Headline & intro
SpaceX is preparing one of the most ambitious tech IPOs ever, and its latest big idea sounds like pure science fiction: data centers in orbit. In an era where AI models are devouring electricity and water, shifting compute to space offers an irresistible narrative. Clean solar power above the clouds, no angry local residents, and a brand‑new infrastructure category that only a handful of companies can touch.
But is this a genuine new industry or a clever way to hang a 1.75 trillion dollar valuation on a rocket company and its satellite network? In this piece we look past the pitch deck and unpack the technical, economic and geopolitical realities behind orbital data centers.
The news in brief
According to TechCrunch, SpaceX has confidentially filed for an IPO that would raise around 75 billion dollars at a valuation of about 1.75 trillion dollars. As part of the story it presents to investors, CEO Elon Musk has been talking up a long‑term plan to build orbital data centers using satellite constellations derived from its existing Starlink infrastructure.
On TechCrunch’s Equity podcast, the hosts note that SpaceX is not alone. A Y Combinator‑born startup originally known as Starcloud has just raised roughly 170 million dollars, pushing it into unicorn territory with a similar orbital data center vision. Jeff Bezos, via Amazon’s satellite work and Blue Origin, is also circling the same idea. TechCrunch’s team stresses that this will require major engineering breakthroughs and huge capital expenditure, and that even the optimistic scenarios frame orbital compute as a supplement, not a replacement, for terrestrial data centers.
Why this matters
The most obvious reason this matters is financial: at 1.75 trillion dollars, SpaceX would debut in the same valuation stratosphere as the largest US tech giants, despite far lower current revenues and margins. To justify that, the company needs more than rockets and residential internet. It needs a story about owning a critical layer of tomorrow’s AI infrastructure.
Orbital data centers serve that purpose beautifully. They extend the logic of Starlink: build a vertically integrated stack where SpaceX designs the satellites, launches them, runs the network, and now also hosts high‑value compute workloads. Every new satellite launched to power these data centers is both infrastructure and internal customer, generating additional launch revenue that flatters the P&L of the core business.
The pitch also taps into a real pain point. Hyperscale data centers have become politically toxic in many regions. They soak up power and water, strain grids, and trigger local opposition. As TechCrunch’s conversation highlights, some executives may genuinely believe that solving orbital thermal management and reliability is easier than fighting zoning battles and public protests on the ground.
Yet the likely winners in the near term are not AI users or ordinary internet customers. The main beneficiary is SpaceX itself, which can use this vision to support its valuation, drive satellite and launch volume, and lock in strategic relevance. The losers could be regulators and communities, who suddenly face a future where the dirtiest part of AI infrastructure is simply moved out of sight – into an orbital shell where rules are fuzzier and externalities are easier to ignore.
The bigger picture
Orbital data centers sit at the intersection of three powerful trends.
First, the AI compute arms race. Training frontier models already consumes enormous amounts of power and specialised chips. Cloud providers and AI labs are scrambling for land, electricity and cooling. This fuels experimentation at the edge: underwater data centers, facilities colocated with renewable plants, and now orbital concepts that promise essentially limitless solar power and very cold deep space as a heat sink.
Second, the growing backlash against data centers on Earth. In many countries, local communities are pushing back against new facilities due to noise, visual impact, power usage and water consumption for cooling. TechCrunch notes that some industry players are starting to reassess just how much capacity they truly need, rather than blindly follow exponential projections. Against that backdrop, the idea of putting the most controversial infrastructure in orbit is politically seductive.
Third, the long‑running game of narrative arbitrage in big tech. Whenever valuations stretch far beyond current cash flows, companies reach for audacious visions: Google’s moonshots, Meta’s metaverse, Tesla’s robotaxis. Some of these bets pay off partially; many quietly shrink. Orbital data centers feel similar. They are not pure fantasy – the physics is hard but not impossible – yet the timelines are long and the path from demo to scalable business is murky.
Compared to competitors, SpaceX has an edge and a handicap. The edge is obvious: it is already the dominant low‑cost launch provider and operates the world’s largest communications constellation. It can iterate in orbit faster and more cheaply than anyone else. The handicap is regulatory and customer trust. Enterprises already struggle with cloud compliance across countries. Moving sensitive workloads into a privately controlled orbital platform raises new questions about jurisdiction, liability and resilience that players like Amazon and Microsoft, with deep enterprise relationships, may be better positioned to navigate.
The European and regional angle
For Europe, orbital data centers cut right across two strategic anxieties: digital sovereignty and space competitiveness.
The EU has spent years trying to reduce dependence on US cloud giants through initiatives like GAIA‑X, the Data Act and stricter GDPR enforcement. At the same time, Europe’s launch sector has been under pressure, while SpaceX has taken a dominant global role. If orbital compute becomes a real market, there is a real risk that it is defined, priced and governed largely from California and Texas.
Jurisdiction is another thorny issue. If European personal data is processed in orbit by a US‑controlled constellation, which law applies? Technically the data will still traverse European ground stations, but the line between transmission and processing becomes blurred when satellites themselves host storage and AI accelerators. Regulators will need to decide whether orbital data centers are treated like any other cloud region or as a new category of infrastructure with additional obligations.
Europe is not starting from zero. Companies like SES, Eutelsat OneWeb and a growing cohort of NewSpace startups already operate constellations and ground segments. The EU’s planned IRIS² secure connectivity programme could, in theory, evolve to include sovereign orbital compute capacity. But that would require political will, long‑term funding and a willingness to move at commercial speed, not traditional aerospace tempo.
For European enterprises and governments, the practical question is simpler: do they trust critical AI workloads to orbit controlled by non‑European entities, or do they insist that any such service be anchored in EU law, with strong guarantees on data locality, encryption and auditability?
Looking ahead
Over the next three to five years, the most likely outcome is not a global shift of cloud workloads to orbit, but a series of increasingly ambitious demonstrations. Expect prototype satellites with modest onboard compute, used for specialised tasks: preprocessing Earth observation data, running AI models for low‑latency decision‑making in remote regions, or serving niche defence and scientific use cases.
The economics will be decisive. Even with reusable rockets, putting hardware into orbit costs orders of magnitude more than deploying it in a terrestrial data center. Cooling in vacuum is radiative, not convective, which complicates design. Radiation hardening, in‑orbit maintenance and higher failure rates will add to lifecycle costs. For orbital data centers to make sense, they will need clear advantages: either workloads that benefit dramatically from proximity to space (for example, processing sensor data before downlink) or customers willing to pay a high premium for isolation and resilience.
From a market perspective, the bigger near‑term impact is on capital flows and regulation. If SpaceX successfully sells this vision to public investors, we will see more startups and incumbents racing to position themselves in the orbital compute stack: chip designers, thermal specialists, debris mitigation companies, space insurance, and of course European launch and satellite firms looking for a slice of the pie.
Key signals to watch include: concrete technical roadmaps from SpaceX and rivals; pilot customers beyond defence; emerging standards for security and data protection in orbit; and early regulatory moves from the EU and national authorities on jurisdiction and environmental impact.
The bottom line
Orbital data centers are not pure hype, but they are far from a near‑term solution to the AI compute crunch. For now, they function primarily as a powerful IPO narrative for SpaceX and a way to channel more satellites – and revenue – through its launch machine. Investors and policymakers should treat them as long‑dated options, not core assumptions in any serious forecast. The real question is whether we learn from past tech bubbles and demand evidence of sustainable economics before rewarding yet another grand space‑age story.
