Will we really put data centers in space?

Avi Parrack; fin

Abstract

Several major technology companies have announced plans to operate AI data centers in orbit. Elon Musk recently claimed: “the lowest-cost place to put AI will be space […] within two years, maybe three.” If a meaningful fraction of new AI compute really is placed in space within a few years, that would be a fairly big deal for AI governance and strategy. Here we try to disentangle the hype from reality and provide a sober assessment of the technical and economic feasibility of orbital data centers (ODCs).

The main case for ODCs is the cost of energy: space solar panels in the right orbits receive more constant and intense sunlight compared to Earth. Moreover, ODCs don’t currently face the same permitting and regulatory delays as on Earth, cause fewer ongoing environmental harms compared to grid or onsite natural gas-powered data centers, and may be more secure against data exfiltration. We find that the cost-competitiveness case for ODCs depends almost entirely on Starship achieving reusability comparable with what SpaceX achieved with Falcon: space-based solar reaches cost parity with present-day off-grid terrestrial power continuously at roughly $250/kg to orbit, and becomes cheaper than any current terrestrial energy source at around $50/kg, from the present-day launch cost of roughly $1,500/kg. Radiative cooling, often cited as a fatal obstacle, appears surprisingly manageable — potentially even cheaper than on Earth. However, ODCs may require substantial (perhaps ~38%) extra non-compute hardware (like solar, racks, and cooling) over 5 years to compensate for their inability to swap out failed chips, and inter-satellite bandwidth limitations likely confine ODCs to inference workloads, at least early on.

Assuming no transformative AI,^[1] but continued demand for data center buildout, we estimate that ODCs are unlikely to represent a meaningful share of compute before 2030, but become cost-competitive with present-day terrestrial data centers within 3–5 years if Starship development stays on track.

Introduction & Takeaways

Some of the world’s largest technology companies continue racing for compute. If progress continues, demand for data centers may more than double by 2030.^[2] Increasingly, though, new data center capacity is bottlenecked by multi-year queues to connect to the power grid.^[3]

The result has been a scramble for workarounds. Leading AI labs have increasingly adopted a “Bring Your Own Generation” model to source power, deploying onsite gas turbines and engines to bypass grid bottlenecks. xAI, for example, reportedly installed hundreds of megawatts of onsite gas generation in Memphis to accelerate deployment, and OpenAI and Oracle have placed large turbine orders for new Texas campuses.

Some argue that energy will become the binding constraint on AI progress, given grid interconnection delays as gas turbines are themselves facing multi-year manufacturing backlogs. But the constraint does not appear fundamentally binding (as Epoch notes): turbine manufacture may expand to meet more demand and companies could go off-grid using combinations of gas, solar, and batteries, scaling power in parallel with compute, albeit at a cost premium. This raises a natural question: if you’re going off-grid anyway, then what’s the best way to get power and where is the best place to put your data center?

Some think the answer will be in orbit. In November 2025, Google announced Project Suncatcher, a plan to put TPU-equipped satellites in dawn-dusk sun-synchronous orbit. In early 2026, SpaceX filed with the FCC for authorization to launch and operate a constellation of up to one million data center satellites.^[4] Other entrants include Blue Origin, Ramon.Space and startups like Starcloud, and Aetherflux while China’s Three-Body Computing Constellation has launched 12 operational satellites and run Alibaba’s Qwen3 model in orbit. Recently, at GTC in March 2026, NVIDIA announced the Space-1 Vera Rubin Module, meant to be a dedicated space-rated GPU platform.

At first glance, it seems very unlikely that any meaningful fraction (say, >10%) of additional data center capacity will be placed in space in the next few years. But if the companies betting on space are right, that would be a fairly big deal, and it could change the landscape of AI governance. For example, terrestrial data centers are subject to national and regional regulations, whereas AI developers could potentially exploit jurisdictional ambiguities around compute in space. Also, the path to low-cost orbital compute likely routes through a single launch company, SpaceX, which also now operates a frontier AI lab since its acquisition of xAI. And that might raise concerns around concentration of power.

