Metal 3D Printing Demand Outpaces Production Capacity
Summary
Velo3D says it needs 300-400 metal printers within one to two years but can build only 100, while 3D Systems sold more metal printers in H1 2026 than in all of 2025.
Between July 30 and August 11, 2026, Velo3D (VELO), 3D Systems (DDD), Protolabs (PRLB) and Carpenter Technology (CRS) each told investors on their quarterly earnings calls that orders for US metal 3D printing now exceed the number of qualified machines available to run them. Velo3D put the gap in units: it needs 300 to 400 machines over the next one to two years and can build only 100 [1].
Buyers now compete for machine slots instead of price
Metal 3D printing, or powder-bed fusion, uses a high-power laser to melt metal powder layer by layer into a finished part. Unlike casting or forging, it makes complex internal channels and thin walls without new tooling, which suits the hard-to-machine parts inside rocket engines, missile components and power-generation equipment. The constraint is qualification: aerospace and defense parts cannot simply move to another line, because the machine, the process parameters and the building each have to be qualified first, and installing, tuning and qualifying a machine typically takes a quarter or more per step.
Over the past two quarters, three sources of demand pushed a set of parts out of prototyping and into rate production at roughly the same time: defense stockpile replenishment, the commercial space launch build-out, and the electrical equipment build-out driven by AI. All of those parts have to land on already-qualified metal printing capacity, which cannot be conjured up inside one quarter, so buyers started bidding for machine slots. Velo3D's CFO attributed the second-quarter revenue increase primarily to higher average selling prices and higher volume in its own printing services, and contract manufacturer Mears ordered its fifth Sapphire XC with options on two more [1]. The same figure of 400 machines was a ten-year capacity plan on the March 2026 call [2]; five months later it had compressed to one to two years [1].
What three companies disclosed, and one datapoint that cuts the other way
Velo3D quantified the shortfall in machine counts. On the August 11 call, the CEO said the company needs 100 machines today just to run the programs it has already won, and that production is lagging; that it needs 300 to 400 machines over the next one to two years but can build only 100; and that even if all 100 are running by mid-2028, they will be sold before they are built [1]. He also said the Fremont plant is fully occupied, which is why the company pulled the opening of its new Livermore site forward to the fourth quarter of 2026 [1].
3D Systems had reported numbers pointing the same way a week earlier. On August 4 it said it sold more metal printers in the first half of 2026 than in all of 2025, that demand was still rising, and that it is significantly expanding its own metal parts production capacity; it currently has roughly 220,000 square feet of space dedicated to metal printing [3]. One tier further upstream, Carpenter Technology, which supplies the alloy powder these machines consume, said on July 30 that its additive business grew both year over year and sequentially on aerospace and defense demand [4].
Protolabs supplied both demand-side confirmation and the strongest evidence against the framing. The rapid-manufacturing company said on July 31 that Lockheed Martin (LMT) expanded its use of Protolabs' 3D printing capability as it builds out a network of strategic partners in additive production. Its own 3D printing revenue, however, fell 2.7% year over year in constant currencies, mainly on a 6.7% decline in Europe, while US DMLS (direct metal laser sintering) and MJF stayed strong in aerospace, defense and electronics, where the company is adding capacity [5].
Revenue timing now follows machine installation
The control point has moved from the order book to whoever holds qualified machines. The first tier is the owners of that capacity: the printer makers' own parts operations, such as Velo3D's printing services and the internal metal parts capacity 3D Systems is expanding, along with qualified contract manufacturers like Mears and Protolabs [1][3][5]. The printer makers' own order books sit behind that, because announced expansions take four to eight quarters to become deliverable machines, and below both sits the alloy powder and optical content fitted into every machine [4].
Financially this shows up in gross margin before it shows up in revenue. Margin moves first as the mix shifts from selling machines to selling printed parts, while the printer makers' revenue waits until new machines are installed, tuned and qualified. The operating metric to track therefore shifts from orders and backlog to machines installed and actually utilized.
The boundary here is clear. Protolabs' total 3D printing revenue declined year over year [5], so the claim holds only for US metal powder-bed work serving aerospace and defense, not for additive manufacturing as a category. The two printer makers also contradict each other on the size of the installed base: 3D Systems alone has roughly 220,000 square feet dedicated to metal printing [3], against the Velo3D CEO's statement that the entire US holds under 250,000 square feet of additive capacity [1]. The scarcity is credible, but Velo3D's scaling of it is not corroborated. What can be checked next is whether Livermore opens on time in the fourth quarter, when 3D Systems' added parts capacity comes online, and what utilization those machines run at once installed.
Companies exposed to the same change
- nLIGHT (LASR): Builds the high-power fiber lasers that melt the powder inside metal powder-bed machines, with several fitted into each multi-laser system, so its unit demand follows how many machines get built rather than how many parts get printed; if build rate really is the ceiling, its demand would move ahead of printed-part volumes.
- Novanta (NOVT): Its Cambridge Technology unit makes the galvanometer scan heads that aim the laser during a build, another component fitted once when the machine is assembled, which puts it on the same per-machine count.
- IPG Photonics (IPGP): The other merchant supplier of fiber lasers to metal printer makers, on a far larger base where additive is diluted, which makes it a useful check on how large the demand change at this tier really is.
Sources
[1] Drillr · Velo3D (VELO) · 2026-08-11 · Q2 2026 earnings call
And we need actually 300 or 400 machines in the next one or two years, but we can only build 100.
[2] Drillr · Velo3D (VELO) · 2026-03-24 · FY2025 Q4 earnings call, prepared remarks
[3] Drillr · 3D Systems (DDD) · 2026-08-04 · Q2 2026 earnings call
[4] Drillr · Carpenter Technology (CRS) · 2026-07-30 · FY2026 Q4 earnings call
[5] Drillr · Protolabs (PRLB) · 2026-07-31 · Q2 2026 earnings call
This is only meant to surface industry changes and companies you may have overlooked - it is not a stock recommendation.