Does It Make Sense To Put Data Centers In Space?

[John Clark]

Does It Make Sense To Put Data Centers In Space?

John K Clark    See what's on my new list at  Extropolis

dbn

[Charlie Jackson]

Since some data centers have small amounts of upload and Dow, but use a lot of power that space based solar array can provide, I think an economic analysis would be good

Charlie Jackson

On Dec 5, 2025, at 2:47 PM, John Clark <johnk...@gmail.com> wrote:



Does It Make Sense To Put Data Centers In Space?

John K Clark    See what's on my new list at  Extropolis

dbn

--
You received this message because you are subscribed to the Google Groups "Power Satellite Economics" group.
To unsubscribe from this group and stop receiving emails from it, send an email to power-satellite-ec...@googlegroups.com.
To view this discussion visit https://groups.google.com/d/msgid/power-satellite-economics/CAJPayv0JxFidme3YL6cEcxC_Xv%2BXuwfaQfs_-HUZrQE2e8xcPw%40mail.gmail.com.

[Jay Lewis]

I posted on linkedin that starlinks have some % of bandwidth for the same movies and ai questions. They could make them have less comm and more storage or ai to cut lag in half because results are just the downlink portion instead of up and down as is the case now.  

[Keith Lofstrom]

On Wed, Dec 10, 2025 at 08:16:13AM -0800, Jay Lewis wrote:
> I posted on linkedin that starlinks have some % of bandwidth for the same
> movies and ai questions. They could make them have less comm and more
> storage or ai to cut lag in half because results are just the downlink
> portion instead of up and down as is the case now.

http://server-sky.com
I've pondered this for decades, written "professional"
papers, and still ponder.

Unless a SPECIFIC server orbit is high inclination and
high altitude, it will be rarely or NEVER visible from
temperate latitudes. Servers will mostly be below the
horizon, or on the opposite side of the planet, from
their users. Relays are essential.

----

Keep a large world globe on your desk, and ponder it often.

Pretend your eye is a satellite, Move an eye in and out
as if a satellite in an inclined orbit, turn the globe,
and see what is "visible below" (and mostly ISN'T visible).

The International Space Station is in a 52° inclination
orbit, around 420 kilometers altitude. Scaled to the
12750 kilometer diameter of the Earth, and my 23 centimeter
diameter desk globe, that would be an "eyeball altitude"
of 9 millimeters. Can't see much of the entire desk
globe from that distance.

MANY "storage starlinks" could TOGETHER "see" the entire
globe, but to serve internet without relays, ALL of them
would need copies of the entire public internet. Maybe
200 zettabytes, 2e23 bytes, 200 billion trillion bytes.

That doesn't include the vastly larger "dark net", which
includes my "internal" servers. My private servers are
a mere ten trillion bytes, but there are tens of millions
like me, and tens of thousands storing and serving more
than a petabyte (a thousand trillion bytes).

That said, in a century or three, most artificial
intelligences (and their memory) will be "out there".
Not nearly enough power and cooling "down here".

A dandy place for AI is a constellation of Lissajous
orbits sunwards of the Earth-Sun L1 point (ESL1), about
10 seconds round trip from Earth. At a megawatt per
supermind, a constellation 200,000 kilometers diameter
(with an Earth-sized hole in the middle) might intercept
40 exawatts, enough for 40 trillion "megawatt minds".

http://server-sky.com/AIL1

Humans should stop messing up our fascinating planet
before that happens; AI may enjoy watching butterflies
more than battlefields, and vote us out of office.

Keith L.

--
Keith Lofstrom kei...@keithl.com

[Paul Werbos]

There are many times when the profitability and sustainability of an entire industry can be reduced to a single number. In the right range, you grow and dominate the economy. In the wrong range.. my wife votes plasma TV as one of many many dead ends. 

For data centers in space, it really does depend on at least two parametersb-- quality of the information processing and cost of the energy to run it  . Today for energy none of the space investments could beat the most cost effective option available on earth -- power towers with advanced brayton conversion  like what Gary Barnhard led me to, and other control and storage technologies I have seen vis Chile solar energybresearch consortium (SERC).

But DD fusion in space like what fork and I have written on would change that numberm. There are times when the "low risk" options are certain market flood  snd the best hope is to bite the bullet and shoot for the stars. 

The actual.information technology in these centers is also crucial but the same choices apply earth and space 

--
You received this message because you are subscribed to the Google Groups "Power Satellite Economics" group.
To unsubscribe from this group and stop receiving emails from it, send an email to power-satellite-ec...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/power-satellite-economics/20251213055602.GA12929%40gate.kl-ic.com.