We’ve been looking into the technical and economic viability of orbital data centers (ODCs). Our core model gives estimates for the total cost of Earth and space-based data centers across several scenarios.

The report focuses on three questions. First, what is the basic economic case for a meaningful fraction of AI compute being placed in space? Second, the most obvious physical blocker: can you cheaply cool a data center in orbit? Third: how fast could the shift to space data centers happen, how soon, and what would have to go right?

Here is our provisional assessment:

SpaceX’s Starship is the only vehicle currently on track to deliver the launch costs and cadence that meaningfully scaling orbital data centers would require. Competitors are years behind, making SpaceX’s Starship the only near-term path to large-scale orbital compute. SpaceX aims to complete Starship development by late 2026, with several necessary milestones still ahead. If development stays roughly on track, Starship could plausibly hit the cost and cadence required to scale meaningful orbital compute within 3–5 years. However, chip production may become the limiting factor by this point, rather than launch capacity.
The cooling problem is more tractable than commonly assumed. Passive radiators using selective coatings and lightweight carbon fibre panels could achieve ~170–360 W/kg at system level, a 13-28× improvement over ISS-era radiators (~13 W/kg).^[5] No radiator at these performance levels has been deployed at the scale an orbital data center would require, but prototype high-conductivity carbon composite panels have demonstrated the material properties required. At these performance levels, thermal hardware is 2-5% of total data center cost, and actually less than what terrestrial data centers spend on cooling over a comparable lifecycle.
If launch costs fall enough, the unit economics could favor space. Solar panels in dawn-dusk sun-synchronous orbit produce roughly 3–5× the energy of the same panel at a good terrestrial site.^[6] Space-based solar becomes cheaper than the best off-grid terrestrial installations once launch costs drop below roughly $250/kg using Starlink-like solar arrays. At a launch cost of $50/kg (corresponding perhaps, to a Starship with full reuse as reliable as Falcon), space solar could fall to between $25–45/MWh, making it cheaper than any current terrestrial option available today.^[7] Beyond the symmetric cost of chips, launch cost is the dominant line item for ODCs while power and op-ex dominate terrestrial costs but would be near zero in space.
The inability to do maintenance would be a large cost. Chips often fail and are swapped out in today’s data centers but a dead chip in an ODC would remain dead, wasting the parts of the supporting infrastructure (power, cooling) and diminishing overall compute. We model this below as a 9% annual bleed causing about 40% overbuy of launch and non-chip hardware over the data center’s lifetime.^[8] Below $100/kg launch cost this might net out against other savings from ODCs but this is a significant uncertainty since the actual rates of chip failure for ODCs could be higher or lower.
All-things-considered we think that, absent transformative AI, orbital data centers probably won’t make up a meaningful fraction of compute before 2030, but it’s credible that space could house much or even the majority of compute buildout throughout the 2030s.

Read the full report on the Forethought website: Will We Really Put Data Centers in Space?

