K2 Space Raises $500M on the Strength of a 20 kW Thruster

$500 million. K2 Space closed a Series D of that size on July 30.

The round more than doubled the startup's valuation to $6.8 billion in seven months. The money funds a production ramp toward 100 large spacecraft a year out of a Torrance, California factory.

K2 Space was founded in 2022 by former SpaceX engineers with a contrarian premise: bigger satellites, not smaller ones, would define the next phase of space infrastructure. For most of its short history, that premise lived in slide decks. Now there is hardware in orbit, a record-setting thruster, and a valuation curve that investors are betting will hold.

TIMELINE: K2 Space
─────────────────────────────────────────────────────────────────
  2022 ──── 2025 ──── 2026 ──── 2026 ──── 2027 ──── 2028
  Founded   Demo      Gravitas  Series D  Trinity   Giga
  🚀        🧪        🛰️        💰        🚀        ⚡
  Co.      1st in-   20kW thr  $500M    3 sats    100kW
  started   space      in orbit  at $6.8B  LEO/MEO/  platform
           flight                          GTO       target

The engine did the talking

Gravitas, K2's first full satellite, launched in March 2026 on a SpaceX Falcon 9 rideshare. It carries a 20 kW krypton-fed Hall-effect thruster, and that number matters more than the funding round.

Before Gravitas, the most powerful electric propulsion system flown on an operational mission was the Aerojet BPT-4000 at 4.5 kW, which entered service in 2010. K2's unit runs at more than four times that power. It lets a satellite raise its own orbit from low Earth orbit to medium Earth orbit entirely under electric power, in under 90 days, freeing the mass budget that a chemical maneuver would consume.

That is the difference between a satellite that carries fuel and a satellite that carries payload. Hall-effect thrusters reach roughly 1,500 seconds of specific impulse versus 300 to 450 seconds for chemical rockets, so a given kilogram of propellant does far more work. At 20 kW, the arithmetic starts to reshape constellation economics: fewer, higher satellites can cover the same ground as many more low-orbit ones.

The engine was ground-tested in July 2025. Flight validation came this year. The gap is not incremental.

The factory math

K2 builds more than 85% of its platform in-house: flight software, avionics, propulsion, power systems, and solar arrays. Its 180,000-square-foot Torrance plant is designed to produce up to 100 satellites a year. The company says the economics follow directly.

The supply chain argument is real. Components for a high-power satellite at that price point barely existed as off-the-shelf purchases when K2 started, so the company built its own manufacturing lines instead of waiting for the market to catch up.

As we wrote in August, Muon Space's $250M Series C is betting on the same constellation-factory logic from the other direction, scaling smaller satellites in higher volume. K2 is the mirror image: fewer, larger spacecraft, vertically integrated in-house.

A backlog that matches the raise

SES, one of the world's largest commercial satellite operators, contracted K2 in March 2026 to build an initial 28 satellites for meoSphere, its medium-Earth-orbit broadband network, which is expected to grow toward 100 spacecraft at roughly 8,000 km altitude. Each satellite delivers Ka-band broadband at up to 1 gigabit per second with optical inter-satellite links rated at up to 100 Gbps. This is the volume order that turns K2 from demonstrator into supplier.

Then there is government work. K2 was selected as bus provider for the Space Force's Protected Tactical Satcom-Global program, its first Pentagon program of record, on a team led by SES Space & Defense. And under the Golden Dome missile-defense architecture, K2 provides satellites to Anduril's interceptor program.

K2 started from a simple insight: in orbit, power determines capability.— Tengbo Li, General Partner, ICONIQ

The pattern is deliberate. High power and high payload mass make a satellite useful to both commercial broadband operators and defense customers, so K2 is building one platform and selling it into both markets rather than maintaining separate product lines.

The production test

None of this validates manufacturing at scale yet. The industry has a documented record of stumbling at exactly this point. OneWeb and Airbus hit supply-chain gaps the moment production moved from prototypes to two satellites a day. K2 is aiming for a different number, up to 100 large spacecraft a year rather than thousands of small ones, but the organizational discipline required is the same.

Trinity deploys three satellites into three different orbital regimes from a single Falcon 9: one climbs from LEO to MEO, another goes straight to MEO, and the third heads to geostationary transfer orbit. It is the first test of whether one platform architecture can be built and flown repeatedly at production quality.

Where the buses go next

The longer-term target is orbital compute. K2's planned Giga platform, rated at 100 kW and scheduled for the second half of 2028, would be five times Gravitas's output and roughly the power budget of a small data center node in orbit. Co-founder and CTO Neel Kunjur has said hyperscalers are the obvious customers for that class of satellite.

The market context is filling in around the bet. SpaceX filed FCC plans in January 2026 for a constellation of solar-powered orbital data centers in LEO and unveiled a 150 kW orbital compute satellite in June. K2's position is narrower and arguably cleaner: it does not plan to run its own constellation. It builds the buses other constellations use.

The Series D funds the ramp, not the thesis. The thesis was proven by a 20 kW thruster flying. What remains is whether a four-year-old company can now become a repeatable manufacturer before the backlog turns into delivery obligations.

One satellite is on orbit and performing. The next two dozen will tell the real story.

Sources