Talk:Space-based data center

Feedback from New Page Review process

I left the following feedback for the creator/future reviewers while reviewing this article: Topic appears notable and has good coverage, even if this article may not totally reflect that.

aaronneallucas (talk) 02:51, 13 December 2025 (UTC)

Challenging the thermal control advantage proposed by User:Wikideas1

In the original 1st revision, it was listed as an advantage the following: "Radiative cooling in space reduces energy needed for thermal control".

I will remove this statement because it isn't an advantage (as I will explain) but a challenge (or one could say disadvantage).

As of today, it is unclear if those space DC (data centre) are going to be in low Earth orbit (LEO) or other type of orbit. As latency could be crucial or on-orbit servicing (OOS) might be needed, I would assume that they will use the LEO orbit.

Now, I am no spacecraft engineer, but I worked about 25 years on space ground system and gathered some knowledge about spacecraft over the years. So in a LEO lane, a satellite is never exposed constantly to sun rays. That has several consequences. One is thermal control, this means that temperature will drop significantly when in eclipse (in the Earth shadow) and significant increase when exposed to the sun light ("moving through temperatures from -170°C to +120°C every 90mins" source: ). This means that you need to cool the data centre when exposed and to warm it when in eclipse. And the other consequence is that the satellite needs to store energy to use during each eclipse.

Thermal control in space or a vacuum isn't the same as on Earth where air can transport heat away. Excess heat needs to be radiated away. On top, In LEO, you need to account of the Earth heat radiation. To cool down, you need to use heat shields to radiate the excess back in space. And to warm up, you need energy to heat the satellite instruments. But when it is cold, it is when you are in eclipse, so no sun light, no electricity from the solar panels. You need batteries. If modern electronic does not like temperature swings, batteries tolerate them even less. For spacecraft, we often use special electronic that can sustain higher and lower temperature than your standard server, but those are definitely not high GHz, high TDP, nm manufacturing process! So if newest generation AI optimised electronic can sustain space environment is to be seen. I don't say it does not work, but we cannot list it as a possibility yet. So talking about thermal control in space as an advantage is just speculation.

Energy storage to use during eclipse is another challenge. In thermal control, I already mentioned that energy need to be stored to maintain temperature during eclipse. But during eclipse you also need energy to power up the DC. I assume the DC in space won't be as big as those on Earth, but I am still wondering how much battery that will require to power it up. This adds up to the weight of such spacecraft.

Of course that my view and I didn't take time to back it by sources online. This is just what I learned over 25 years of career. I am open to feedback on this. Huygens 25 (talk) 12:31, 3 February 2026 (UTC)

Agreed on the removal as written, and I think there is better sourcing available for your position than the page currently linked.

One distinction worth adding: the claim is orbit-dependent rather than simply wrong. Your eclipse-cycling figures hold for a generic low Earth orbit. Several of the filed constellations specify sun-synchronous orbits, and a dawn-dusk sun-synchronous orbit rides near the terminator and largely avoids eclipse, which is where the near-continuous illumination claim comes from. The removed sentence did not say that, which is why it read as an unqualified advantage.

On the substance you are closer to right than the original text was, and two sources would let the article be specific:

• Shi, Zhang, Yang and Liu (2026), "Thermal Management Technologies for Space Data Centers: Current Status and Prospects", Journal of Refrigeration, open access. A peer-reviewed review that treats large-scale heat rejection in microgravity as a current unsolved engineering problem. It is published in Chinese with an English title and abstract, so WP:NONENG applies, but it is directly on topic.
• NASA's ISS Active Thermal Control System overview, which gives a hard number: the External ATCS rejects up to 70 kW through two independent 35 kW ammonia loops, and is the highest-power heat-rejection system with flight heritage. https://www.nasa.gov/wp-content/uploads/2021/02/473486main_iss_atcs_overview.pdf

Against that ceiling, publicly reported per-satellite designs are higher: SpaceX's AI1 has been reported at 120 kW sustained and 150 kW peak. That is roughly twice the demonstrated flight class, and without the ISS's tolerance for mass. A comparison like that seems more useful to a reader than sorting radiative cooling into advantages or disadvantages.