^{^}
We hope to do more analysis on how transformative AI might change this picture in the future. Speculatively, our initial thinking is TAI could accelerate the timeline over which compute transitions to space but this is not necessarily the case. In particular, during an industrial explosion pressure to grow rapidly might be so strong as to incentivize aggressive usage of non-renewables on Earth like oil and gas. If so, transition to space might be delayed for a one time boost on Earth, in which case the picture may look similar to the one we outline here, but with the added prologue of a large-scale terrestrial buildout.
^{^}
McKinsey projects demand growing to 171–219 GW by 2030, roughly doubling from today, in a buildout they estimate will require up to $7 trillion in investment.
^{^}
See, directing FERC to address large-load interconnection (2025), Reuters, Google says US transmission system is biggest challenge for connecting data centers (2026), Bain & Co., Next phase of data center growth (2025)
^{^}
Starcloud subsequently filed for authorization to operate 88,000 satellites and Blue Origin has filed for 51,600.
^{^}
The ISS External Active Thermal Control System achieves roughly 13 W/kg. Our improvement comes from three sources: selective coatings (high emissivity, low solar absorptivity, off-the-shelf AZ-93 paint), carbon fibre composite construction (2.4 kg/m² vs ISS’s ~14-17 kg/m²), and optimised operating temperature (40°C vs ISS’s -40°C, exploiting the T⁴ dependence). Each factor is independently demonstrated; their combination at scale is not.
^{^}
The solar constant at Earth’s orbit is approximately 1361 Wm^-2. A solar panel in a dawn–dusk sun-synchronous orbit receives nearly continuous illumination (capacity factor ≈ 90–95%), yielding an average power of roughly 1220–1290 Wm^-2 before panel efficiency losses. By contrast, even excellent terrestrial solar sites typically achieve ~20–30% capacity factors due to night, weather, and atmospheric attenuation, corresponding to an average incident power of roughly 270–410 Wm^-2. Thus, a panel in a dawn–dusk orbit produces roughly 3–5× more energy annually than the same panel on Earth.
^{^}
This wouldn’t be true if you were then beaming the energy back to Earth, but would apply to orbital compute, where only data needs to be sent to Earth.
^{^}
Both terrestrial data centers and ODCs will pay symmetric costs to replace dead chips but ODCs would have to pay the additional cost from lost overhead, i.e. in the earthbound case a technician swaps the dead chips, in the space case you launch entire additional satellites to compensate for chip bleed. We assume you would not send a mission to do maintenance and instead simply let the excess power and cooling go to waste doing no useful compute. Extra power and cooling over fewer chips may increase operating efficiency somewhat but this seems fairly negligible. The figure for chip bleed of ~9% per year is derived from Meta’s The Llama 3 Herd of Models (2024). We cover radiation and other forms of damage in more detail subsequently.

The main case for ODCs is the cost of energy:

I suspect one motivation for orbital datacenters is that you get to do whatever business you want with anyone on earth, and nobody can stop you without a military strike (and the risk of Kessler syndrome). This is a position of significant political power; it is a castle on top of the biggest hill. It is a position that allows the operator to circumvent quite a lot of earthly laws — including AI regulation.

Sure: I agree that if other actors fail to negotiate for control over an ODC before you launch it, then they would need anti-satellite capability to forcibly shut it down. But presumably there are plenty of standard sources of leverage the ODC operator, e.g. fines or sanctions on the company, or legal action against individual principals.

In fact, there's a sense in which ODCs (unlike castles) are extremely vulnerable: you can't afford to reinforce them, you can't easily repair them, and (importantly) the human and political costs of destroying them seem lower than on Earth (no possibility of collateral damage in the form of human life, far less collateral damage to surrounding infrastructure, no possibility of hostage-taking). And they are more vulnerable to states and sophisticated private actors, but not mercenaries or lone actors. I agree space debris makes things less clean, though directed energy weapons might make it possible to disable the ODC without blowing it up. I feel antsy talking about such things because I think our best governance mechanisms won't involve physical destruction, but as far as the threat of physical destruction is a backstop, then I think ODCs in space might not be so bad.

It is a position that allows the operator to circumvent quite a lot of earthly laws — including AI regulation

I'm not quite sure what the suggestion is here. National and international law is at least somewhat clear on who is liable for what, and if ODCs become a bigger issue then I expect the relevant legislation will close off really obvious loopholes. For instance, if SpaceX operates ODCs, then I would assume those ODCs would still be subject to US federal regulation since SpaceX is US-domiciled. And if they cause an international incident, for example if one of their ODCs blows up and creates a bunch of debris, then the US itself is on the hook as the parent country (per Article VI of the OST + the Liability Convention). I am somewhat worried about loopholes, for example maybe an AI company could domicile in the most lenient regulatory environment for AI. But that applies to terrestrial data centers, too.

