Torrance, California · Digital Manufacturing
Divergent
The company that turned a factory into software - and proved it by building a record-setting hypercar and a missile airframe on the same line.
The company that turned a factory into software - and proved it by building a record-setting hypercar and a missile airframe on the same line.
Walk into Divergent's plant in Torrance and you will not find a stamping press for one car and a different press for another. You will find lasers fusing metal powder into shapes no human would draw by hand, and robots that join those shapes by following code - not steel jigs.
Change the part, change the file. The line does the rest. On a given day that line might be turning out a chassis node for a multimillion-dollar hypercar, a bracket for a spacecraft, and an airframe for a cruise missile. Same robots. Different software. That is the entire pitch, and for most of a decade it sounded like science fiction.
The skeptic's question is fair: lots of companies promise the factory of the future. What makes Divergent different is that its factory of the future has a customer list, a Pentagon order book, and a hypercar that beat a Koenigsegg around Laguna Seca. The future, inconveniently for the skeptics, already shipped.
Here is the dirty secret of making physical things. The hard part is rarely the design. The hard part is the tooling - the dies, molds, jigs, and fixtures that must be built before a single part exists. Tooling costs millions, takes months, and locks you into one shape. Want to change the design? Build new tooling. Want a different part? Build different tooling.
That math kills iteration. It is why cars look the way they do, why aerospace programs run late, and why a small design tweak can cost a fortune. The tooling tail wags the engineering dog.
Divergent's founders looked at that constraint and asked an awkward question: what if the tooling simply did not exist? Not reduced. Gone. If a machine could print any shape and a robot could assemble any combination of shapes, then the cost of changing your mind would collapse toward zero. Engineers could optimize for performance and weight instead of for whatever the press could stamp.
Kevin Czinger had already lived several careers before Divergent: U.S. Marine officer, federal prosecutor, Goldman Sachs executive, and co-founder of the early electric-car company Coda Automotive. Coda taught him a brutal lesson - that building cars the traditional way is enormously capital-intensive and environmentally costly, mostly because of everything that happens before the car exists.
So in 2014 he founded Divergent Technologies to attack the cause rather than the symptom. The bet was not "make a better car." The bet was "make the way we build anything obsolete." His son Lukas joined as co-founder and chief operator, and in 2025 stepped up as CEO while Kevin moved to executive chairman.
It was a long bet. Software that designs metal differently than humans do. Printers that build in application-specific alloys. Robots that assemble without fixtures. None of it was off the shelf. They had to invent the whole stack and convince investors that a hardware-and-software moonshot could become a real business. Eventually they raised more than a billion dollars trying.
DAPS - the Divergent Adaptive Production System - is the whole company in three verbs. Design: AI-driven generative software takes a performance target and a set of constraints and computes the optimal structure, usually something organic-looking and lighter than anything a human would specify. Print: metal 3D printers materialize that structure as nodes in application-specific alloys. Assemble: a universal, software-defined robotic cell joins the nodes - no part-specific fixtures, reconfigured by code.
Generative, AI-enabled engineering software optimizes for performance, weight, and manufacturability at once.
Laser powder-bed metal printing builds complex nodes in alloys chosen for the specific job.
A universal robotic cell joins parts with zero design-specific tooling - swap the file, not the line.
The output is the point. Fewer parts, less material waste, lighter structures, and almost no upfront capital expenditure for the customer - because there is no tooling to pay for. The Czinger 21C, the brand's flagship, is the showroom version: a tandem-seat hybrid hypercar around 1,250 horsepower whose unusual front-to-back seating exists because the chassis was optimized by algorithm, not tradition.
A factory-of-the-future story is easy to tell and hard to verify. Divergent's defense against the skeptic is a pile of receipts. Its system has produced parts for Bugatti, Aston Martin, and McLaren on the automotive side, and works with aerospace and defense customers including General Atomics, with reported relationships across the broader defense and space sector. By 2025 it was running 600+ unique part numbers.
The 2025 Series E - $250M in equity plus $40M in debt, led by Rochefort Asset Management - was raised explicitly to scale defense production as demand for U.S.-made missile airframes and drone structures climbed. Revenue grew more than five times that year. The hypercar gets the headlines; the defense order book pays the bills.
Divergent frames its work as bigger than parts. The pitch is to rebuild a 21st-century industrial base in the United States - one where complex structures are designed and produced on demand, where supply chains are software, and where you do not need a billion-dollar plant and a year of lead time to make something new.
There is an environmental argument underneath it, and it traces straight back to Kevin Czinger's Coda years. Optimized designs use less material. Printing adds metal only where it is needed instead of machining it away. No tooling means no tooling to manufacture, ship, and eventually scrap. The cleanest part is the one you never had to over-build.
Imagine an industrial world where iteration costs nothing. Where a defense program can revise an airframe in days, where an automaker can localize a part without retooling a plant, where the gap between an engineer's idea and a finished metal structure is measured in hours. That world rewards whoever can change their mind fastest - and Divergent built the machine that makes changing your mind cheap.
The competition is real - Relativity Space, Velo3D, Desktop Metal, and digital-manufacturing peers like Machina Labs are all chasing versions of this. But few have a single platform that spans a road-legal hypercar and a Pentagon contract. That breadth is the moat.
Return to that floor in Torrance. The lasers are still running, the robots are still following code, and the part coming off the line tonight did not exist as a tool yesterday and will not need one tomorrow. The hypercar in the lobby is not the product. It is the proof. The product is a factory that forgot how to need tooling - and that, quietly, is a much bigger idea than any car.