Suggested treatment: keep heat rejection under Disadvantages, sourced to Shi et al. and the NASA document, and where the article describes orbits, note that near-continuous illumination is a property of dawn-dusk sun-synchronous orbits specifically rather than low Earth orbit generally.

Disclosure: I have a conflict of interest on this topic, set out on my user page. I am not editing the article. Vaquerito (talk) 07:18, 19 August 2026 (UTC)

Article is too hypothetical

Almost all of it is speculation or promises made by companies that are seeking investment for space based DCs.

Some parts are contradictory, eg. pro: lots of room in orbit, con: limited room in orbit.

Some parts are not relevant to SBDCs, eg. the communication constellations, which serve an entirely different purpose.

The advantages and disadvantages need a lot more fleshing out to weigh their relative importance, eg. the lack of property taxes is listed as an advantage, but property taxes are an insignificant portion of the cost of running a DC.

Frankly this reads too much like a mashup of press releases from companies that have a vested interest in making the idea of SBDCs seem viable. - Wikkiwonkk (talk) 15:22, 7 April 2026 (UTC)

Can you give some examples of press release sources? WeyerStudentOfAgrippa (talk) 21:20, 21 June 2026 (UTC)

History section omits Blue Origin's Project Sunrise (51,600 satellites, March 2026)

The History section currently covers Blue Origin's TeraWave constellation (~5,400 satellites, January 2026), described as "high-throughput networking for data centers, enterprise, and government customers." That is accurate, but TeraWave is Blue Origin's connectivity constellation. The article does not mention Project Sunrise, which is Blue Origin's orbital data centre constellation proper, and is considerably more relevant to this article's subject:

• FCC application SAT-LOA-20260310-00118, filed 10 March 2026
• Up to 51,600 satellites
• Circular sun-synchronous orbits, 500–1,800 km
• Connects to TeraWave by optical link; the two are complementary, not alternatives

Sources: Ma, Jason (20 March 2026), "Bezos-backed Blue Origin files for approval to put data centers in space", Data Center Dynamics; Boyle, Alan (21 March 2026), "51,600 more satellites? Blue Origin adds another twist to the data center space race with Project Sunrise", GeekWire; also covered by SpaceNews and Tom's Hardware.

Suggested addition to the History section, after the existing TeraWave sentence:

In March 2026, Blue Origin filed a separate application with the FCC for Project Sunrise, a constellation of up to 51,600 satellites in circular sun-synchronous orbits between 500 and 1,800 km, intended to host orbital data centres and to connect to TeraWave by optical link.<ref>{{cite news |last=Ma |first=Jason |date=20 March 2026 |title=Bezos-backed Blue Origin files for approval to put data centers in space |work=Data Center Dynamics}}</ref><ref>{{cite news |last=Boyle |first=Alan |date=21 March 2026 |title=51,600 more satellites? Blue Origin adds another twist to the data center space race with Project Sunrise |work=GeekWire}}</ref>

Relatedly, Cowboy Space appears in the "Companies pursuing space-based AI infrastructure" list, but its Stampede filing (up to 20,000 orbital data centre satellites, May 2026, covered by SpaceNews, Data Center Dynamics and Via Satellite) is absent from the History section. The Secure World Foundation also filed comments on that application, which may be a useful independent source for the orbital-sustainability points in the Disadvantages section, several of which are currently uncited.

Disclosure: I have a conflict of interest on this topic. I research this area under a paid engagement with Turion Space, a company with a programme in adjacent in-space manufacturing. I am therefore not editing the article directly and am raising this here instead. Details on my user page. Vaquerito (talk) 07:02, 19 August 2026 (UTC)