No, data centers in space aren't Science Fiction. That doesn’t mean it’s that useful to you, either

You may have noticed a flurry of articles in the last few weeks which all ran variations of the following headline: ‘Data centers in space are no longer science fiction.’ 

Short answer, no, they’re not—but that also doesn’t mean you should be looking to wind down your Northern Virginia or Docklands data center lease quite yet.

The context: last year, Amazon founder Jeff Bezos predicted orbital facilities could be built at scale within 10 years, but “for sure” in under 20, and at least one source says his company is actively pursuing the plan. 

Not to be outdone in the cool stakes, SpaceX’s CEO Elon Musk has claimed that satellites with localized AI compute, where just the results are beamed back from low-latency, sun-synchronous orbit, “will be the lowest cost way to generate AI bitstreams in <3 years” - amplifying his idea at last month’s Davos, where he claimed that “building solar-power data centers in space” was a “no-brainer,” and inside two to three years “at the latest.”

The usual Musk bluster (after all, we’re still kind of waiting on all those self-driving cars). But it turns out SpaceX is far from the only US company claiming to be actively pursuing this idea. In November 2025, Google announced Project Suncatcher, for example, which it says will explore how an interconnected network of solar-powered satellites, equipped with its Tensor Processing Unit (TPU) AI chips, “could harness the full power of the Sun” up there in LEO (Low Earth Orbit); working with a partner called Planet, it plans to launch two prototype satellites by early 2027 to test its hardware above the atmosphere.

A growing cast of characters

Smaller-scale players are also in the orbital data center mix. A space data center start‑up called Starcloud (previously, Lumen Orbit) is proposing satellite clusters with central racks of AI chips surrounded by literally miles of solar arrays as way to “build large data centers in space that are much more rapidly scalable and cost competitive than terrestrial data centers.” Other contenders include Aetherflux, which says it wants to build “an American power grid in space” named, er, the Galactic Brain.

All of this might have been dismissed as tech bro grandstanding (and some Science Fiction fanboy dreaming), when at the end of January, AI, industrial and space player China said it was going this way too.

Specifically, its state broadcaster, CCTV, announced plans for its own “gigawatt-class space solar-powered hubs” to create an industrial-scale "Space Cloud" by 2030. A Reuters summary of the announcement says the aim is an integration of cloud, edge and terminal (device) capabilities" to achieve the "deep integration of computing power, storage capacity and transmission bandwidth.”

That would enable data from Earth to be processed in space, while China also plans to shift the energy-intensive burden of AI processing into orbit - with the integration of space-based solar power with AI computing as a core pillar of the country’s upcoming 15th national Five-Year Plan.

It’s not (brace yourself for the pun) rocket science to see why it would be great if we could use all that near‑constant sunlight for free (solar) juice; we’d also have no need for water or other cooling and all the environmental concerns to do all that power-intensive AI training that has raised, as we’d have a naturally radiative environment that could simplify cooling. 

It would arguably also be more cost-effective. At Davos, Musk said solar generation in orbit can produce five times more power than panels on the ground - and as rising AI workloads are expected to push global data-centre power demand sharply higher by 2030, intensifying pressure on national grids, freshwater supplies and urban real estate, that is a factor that could make this happen sooner rather than later. There is also the potential for not just energy and cooling advantages but the possible emergence of new classes of applications out of computing in Zero G/the vacuum of space, as Michel Muizer, Director of Product at Expereo, a global provider of managed network solutions, predicts:

Moving above the atmosphere removes the limits of land and power grids, replacing them with sunlight for energy and a vacuum for cooling. This shift turns physics into an advantage, allowing enterprises to scale at the speed of launch rather than the pace of construction, and as latency stretches from continental to planetary, enterprise behavior will change dramatically; real-time global inference becomes possible, automation can operate without geographical constraints, and continuous climate modelling moves from theory to practice.

