Before data centers move to orbit, Gravitas has to prove the power is there

K2 Space is treating orbit less like a place to visit and more like a place to build. Its first high-powered satellite, Gravitas, is meant to demonstrate the hardware needed for space-based compute infrastructure: enough electricity, enough surface area, and enough payload capacity to make orbital data processing look less theoretical.

According to TechCrunch’s report, Gravitas is scheduled to fly on a SpaceX Falcon 9 by the end of the month. The spacecraft has a mass of about 2 metric tons and a 40-meter wingspan once its solar panels are deployed, giving it a physical scale closer to infrastructure than a shoebox satellite with ambitions.

The central figure is 20 kW. K2 says Gravitas can generate that much electricity, putting it in the same rough power class as SpaceX’s next-generation Starlink satellites and below the 25 kW class of ViaSat-3, but above what most spacecraft carry. In space, power is not a footnote; it is the budget that decides what can compute, transmit, maneuver, and survive.

The source material also shows that gravitas also carries 12 undisclosed payload modules, including a 20 kW electric thruster that K2 expects to be the most powerful ever flown in space. If the system performs as planned, it would validate more than one satellite design. It would test whether a commercial spacecraft can support the kind of energy-hungry systems that orbital data centers would require.

K2’s broader claim is not that one satellite becomes a data center overnight. It is that high-power buses can turn satellites into expandable platforms for communications, sensing, edge processing, and eventually orbital compute. The company, founded by Karan and Neel Kunjur in 2022, has reportedly raised $450 million and reached a $3 billion valuation in December 2025, which gives this mission real financial gravity as well as orbital mass.

There are still boundaries around what is confirmed. The payload customers are undisclosed, the commercial model remains early, and the economics of running data infrastructure in orbit are not settled by a successful launch. But Karan Kunjur’s line that “the future is higher power” captures the technical hinge: without much more electricity in orbit, space compute remains a diagram. With it, the diagram starts asking for an operations manual.

The real signal here is not that Earth’s data centers are about to migrate above the clouds. It is that orbital infrastructure is beginning to borrow the language of power budgets, modular payloads, and iterative platforms from terrestrial computing. Space, politely, is becoming a systems engineering problem at industrial scale.