That's not true if you and your employees are on earth. It's like saying, "I'm building data centers in neutral waters, and no one can manage me." Nope. It doesn't work that way. You're subject to the jurisdiction of the country whose flag your ship flies

You're absolutely right that maritime law works this way - but actual shipping companies manage to get around it all the time.

1) Poorer nations compete with one another to have the absolute most permissive maritime regulations they possibly can so as to attract shipping companies registering with them as a flag state. (The money from such registry ain't great but it makes a significant difference to certain economies).

2) The shipping companies register ships under one flag state then, if they're ever forced to submit to regulations or go to court or anything they simply re-flag the vessel and say "Sorry, we're Panamanian now, not Nigerian, we're out of your jurisdiction". Within a few years the same vessel will fall afoul of Panama's authorities and be re-flagged as Liberian, then Bermudan, and so on.

3) When you're deep sea, you can do all sorts of illegal stuff - under both international law and the laws of even the most permissive flag states - including gross environmental damage, forced labour, human and animal rights abuses, and nobody will ever know. This stuff happens all the time (source: spent half my life at sea). There's very little money, political will, public demand, and practical ability to police the behaviour of vessels on the other side of the planet, the best part of a thousand miles from the nearest inhabited land and ten thousand miles from your country's nearest government asset.

The EU is experimenting with detecting certain kinds of common deep-sea environmental crimes by satellite remote sensing - but A) this has the same jurisdictional problems as everything else and you can be sure the rest of the world won't ever spend money on it, and B) it's only effective for detecting one type of crime, and even then only when the resulting pollution is big enough to be seen from space. The Bermudan government developing the capability to remotely detect illegal activity inside a USA-headquartered (but Bermuda-flagged) megacorp's satellite seems even more unlikely.

4) You don't even need to go to sea (or into space) for this sort of thing. Note for example that Meta's digital sweatshops and workforce-wide human rights abuses are located in Kenya and Ghana rather than San Francisco.

I really hope you're right and that flag-state controls can be relied-upon to prevent illegal activity in space - but I'm afraid maritime law (and the behaviour of entirely terrestrial corporations..) doesn't offer a very promising case-study.

The Bermudan government developing the capability to remotely detect illegal activity inside a USA-headquartered (but Bermuda-flagged) megacorp's satellite seems even more unlikely.

... but that only applies if the illegal activity stays inside the satellite. Presumably your data center is interesting because it communicates with something on Earth. People can say, "Hey, Bermuda, we're getting spam from your satellite, clean it up.". Or cut off the downlink. Not to say that the international thing wouldn't be a giant impediment to enforcement, but I don't think it's the same as somebody dumping fuel in the middle of the Pacific Ocean.

nobody can stop you without a military strike

This is only true if the development is being done autonomously from space in a way which can't be stopped from Earth, even by the Earth-bound organization which owns it. Otherwise the threat of law or force can be applied as usual to the operator/organization on Earth.

If things have gone that far, AI regulation is likely doomed anyway, space or no space.

nobody can stop you without a military strike (and the risk of Kessler syndrome).

Would it work to send a robot to rendezvous with it and turn it so the solar cells aren't facing the sun? Or cut the cables collecting the electricity?

One can think up countermeasures. And countermeasures to those countermeasures.

You can do things like this, China's Shijian-21 docked with a defunct Beidou navigation satellite and towed it into a graveyard orbit in 2022. I think this capability is still being expanded/developped.

There is a question of if soft-kill or even ASAT weapons in general (barring Kessler) can keep up with the scale of new satellites being placed in orbit however. Already we launch several thousand per year. I'd guess the economics favor an actor who can spam satellites into orbit.

This is the main driving case for orbital data centers. Space development timelines are long: the irreplaceability of hardware means that as compute efficiency increases, ground based systems can swap out compute but space based ones cannot.

Datacenters are immensely unpopular on the ground and AI companies are seeking alternate places to build them.