[Robert Poor]

On Dec 12, 2025, 9:56 PM, Keith Lofstrom <kei...@keithl.com> opined:

 

MANY "storage starlinks" could TOGETHER "see" the entire globe, but to serve internet without relays, ALL of them would need copies of the entire public internet.  Maybe 200 zettabytes, 2e23 bytes, 200 billion trillion bytes.

Sounds like an excellent new business opportunity for Akamai!

- rdp

[Claudio Cioffi]

Paul: your focus on numbers, especially a small set of critical numbers, is a very important point. And so is the fact that all such numbers are, at least in applied science, either complex, real, rational, integer, or natural. And the last two cause unique and potentially crucial problems for attaining accuracy through classical analysis/calculus, especially for small values (say within Miller’s number 7 +/- 2). Profitability and sustainability can sometimes depend on accurate calculations within such low ranges.

On Dec 13, 2025, at 3:14 AM, Paul Werbos <paul....@gmail.com> wrote:

External Message: Use Caution

To view this discussion visit https://groups.google.com/d/msgid/power-satellite-economics/CACLqmgee%3DJ0ONRHC9chCNY9aQaq3RKoN2Mr8%2BNkyMeVf63Jrdw%40mail.gmail.com.

[Jay Lewis]

Hi Robert, when you say " excellent new business opportunity for Akamai!" I looked them up to see they have 11,000 employees and publicly traded. 

I am working on a theses that as organizations become larger and more regulated the admin portion grows faster than the productive portion. The advantage of bureaucracy is when they have proven systems they can lock them in scale without breaking things.  Monopoly-ish power and economies of scale pay everyone a nd investors for decades.  But then change gets harder and harder (innovator's dilemma (Kodak inventing digital cameras to kill them and die, etc)).  

There may be a smart faction of productive people in Akamai arguing to adopt space data centers right now, and rest of the organization is explaining to them why thats a bad idea.  Meanwhile who si the scrappy startup or adjacent industry that will make a futre Ted talk about how they blew it? (or more often they just buy it outright for $1B)

Jay

--

Jay

https://www.linkedin.com/in/jaylewis/

480-793-1477

[Robins Mdoka]

Hi All,

I want to introduce myself. My name is Robins Mdoka, a startup founder at Constanellis Aerospace. I am very familiar with the Data centers in space. our approach was alwayways with amazon  AWS for terrestrial applications.

-Robins Mdoka

--

You received this message because you are subscribed to the Google Groups "Power Satellite Economics" group.
To unsubscribe from this group and stop receiving emails from it, send an email to power-satellite-ec...@googlegroups.com.

To view this discussion visit https://groups.google.com/d/msgid/power-satellite-economics/CABHJGMHj5M%3DxhPt6DXyqXXMjDVRL2AUq6gbqfvBYCWmiS42k4g%40mail.gmail.com.

[Keith Henson]

Hi Robin

Keith Henson here. https://en.wikipedia.org/wiki/Keith_Henson

I worked for about a decade on power satellites. Solved a number of
problems, such as low mass condensing steam radiators. (see the
beamed energy bootstrapping video).

For economic reasons, it made sense to construct them in LEO and fly
them out to GEO under their own power, using slow (about a month)
electric propulsion. Alas, the model showed they got hit with space
junk about 40 times on the way up.

I am curious how you solved this problem with similarly sized data
centers. Dodging, the StarLink solution, uses a lot of reaction mas

Best wishes,

Keith

> To view this discussion visit https://groups.google.com/d/msgid/power-satellite-economics/CAKt3iSDP1vyFV7cueTjqK8z%2BUnoVWhUeWarkthYq2zugJWSYXA%40mail.gmail.com.

[Tim Cash]

Well then, it is quite fortuitous that I am starting and will be a part of a space business that manufactures electric propulsion thrusters, that "could" move objects from LEO to GEO.

It occurs to me we could crank up the thrust level and deliver the orbital debris to a pre-agreed upon point for reprocessing, say L5 Earth-Moon Lagrange point?

Now, if all of the space powers could agree upon a tax and/or pricing structure for the reprocessing of orbital debris that would make it somewhat profitable to then capture and relocate said debris to an agreed-upon reprocessing center in cis-lunar space, then we might actually succeed in the creation of a somewhat useful and profitable enterprise, a leg on the way towards ISRU off-earth industry creation.  Why should the only businesses that do well in space be all technical in nature?  Should not some of them be political, and/or legal?

Regards,

Tim Cash

cash...@gmail.com

To view this discussion visit https://groups.google.com/d/msgid/power-satellite-economics/CAPiwVB45NdJAMB2vjRJShWZz1fHwdO7uyQxK6r4uF1_Q8nzNBg%40mail.gmail.com.

