Robotic Metal Forming: Where It Actually Lives in Architecture

Robotic incremental sheet forming does something that sounds impossible if you have spent time around a press brake. A robot arm with a hardened spherical tool pushes a flat sheet into a complex three dimensional shape, incrementally, with no die and no mold. Change the CAD file and you change the part, with no tooling to build.

The question for anyone in construction is whether this is coming to architectural metal. The honest answer has two parts, and the second is more interesting than the first.

Machina Labs is a defense company now

Machina Labs, founded in 2019 in Los Angeles by a team drawn from SpaceX, Microsoft, Google, Northrop Grumman, and Stratasys, is the company most associated with this technology. Its RoboCraftsman platform combines forming with laser metrology and closed loop correction, trimming, drilling, and finishing.

The published capability is genuinely impressive. Production materials include aluminum in the 2000, 5000, 6000, and 7000 series, mild steel, and 304 and 316 stainless. Thickness up to a quarter inch. Maximum part size 12 feet by 5 feet, with deep drawn parts to 4 feet. Forming can begin within hours of receiving a CAD file, with first parts inside a week.

One specification deserves a closer look than the marketing gives it. Tolerance is stated as plus or minus 0.3 percent of the maximum dimension, or plus or minus 1 mm, in the roboformed state. On a 12 foot part, 0.3 percent is about plus or minus 0.43 inches. That is a proportional tolerance, and the sub millimeter language does not describe large parts.

The company has raised roughly $169 million across four rounds, most recently a $124 million Series C in February 2026 with Toyota’s Woven Capital and Lockheed Martin Ventures participating. It is building a 200,000 square foot US factory with capacity for up to 50 forming cells.

And it has no construction business. The Series C release frames the company as manufacturing infrastructure for defense and advanced mobility. Named customers are the Air Force Research Laboratory, the Air Force Rapid Sustainment Office, an unnamed defense prime working on missiles and hypersonics, Toyota, and NASA. The word architecture appeared once, in a 2023 list of markets served, attached to no named project, and has since disappeared from company messaging.

If you read a trade piece claiming Machina Labs serves construction, it is recycling that single line. There are no verifiable construction or facade projects.

Where robotic forming does show up in architecture

This is the part worth knowing, because it is real and it predates the current wave of funding announcements.

A. Zahner Company in Kansas City, one of the best known architectural metal fabricators in the country, has been building its own robotic forming capability in house.

For the Cornell Tech Bloomberg Center on Roosevelt Island, designed by Morphosis, Zahner retrofitted a Fanuc arc welding robot with a custom 3D printed head and reprogrammed it to push precut tabs in perforated stainless steel outward, producing a louvered facade effect across the panels. That was developed in the summer of 2015 as a second generation of computer controlled tab pushing, following manual methods used on an earlier project.

Zahner has also worked on robotic English wheel forming, pairing a nineteenth century metal shaping tool with a six axis robot and computational software that translates geometry into forming instructions. The stated motivation was architects’ rising demand for double curved forms and the need to produce them affordably.

That is the honest architectural story for this technology: a specialist fabricator solving specific project geometry with robots it adapted itself, rather than a venture backed platform company selling into construction.

The economics, and what nobody will tell you

The qualitative case for die free forming is clear. Machina Labs claims tooling cost savings that can exceed $1 million per unique part design and more than a tenfold reduction in lead time against conventional die manufacturing. Both are savings on avoided tooling rather than claims about cost per part.

The counterweight is straightforward physics. A robot moving incrementally is slower per part than a press. Stamping wins decisively at volume because cycle time is a fraction of forming time. Incremental forming wins at low volume and high mix, where die cost dominates and there is no volume to amortize it against.

One real technical advantage over press brake work: springback is largely absent because the geometry is produced incrementally rather than in a single bending operation.

Where is the crossover point? No published break even study exists, from Machina Labs or from any independent source. The company’s own technical material contains no volume threshold. Anyone quoting a break even number is estimating. If that number matters to your project, it has to come from a fabricator quoting your specific part.

The code question for formed architectural metal

Short answer: formed metal panels are cladding, and there is no code prohibition. The cost is in testing, not in approval.

The International Building Code requires that exterior wall coverings be approved for their specific application. That is a performance and approval requirement rather than a prescriptive list, so custom geometry does not by itself create a problem.

Wind resistance is established through ASTM E330 testing, with the designer confirming rated resistance meets design wind pressures per ASCE 7 Chapter 30 for the building height and location. For a bespoke formed panel that means project specific mockup testing, which is the real cost driver.

On fire, solid metal panels in aluminum, steel, or zinc are noncombustible and do not by themselves trigger NFPA 285. That standard is triggered by combustible components in exterior wall assemblies on buildings over 40 feet in Type I through IV construction, most notably composite panels with polyethylene cores and combustible insulation behind the panel. A solid formed metal skin is the easy case. The assembly behind it is where the risk lives.

So the practical difference between a bespoke formed facade and a catalog system is not approvability. It is that catalog systems arrive with manufacturer test data and yours will not, so budget and schedule have to carry the testing.

What this means if you are specifying

Robotic forming is real, it is in architectural metal today, and it is concentrated in a small number of specialist fabricators rather than available as a commodity service.

If your design calls for double curvature or panel by panel variation, ask fabricators directly what forming methods they hold in house and what geometry they have actually produced. Ask for the mockup testing scope early, because that is where the schedule risk sits. And treat any claim that a venture backed forming platform is serving construction with skepticism until someone can name the building.

Sources

Machina Labs, published capabilities and tolerances
Machina Labs, $124 million Series C, February 2026
The Robot Report, Machina Labs Series C coverage
The Robot Report, Machina Labs Series B, October 2023
Machina Labs, technical overview of incremental sheet forming
Zahner, robotic tab forming for the Cornell Tech Bloomberg Center
Penn State, robotic English wheel research with Zahner, February 2018
Zahner, efficient fabrication of double curves in architecture
Engineering.com, the what, why, and how of roboforming
Metal Construction Association, metal panels and NFPA 285

Keep reading

3D printing beyond concrete: tools work, buildings do not yet
Mass timber: codes, costs, and market growth
Wall assembly guide: fire ratings, STC, and code resources