Phantom Space reportedly acquired thermal management hardware provider TMT in early April, the latest move in the Tucson-based company’s effort to build a vertically integrated space infrastructure business capable of computing in orbit. The deal directly addresses one of the hardest engineering problems facing anyone who wants to run high-performance computing workloads above the atmosphere: getting rid of the heat.

TMT, a company specializing in satellite thermal control systems including radiators designed to dissipate heat from electronics in the vacuum of space, operates office and laboratory space in North Logan, Utah. For Phantom Space, which is reportedly building a satellite constellation called Phantom Cloud, the acquisition fills a specific technical gap that would otherwise require relying on outside suppliers for a mission-critical subsystem.

The Thermal Problem Nobody Wants to Talk About

Heat dissipation in orbit is a fundamentally different challenge than on Earth. On the ground, data centers push hot air through cooling towers and chillers, using convection and sometimes nearby rivers or lakes. In space, there is no air. The only way to shed heat is through radiation, which requires large surface-area components engineered to emit infrared energy into the cold of space. The higher the computing load, the more heat you generate, and the bigger your thermal management problem becomes.

This is where TMT’s expertise matters. Phantom Space CEO Jim Cantrell has described the company’s technology as critical for the constellation’s development, noting TMT’s expertise in satellite thermal components. TMT’s founder, Scott Schick, has stated that joining Phantom brings together complementary strengths that will expand technical capabilities and accelerate hardware development.

Any company serious about orbital computing will eventually have to solve this problem. Phantom is trying to solve it by owning the solution.

Vertical Integration as Strategy

The TMT acquisition fits a pattern. Phantom Space has been on an acquisition spree for several years, buying companies that fill different layers of the space infrastructure stack. The company acquired StratSpace in 2021 to expand its services portfolio. It has also reportedly purchased Vector Launch assets. Now TMT adds thermal hardware.

The logic is straightforward. If you’re building a constellation, a launch vehicle, and an orbital computing service, you don’t want to be at the mercy of component suppliers who might prioritize larger customers or raise prices. You want to own the critical technology. SpaceX proved this model works at scale with Starlink, manufacturing everything from the satellite bus to the phased array antennas in-house.

Phantom is much smaller than SpaceX. The company has raised tens of millions of dollars. Compare that with the billions flowing into orbital computing from larger players. Vertical integration is a sensible strategy for a company with limited capital, but only if the acquired companies actually deliver the performance needed at the system level.

Phantom Cloud’s Position in the Orbital Computing Race

Phantom Cloud is reportedly designed as a satellite constellation focused on data-backhaul and edge-processing. Cantrell has framed the ambition in broad terms, describing how the service will unlock a new generation of data-driven applications in space.

But the positioning is actually narrower and more focused than what some competitors are attempting. SpaceX has talked about large-scale orbital computing platforms. The broader industry is buzzing with proposals for full AI inference in orbit. Phantom Cloud appears to be targeting something more specific: processing and relaying data at the edge, closer to where sensors and instruments are collecting it, rather than trying to replicate a terrestrial hyperscale data center in space.

That narrower scope may be smart. The power requirements for running full AI workloads in orbit are enormous and, as of now, largely unsolved at scale.

The Power Problem Hanging Over Orbital Computing

A recent SpaceNews opinion piece by Oleg Demidov of Beyond Earth Ventures made a blunt argument: the race to build orbital data centers is missing its biggest variable, which is power. While companies announce compute architectures and satellite constellations, almost nobody is seriously addressing where the electricity comes from to run those workloads.

On Earth, this same bottleneck is already constraining the AI buildout. Industry analysts estimate that trillions of dollars in data center investments will be required by 2030. Energy agencies project that data center power consumption could double by 2030, potentially reaching hundreds of terawatt-hours annually. Terrestrial hyperscalers are signing multi-gigawatt nuclear power purchase agreements and building their own reactors because the existing grid can’t keep up.

In orbit, the physics of solar power are more favorable. A solar panel in space produces roughly five times the electricity of the same panel on Earth, with no atmosphere, no weather, and no day-night cycle for most orbits. But converting that advantage into reliable, high-density power for computing hardware requires engineering that doesn’t exist at the scales being discussed.

Phantom Cloud’s edge-processing approach sidesteps some of this problem. Edge computing workloads are typically lighter than full AI training or large-scale inference. A satellite constellation doing data backhaul doesn’t need the power density that a true orbital data center would. But it still needs thermal management, which circles back to why Phantom bought TMT.

Timeline Questions

The most conspicuous detail in Phantom’s story is the timeline. The company has experienced delays in its development schedule, with current targets for both rocket launch and initial Phantom Cloud deployment now reportedly set for the second half of 2027.

Significant delays are not unusual for small launch companies, and Phantom is trying to build both a launch vehicle and a satellite constellation simultaneously. But it raises obvious questions about execution capability and whether the capital raised to date is sufficient to reach those milestones.

For context, Phantom had also announced plans in 2021 to build and launch a 72-satellite constellation for Ingenu. It’s not entirely clear how these constellation projects relate to each other or whether the Ingenu contract is still active.

The company raised seed funding in 2021, and the ambition has grown considerably since then. Building a launch vehicle alone typically costs hundreds of millions of dollars. Building a constellation on top of that, while also acquiring companies, requires either remarkable capital efficiency or additional funding rounds that haven’t been publicly announced.

Where Orbital Computing Actually Stands

The broader orbital computing industry is in a strange phase. The idea has tremendous momentum. Nvidia has been developing more efficient computing architectures, with significant improvements in performance per watt that matter enormously for space applications where every watt counts.

Industry observers have noted that space could become an increasingly attractive location for computing infrastructure as efficiency improvements continue. These are real signals from real companies with real resources.

But signals and operational systems are different things. The gap between announcing an orbital computing concept and actually running production workloads on satellites in orbit is vast. Thermal management, power generation, radiation hardening, latency management, and orbital mechanics all impose constraints that don’t exist for a data center in Iowa.

Phantom Space is attacking one of those constraints directly with the TMT acquisition. Whether the company can solve the rest of them on a startup budget and a delayed timeline is the central question.

What the TMT Deal Actually Tells Us

Acquisitions like this are useful signals precisely because they’re boring. Nobody buys a thermal radiator company for the press release. You buy it because you’ve identified a specific technical bottleneck and you want to own the solution rather than depend on someone else to provide it.

TMT’s years of experience in satellite thermal control represents real institutional knowledge. Thermal engineering for space is specialized work. The number of companies that do it well is small. For Phantom, bringing that capability in-house removes a dependency and gives the company more control over system-level design trades.

The risk is integration. Acquiring a small specialized firm is one thing. Getting its technology to work seamlessly within a new satellite architecture, on a tight schedule, with limited funding, is something else entirely.

Phantom Space is making a bet that vertical integration will give it a competitive edge in a market where larger, better-funded competitors are taking different approaches. The TMT acquisition is a coherent move within that strategy. But coherent strategy and successful execution are separated by a large and unforgiving gap, and in the space industry, that gap is measured in years, dollars, and hardware that either works in orbit or doesn’t.

Photo by Ray Strassburger on Pexels