As far as power costs go, the real tradeoff is energy storage; if batteries are cheaper than launch costs, terrestrial will beat SSO. (But realistically, everyone will keep using the short-lead, cheap and scalable option of natural gas turbines -- its far easier to scale manufacturing of these than bespoke satellite buses and launch hardware)

Increasingly, though, new data center capacity is bottlenecked by multi-year queues to connect to the power grid.[3]

Interconnection timelines have lengthened substantially in recent years. Lawrence Berkeley National Laboratory reports that projects built in 2023 waited a median of roughly five years from interconnection request to commercial operation.

You appear to have relied upon a hallucination. Reading the chatgpt-provided citation here reveals this to be a report about connecting power plants to the grid. Interconnection for power consuming projects is a separate problem with wait times not described in the linked source. Additionally, five years is what you get by rounding the report's figure for median interconnection times of power plant projects built in 2024, the 2023 stat is lower.

You're correct. In fact I see this was caught by review but I see the replacement went to an old draft, apologies. Thanks for this.

Correcting footnote to: See, directing FERC to address large-load interconnection (2025), Reuters, Google says US transmission system is biggest challenge for connecting data centers (2026), Bain & Co., Next phase of data center growth (2025)

Claim stays the same, and we compare ODCs to both on- and off-grid options in case queues can be cut or avoided.

Permitting in space seems easy now, but once datacenters in space become a big business, regulators will be tempted to extract more rents.

As you note in the report, satellites are vulnerable to missiles. And you note they can be hacked through ground stations; once hacked, they'll be difficult to recover without physical access. Also, space pirates!

So some of the current permitting and security advantages of space may evaporate, which makes terrestrial options comparatively more appealing: Federal land in the US, countries with easier permitting, uninhabited wastes, Antarctica, and the entire ocean. These have their own engineering and security issues, but maybe easier than in space?

Kessler syndrome might be a feature not a bug. It makes it so that it's easier to destroy all the space datacenters. It might even be done in a way that can be presented as an accident.

Very interesting report!

I think the assumptions regarding vulnerability to debris are questionable, I assume that the Starlink "zero confirmed failures due to debris strikes" thing probably means complete failure of the satellite, not some sort of degradation?

Also, Starlinks don't (afaik) use radiators with coolant for heat rejection, so loss of coolant due to a radiator being hit by debris is a risk that ODCs would face and Starlinks don't. I think mitigating that would be a barrier to getting really low mass radiators, though I don't expect it's a showstopper for ODCs overall.

Regarding competition, Stoke Space, while it's a startup not a pre-existing launch provider, is also developing a fully reusable rocket (Nova). I'd also mention Rocket Lab's Neutron, which while it will have an expendable upper stage, may be able to save some costs by making the non-reusable stuff cheaper, so an interesting competitor if the fully reusable rockets (Starship, Nova) don't work out.

Wow, I did not expect those ODC cost per GW numbers to end up as close to Earth-based DCs as that graph shows. Even if the launch costs stayed above $500/kg, there would still be an argument for the right buyer to pay it, if other options are sufficiently land and energy constrained.

The calculation in Appendix E is wrong. The Earth emits longwave radiation, which are at roughly the same wavelength as the emissions of the radiator itself. Per Kirchhoff's law, the spectral emissivity equals the spectral absorptivity. Therefore, the correct heat flux due to Earth itself radiating is epsilon x F x sigma x T_earth^4, not alpha x F x sigma x T_earth^4. The contribution from reflected solar radiation due to Earth's albedo is correct, however. The heat flux from infrared radiation from Earth is then 53W/m2 instead of the ~5W/m2 your model spits out. It reduces the net rejected power by about 8% in the 20C case, from 633 W/m2 to 585 W/m2. Not enough to change the overall argument, but having the correct physics is always good.

Sun-synchronous orbits are already very popular, have you considered how large a constellation can they safely sustain before everyone placing satellites would have to pay some kind of congestion fees?

I'm aware of a few studies working on modeling this topic in some detail! We'll be very interested to read them when they publish. One consideration is that the congestion will ostensibly be intentionally high with satellites packed closely together to improve inter-satellite bandwidth.