The high frontier. Possibly, too high

Great stuff - but before we go any further, it just has to be said that this is all, shall we be polite, a way off. Tl/dr for why: we still have a launch cost per kilogram of physically getting things into the sky, you’d have to massively harden all your components against radiation, we’d need ways we don’t know yet on how to deliver reliable long‑term cooling in vacuum via radiators rather than air, there’s the slight ongoing challenge of replacing hardware periodically, there’s a heck of a lot of space junk up there that could prove inconvenient if it crosses your space array, don’t even ask about all the regulatory space law that needs to be worked out, and so on. At least some skeptics, like Stewart Laing, CEO of Asanti Data Centres, also point out that we still have some problems to work out with terrestrial data centres, let alone stellar ones:

From where most CIOs sit, this feels like AI bubble chatter on steroids, and before we export our technical debt into space, we would be better off fixing the basics on the ground. The real work in 2026 should be on greener power, more regional colocation in the North and Scotland, and a sober look at which AI projects actually earn their keep. Until then, most organizations need less ‘final frontier’ and more ‘fix the fundamentals.’

Aviad Algamor, VP of Technology Innovation at US technology vendor Trimble, adds:

Space will almost certainly factor into the computer landscape in the future, and big, ambitious ideas are essential to move the industry forward. Right now, however, the urgent work is here on Earth: keeping up with accelerating demand, improving efficiency, shortening delivery times, and addressing constraints around labour, power, and sustainability. 

Solving these challenges is what will lay the groundwork for whatever comes next—including the possibility of extra-terrestrial data centres.

There is also the slight (but inconveniently real-world) issue of how you’d reliably get all that power and data back down to your HQ, as very high frequencies such as Q/V-band and optical links are highly sensitive to weather and atmospheric conditions, and so on. 

Nonetheless, while significant technical hurdles remain and implementation is still several years away, is this an approach that could eventually offer an effective way to achieve all the benefits of space-based computation? Should a practical CIO be putting any of this on their ‘watch and see’ pile?

A pretty positive vote in favour comes from, of all possible stakeholders, the lawyers. Hayley Blyth is a space expert at international law firm Bird & Bird:

Though skepticism remains around technical challenges and economic viability, orbital computing and data centre technology are gaining mainstream momentum. Early commercial traction will likely centre on high-value niches such as defence, edge AI for rapid imagery analysis, and secure backup, with progress likely to be industry-led given tempered public sector support due to viability concerns. 

Businesses should monitor regulatory shifts, invest in adaptable architectures, and align risk management with evolving regulatory frameworks.

This may not be relevant to you for quite some time

In other words: yes, this is coming and is worth opening a watching brief on. But the last word may be best offered by Tom Winstanley, CTO and Head of New Ventures at the UK & Ireland IT services arm of Japanese engineering and comms giant NTT, NTT Data—which is already deadly serious about using space for IT, but under today’s constraints:

We have many projects going on right now involving space, which is why we have put our space and energy teams together as one, joined-up research division. But even though there are companies out there that are talking about doing it at scale, you’re just not going to put your data center in space in the next couple of years.

For sure, there are logical, worthwhile use cases for the use of data at the edge in space, Earth observation being one key area for that. If you're in defense, if you're in maritime, if you're in insurance, if you're in the telecommunications industry, there are relevant reasons why you would be doing this; if you're a communications provider, space ICT is very real today, Starlink and lots of others already have micro data centers as part of what they are doing.

In the midterm we could well see an interconnected space ICT infrastructure which would be complementary to the communications and data processing infrastructure on Earth. But at least for today, the focus must be on use cases where space is already a relevant component of what you're seeking to achieve. 

If you’re a bank, data centers in space are probably not the next thing that you're going to need to be worried about.

My Take

Spoilsport. 

But he’s right; something will happen here - but as a Boomer space fan myself, it’s probably on the same timescale as Moon bases or asteroid mining. And on the disappointing front, we might as well also share now that if you have any romantic notions of applying for the work, sorry to disabuse you, but the direction of travel is to use a lot of robots to build a lot of modular, self‑assembling structures - so no Space Jockeys need apply.

That doesn’t mean - as with the 1960s Space race itself - that investment and R&D in all this stuff won’t return potentially hugely useful ideas and breakthroughs for core IT and data center operations and engineering.

So, let’s not get too excited… at least, not until the Chinese invite us to the champagne opening of the first working Data Center in the Sky, maybe?