Kept Appearing in Places I Couldn't Ignore
My first search was simple: "military additive manufacturing Velo3D." I expected white papers and glossy ads. Instead, I kept landing on references to actual hardware. According to Velo3D's public announcements, the company has supplied metal AM systems and parts to customers including SpaceX and Anduril. "Velo3D supplier to Anduril SpaceX" isn't just a flex. It means the company's Sapphire systems have been used for parts that need to survive a rocket launch or a defense platform, not just a trade show demo.
I get why people ignore vendor customer lists. Any company can publish a logo if they've sold a single test part. But I'd argue that in defense and aerospace, those two names carry more weight than a hundred marketing slides. The way I see it, if your part passes SpaceX's internal requirements, you're already ahead of most suppliers.
Still, I've watched enough "qualified suppliers" fail quality audits over the past six years to know that logos aren't proof. We had to run our own build.
The Cheap Quote Wasn't Cheap
We sent the same part—a small manifold with internal cooling channels—to four vendors. The part is about the size of a soda can, but it has complex internal geometry that would be nearly impossible to machine. We were quoted $680 per unit from Vendor A, $620 from Vendor B, and $890 from Vero3D—no wait, from Velo3D. I'm mixing up the quotes; let me be precise. Velo3D's per-unit price was actually the highest, around $1,150. But that's a silly way to compare.
Last year, I made the same mistake when I searched "co2 laser los angeles cost" for a separate cutting project. One shop quoted $45/hour, another $90, and another $120. The cheap price didn't include material handling, nesting, or cleanup. The expensive one included everything. The cleanest comparison made the "expensive" shop cheaper in the end. I should have learned that lesson earlier.
With metal additive, the unit price is the smallest part of the story. I built a TCO model with rows for build preparation, support structuring, removing supports, heat treatment, machining, inspection, failed builds, and operator time. I didn't invent these categories. I copied them from our five-year record of machining and post-processing costs. Those numbers are from our 2024 procurement files, and they won't match every shop on Earth. Verify current rates before you build a budget around them.
Here's where it got interesting. The lower-priced vendors required us to redesign the manifold to fit their support-removal process. Redesign costs are not free. Engineering time, drawing changes, validation, and risk all have price tags. One vendor told me, "You'll save $30,000 on the machine, maybe more." They were right about the machine. They were less right about the eleven weeks of engineering time we'd need to make the part printable.
I'm not 100% sure of the final total, but the redesigned path came out near $780,000, while the Velo3D path was around $740,000. The Velo3D quote also included a build analysis before we committed. That pre-build review caught a wall thickness issue that would have failed a leak test. Without it, we might have spent $60,000 on a qualification run that went nowhere. To be fair, the other vendors had good machines. For simpler parts, they'd probably be fine. But for this part, they weren't.
VMC Weather Minimums and a Paloma Walk Into a Spreadsheet
I know this sounds like the setup to a joke, but bear with me. During the same period, I was studying for a private pilot checkride. I looked up "VMC weather minimums" more times than I care to admit. VMC—visual meteorological conditions—are the bare minimum visibility and cloud clearance specs for visual flight. According to FAA 14 CFR 91.155, the minimums vary by airspace and altitude. A lot of pilots will tell you that legal minimums are not safe minimums. You need margin for wind, terrain, and fuel.
Meanwhile, my wife asked me to research "how many carbs in VMC Paloma" because she wanted to order a fancy cocktail without wrecking her day. I typed it into Google and got a bunch of recipe sites. The answer, as far as I can tell, depends on the mixer: a traditional Paloma with grapefruit soda is around 20–30 grams of carbs; one with sparkling water and fresh grapefruit can be closer to five. Take that with a grain of salt—I'm not a nutritionist.
Why am I telling you this? Because those two searches had the same lesson as every quote we received. Numbers without context are dangerous. VMC weather minimums are legal, but they're not a guarantee of safety. A Paloma's carbs depend on how it's made. CO2 laser Los Angeles cost depends on what's included. And Velo3D's cost per part depends on build preparation, supports, post-processing, and whether the part is actually designed for the process.
The people who search "velo3d supplier to anduril spacex" are probably asking a deeper question: "Can this company handle parts that cannot fail?" Based on the public track record, the answer is yes. But my job isn't to trust the answer—it's to verify it with our own test coupon.
What We Ultimately Did
We didn't buy a Velo3D Sapphire. At least not yet. Our volume didn't justify the capital expense. Instead, we used Velo3D's contract manufacturing services for the first qualification batch. I know that sounds like a cop-out, but it was the right procurement decision. It let us evaluate the technology without owning a machine that would sit idle 60% of the year.
The parts came back with dimensional readings that held within spec. Internal channels were cleaner than any of our CNC-based alternatives. We'd still use machining for simple housings and brackets—metal additive manufacturing will not replace our entire machine shop. But for parts we literally cannot make any other way, Velo3D's approach has a real advantage.
Lessons I'd Hand to Another Procurement Person
If you're evaluating Velo3D or any metal AM vendor, build the full cost model before you fall in love with a quote.
- Include engineering time for support-free design reviews. If a vendor forces you to redesign the part, that's hours of hidden cost.
- Include failed builds. On complex parts, a 20% failure rate is not an anomaly; it's an assumption.
- Include post-processing support removal. It's the task everyone forgets.
- Include qualification costs. A printed part that isn't documented and inspected isn't an aerospace part.
I'm not a Velo3D salesperson. My experience is based on a narrow slice: about 60 candidate parts over 9 months, most with internal channels and organic shapes. If you're working with simple, conventional designs, your experience might differ. The cheapest option might genuinely be the best option. But if your parts look anything like ours, ask the TCO question early.
In the end, the best procurement decision I made this year wasn't choosing a vendor. It was forcing myself to compare apples to apples—and remembering that even a Paloma has hidden carbs.