Seven months. That is how long it took K2 Space to go from a $3 billion valuation to a $6.8 billion one. The catalyst was a $500 million Series D, co-led by Kleiner Perkins and ICONIQ, for the largest high-power satellites currently flying. The round more than doubled the company's valuation since December, and it funds a factory built to turn out a hundred of these spacecraft a year.
The Gravitas platform is already in orbit with 12 hosted payloads, proving a 20-kilowatt spacecraft can launch, raise its own orbit and carry national-security hardware.
The open question is production. Whether space infrastructure can be built at industrial scale will be tested by a 100-satellite-a-year factory and a 100-kilowatt Giga platform.
The industry spent a decade shrinking satellites. Starlink runs on thousands of small craft. Startups built lighter, cheaper, faster. The founders, brothers Karan and Neel Kunjur, went the other way in 2022: build fewer, far more powerful spacecraft, and make enough of them to matter.
That bet is now measurable. The company was valued at $3 billion in December on a $250 million Series C. In July it raised $500 million at $6.8 billion, with CapitalG, Lightspeed, Altimeter, Spark Capital, Sands Capital and ARK Invest joining the two lead investors. The logic investors bought: power in orbit is a bottleneck, and the company is the only one selling it in volume.
Why bigger keeps winning in orbit
Almost every meaningful mission constraint in space reduces to two things: power and mass. A communications payload needs kilowatts. A radar aperture needs mass. A computing stack needs both. The small-satellite wave worked because early LEO applications — broadband, imaging, IoT — were built for constrained budgets. The next applications are not.
Its answer is a platform line that starts at the Mega class: up to 3,000 kilograms of payload and tens of kilowatts of power. Its Gravitas satellite, launched in March 2026, is roughly two metric tons and carries 12 payloads. It is running in orbit now, testing a dual-channel Hall-effect thruster the company calls the most powerful electric propulsion currently operating in space.
The ship has flown. The question is the shipyard.
Valuation after Series D
The company was valued at $3 billion in December 2025. The July round priced it at $6.8 billion, a 2.3× jump in seven months. SpaceNews, 2026
Gravitas proved the platform. Now comes the factory.
Its Torrance, California facility covers 180,000 square feet and is sized for 100 satellites a year. The company builds more than 85% of each platform in-house — the solar arrays, the reaction wheels, the avionics, the propulsion. That vertical integration is the entire investment thesis: buy less from a supply chain that never built anything this big, and control the cost curve yourself.
The round pays for scaling that factory and its workforce, now approaching 300 people. Next up is Trinity, a mission slated for the second quarter of 2027 that will put multiple spacecraft in different orbits and test long-range optical communications. It answers a simple question: can the same platform work in LEO, in MEO, in GEO?
Delivery dates are where space companies earn their reputations. Its own history is instructive. The company said Gravitas would fly in 2026; it launched in March. The first demonstration mission went up in January 2025. Nothing has slipped catastrophically. But the difference between one satellite and one hundred is not engineering. It is a production system that has not been built yet by anyone at this scale.
The contracts behind the valuation
The contract book now exceeds $1 billion in signed orders, roughly split between government and commercial customers. Three are worth naming.
Commercial anchor is SES, which tapped the company to build satellites for its meoSphere network in medium Earth orbit — an initial tranche of around 30 spacecraft, with the network eventually planned to grow to 100. Defence anchor is the Golden Dome space-based interceptor program, where the company supplies satellite platforms to an Anduril-led team. And in June, it was named satellite-bus provider for the Space Force's Protected Tactical Satcom-Global program, its first slot in a formal Pentagon program of record.
The SES deal, the Golden Dome team, the Pentagon program — each of those is a different customer type, and the company is serving all three with the same platform. That diversity is why the valuation jumped faster than the engineering did.
These are big numbers, but they're big numbers because we have big revenue numbers that are growing really fast.— Karan Kunjur, co-founder and CEO, K2 Space
Giga and the orbital compute question
The most consequential part of the roadmap is not a satellite at all. It is a market.
The next platform, Giga, is designed around 100 kilowatts of power per spacecraft, with introduction targeted for the second half of 2028. The company is explicit about why: orbital compute. A 100-kilowatt satellite carrying AI hardware is the enabling step for space-based data centers, a market SpaceX, Blue Origin and Axiom are all positioning for.
Whether that demand arrives is unproven. Compute in orbit carries costs: power generation, thermal management, radiation. Data still has to come down. The company's position is that the infrastructure must exist before the market does, and that being first movers on orbital compute is worth the production bet.
| Parameter | Mega class (Gravitas) | Giga class (planned) |
|---|---|---|
| Payload power | 20 kW on orbit | ~100 kW |
| Payload mass | up to 3,000 kg | not disclosed |
| First flight | March 2026 | target H2 2028 |
| Primary role | comms, sensors, hosted payloads | orbital compute, super-heavy lift |
What happens when a factory this ambitious meets reality?
Probability: 65% — the current factory peaks at that rate, Trinity only flies in Q2 2027, and Giga is not due until late 2028. Ramping a new platform to full rate within one year of first flight has no precedent at this power class.
✅ Arguments for a fast ramp
Contract revenue is booked and split across three customer types, so the factory has demand to fill.
Confirmation criteria: Trinity flies on schedule in Q2 2027, and SES converts its initial tranche into follow-on orders within the year.
❌ Arguments against a fast ramp
The valuation assumes orbital compute demand that has not been proven, and Giga's 2028 date depends on super-heavy lift availability.
Disconfirmation criteria: Giga slips past 2029, or a prime acquires the company at a discount before production scale is reached.
Gravitas completes its orbit-raising maneuver to MEO
Trinity launches on schedule in Q2 2027 across multiple orbits
SES converts the meoSphere initial tranche into follow-on orders
Giga-class platform unveils in H2 2028 as planned
Development scenarios
🟢 Optimistic scenario (30%)
Implications: The company raises at a double-digit-billion valuation and pulls the entire supply chain it built in-house along with it.
🟡 Base-case scenario (55%)
Implications: The company raises a Series E at $10–12 billion and becomes the reference supplier for high-power buses, without owning the orbital compute market yet.
🔴 Pessimistic scenario (15%)
Implications: The valuation re-rates down, or a defence prime acquires the company at a discount to lock up its high-power platform.
As we wrote in July, when Swiss startup SWISSto12 raised $70 million to scale 3D-printed small GEO satellites, the manufacturing side of this industry is where the value is concentrating. The company is the same story in the opposite direction — a bet that the biggest platforms, not the smallest, are the scarce resource. One of those two theses will be wrong. Production data over the next two years will say which.