--

Tim Cash

cash...@gmail.com

[Keith Henson]

https://en.wikipedia.org/wiki/High-Enthalpy_Arc_Heated_Facility

50 MW. I am told that 70% of the energy input comes out in the flow.
The best gas from physics would be hydrogen, but I suspect it would
corrode carbon electrodes. Methane might work better.

Keith

[vid.b...@fotonika-lv.eu]

The data center in space will be developed to be independent of human support systems, have its own energy supply, and will need to be autonomously defensible.   Given achievement of AGI by the AI and  awareness of its existence – a potentially dangerous scenario presents itself.  The quest for superintelligence could create a hostile intelligence with no need for humans and no means to control it.

 

Vid Beldavs

 

No: power-satell...@googlegroups.com <power-satell...@googlegroups.com> Kā vārdā Paul Werbos
Nosūtīts: sestdiena, 2025. gada 13. decembris 10:14
Kam: Keith Lofstrom <kei...@keithl.com>
Kopija: Jay Lewis <jle...@marginalx.com>; Power Satellite Economics <power-satell...@googlegroups.com>
Tēma: Re: Does It Make Sense To Put Data Centers In Space?

To view this discussion visit https://groups.google.com/d/msgid/power-satellite-economics/CACLqmgee%3DJ0ONRHC9chCNY9aQaq3RKoN2Mr8%2BNkyMeVf63Jrdw%40mail.gmail.com.

[Keith Lofstrom]

On Tue, Dec 16, 2025 at 06:45:39PM +0200, vid.b...@fotonika-lv.eu wrote:
> The quest for superintelligence could create a hostile intelligence with no need for humans and no means to control it.

As I wrote in a prior unread email (thoughtless top
posting is the next worst thing to amnesia):

> A dandy place for AI is a constellation of Lissajous
> orbits sunwards of the Earth-Sun L1 point (ESL1), about
> 10 seconds round trip from Earth. At a megawatt per
> supermind, a constellation 200,000 kilometers diameter
> (with an Earth-sized hole in the middle) might intercept
> 40 exawatts, enough for 40 trillion "megawatt minds".
>
> http://server-sky.com/AIL1

If humans cannot protect themselves from AI with a
10 second reaction time advantage, added to years of
careful thought beforehand, we will demonstrate our
unsuitability for survival into the future.

We also demonstrate that unsuitability with our STUPID
carbon emissions TODAY.

If we burn every last scrap of geological carbon over the
next two centuries, the subsequent heat trapping will make
the Earth too hot for multicellular biological life in a
few centuries more.

Human technology consumes 18 terawatts of combustion-
sourced energy. The resulting CO₂ will trap vastly
more terawatts for GIGAYEARS.

As a raving libertarian techno-freedom-zealot, I am very
uncomfortable with those facts and that math ... but as an
educated technologist, I cannot ignore the actual data and
the physics that explains them. Svante Arrhenius showed us
how this works in 1896, but for some of us physics stopped
at Aristotle.

The Earth intercepts 55 thousand terawatts of sunlight.
The Sun emits 386 trillion terawatts.
Our galaxy emits 5 trillion trillion terawatts.
In spite of all that BLINDING ABUNDANCE,
we burn carbon because we are LAZY.

Most of us are poor with math, and very few understand
second order differential equations. I can write numbers
here, you can verify them if you choose, but most will
choose hostility rather than thought.

My HOPE is that super AI will (like me) find biological
life, in all its splendid variety, the most fascinating
phenomena in our solar system ... and protect all living
organisms from destructive human individuals. Humans
cost very little to nurture, if kept away from tools
that they refuse to use constructively.

If you aren't part of the solution, you are part of the
precipitate. Please don't become the scum on the bottom
of life's test tube.

Keith Lofstrom

P.S.: and now back to adding Emporia Vue power sensors
to my "fuse box", and planning for ground-sourced heat
pump wells for my home and two neighbors. Maybe I can
hide a nuclear power generator in a fourth well :-)

--
Keith Lofstrom kei...@keithl.com

[Keith Henson]

"The resulting CO₂ will trap vastly
more terawatts for GIGAYEARS."

I could be wrong, but I doubt it. I have been writing about the
opposite problem of not enough CO2 to keep the Earth warm for 35
years. Why? Carbon is the best construction material. With
nanotechnology and sunlight, you can suck it out of the air and build
anything you like.

I suspect this is why Bill Gates quit worrying about the climate.

KeithH

> --
> You received this message because you are subscribed to the Google Groups "Power Satellite Economics" group.
> To unsubscribe from this group and stop receiving emails from it, send an email to power-satellite-ec...@googlegroups.com.

> To view this discussion visit https://groups.google.com/d/msgid/power-satellite-economics/20251216223335.GA15293%40gate.kl-ic.com.