The aerospace company that decided the rocket isn't the hard part. The factory is. So it built a printer the size of a building and pointed it at orbit.
Walk into Relativity Space's Long Beach headquarters - a million square feet that once stamped out military cargo planes, now nicknamed "The Wormhole" - and the strangest thing is what you don't see. No endless racks of fasteners. No wall of suppliers' crates. Instead, robotic arms trace glowing metal in slow, deliberate passes, laying down a rocket the way a 3D printer lays down a desk toy, only several stories tall.
This is the bet, made physical. Relativity argues that the bottleneck in spaceflight was never really the physics - it was the manufacturing. Thousands of parts, thousands of suppliers, years of lead time. So the company built machines called Stargate to print most of a rocket from raw metal wire, and a software stack to design it fast. The rocket, in this telling, is almost a byproduct. The factory is the invention.
Most rocket companies build a rocket and call the factory a detail. Relativity built the factory and treats the rocket as the output./ the thesis, in one line
A traditional orbital rocket is a sprawling supply chain wearing a metal skin. Each vehicle can carry tens of thousands of parts, each with its own tooling, vendor, and waiting list. Change the design and you change a hundred contracts. The result is that rockets, those supposed symbols of the future, are often built with the patience of a cathedral.
Relativity's founders looked at that and saw something almost heretical: the part count itself was the enemy. Fewer parts means fewer suppliers, fewer failure points, faster iteration. And the cleanest way to slash part count is to stop assembling and start printing - to grow a rocket as a handful of large structures rather than fasten it together from thousands of small ones.
The conventional rocket has tens of thousands of parts. Print it, and that number can fall by orders of magnitude. Complexity, it turns out, was a choice./ why additive manufacturing
Tim Ellis and Jordan Noone met running the USC Rocket Propulsion Lab, the kind of place where undergraduates set things on fire for academic credit. In 2015 they took their wager - that 3D printing could one day produce an entire orbital rocket - through Y Combinator and into a company. It sounded, at the time, like science fiction with a business plan.
The audacious part wasn't building a rocket. Plenty of startups try that. It was insisting the printer came first. Relativity poured years and capital into Stargate before it had a flight-worthy vehicle to show for it - betting that owning the means of production would matter more, long-term, than any single design. Investors, eventually, agreed to the tune of more than a billion dollars.
Co-founder and former CEO, now executive chairman. The propulsion engineer who would not let go of the printer-first idea.
Co-founder and former CTO. Co-architect of the early vision before moving on to back other space ventures.
Former Google CEO who became Relativity's CEO and chairman in 2025, with a major personal investment to steady the company.
The hard sell wasn't "we'll build a rocket." It was "we'll build the machine that builds the rocket, and you'll have to wait."/ the printer-first wager
In March 2023, Terran 1 lifted off from Cape Canaveral on a mission cheekily named "Good Luck, Have Fun." The rocket - the great majority of it printed - cleared the violent pressure peak of Max-Q, then its upper stage faltered and it never reached orbit. By most ledgers, a failure. By Relativity's, a proof: a printed structure had survived the worst loads of launch. The point was made; the program was promptly retired.
What replaced it is the real commercial play. Terran R is a medium-to-heavy-lift, fully reusable rocket - first stage, second stage, and fairing all meant to come home and fly again. It rides on seven Aeon R engines, themselves 3D-printed and fed by liquid oxygen and methane, each rated around 302,000 pounds of thrust. The target customer is anyone who currently waits in line for a Falcon 9.
Reusable medium-to-heavy-lift launcher. First flight targeted late 2026, Cape Canaveral. Built to compete for the workhorse launch market.
3D-printed, methane-and-oxygen, ~302,000 lbf of thrust, designed for reuse. Seven of them power the first stage.
Among the largest metal 3D printers in the world. The 4th-gen prints horizontally, feeding multiple wires into one head.
The first largely 3D-printed rocket to launch. Cleared Max-Q in 2023; its job was to prove the structure, and it did.
Terran 1 was an existence proof with a sense of humor. Terran R is the invoice./ proof, then product
Here is the part skeptics find hard to wave away. Before Terran R has left the ground, Relativity has signed launch agreements reportedly worth more than $3 billion - with names like SES, Intelsat, NASA, the U.S. Space Force, and space-logistics partner Impulse Space, with whom it plans a privately funded mission toward Mars. Buyers, in other words, are wiring deposits on a future the company has yet to fully deliver.
That is either reckless faith or a market starved for launch capacity. Probably both. With one dominant provider setting the pace for the entire industry, customers have an obvious incentive to fund a credible second source - even one that prints its hardware and is still working toward its first orbital flight.
Relativity has always framed itself in the long arc - multiplanetary ambitions, an "interplanetary society," the usual stars-in-the-eyes vocabulary of the space set. Strip away the poetry and a sharper idea remains: an adaptive, software-defined factory that can print complex flight hardware on demand. Rockets first, because rockets are the hardest thing to print. After that, who knows.
Under Eric Schmidt, the company has begun talking about something further out still - orbital infrastructure for artificial intelligence, even data centers in space, where power and cooling obey different rules than they do on a crowded planet. It's speculative. It's also exactly the kind of swing a company built around a giant printer is positioned to take.
The printer doesn't care whether it's building a fuel tank or a future. That ambiguity is the entire point./ the factory as platform
Return to that factory floor. The robotic arms are still tracing metal, still building a rocket the slow, glowing way. Nothing about the scene has visibly changed - and that's precisely what's changed. What looked like a science-fair stunt a decade ago is now a manufacturing method with billions in contracts riding on it and a former Google CEO at the controls.
The skeptic's question is fair: Terran R still has to fly, and a printed rocket is only as good as its first orbit. But the wager underneath has already half paid off. Relativity made the industry take seriously an idea it once dismissed - that you don't assemble the future. You print it, one deliberate pass at a time. The arm finishes another layer. The rocket grows a little taller. The bet keeps building.