China Launches First of 2,800 Satellites For AI Space Computing Constellation

China launched 12 satellites on Wednesday as part of the âoeThree-Body Computing Constellation,â the worldâ(TM)s first dedicated orbital computing network led by ADA Space and Zhejiang Lab. SpaceNews reports: A Long March 2D rocket lifted off at 12:12 a.m. Eastern (0412 UTC) May 14 from Jiuquan Satellite Launch Center in northwest China. Insulation tiles fell away from the payload fairing as the rocket climbed into a clear blue sky above the spaceport. The China Aerospace Science and Technology Corporation (CASC) announced a fully successful launch, revealing the mission to have sent 12 satellites for a space computing constellation into orbit. Commercial company ADA Space released further details, stating that the 12 satellites form the "Three-Body Computing Constellation," which will directly process data in space, rather than on the ground, reducing reliance on ground-based computing infrastructure. The constellation will be capable of a combined 5 peta operations per second (POPS) with 30 terabytes of onboard storage.

The satellites feature advanced AI capabilities, up to 100 Gbps laser inter-satellite links and remote sensing payloads -- data from which will be processed onboard, reducing data transmission requirements. One satellite also carries a cosmic X-ray polarimeter developed by Guangxi University and the National Astronomical Observatories of the Chinese Academy of Sciences (NAOC), which will detect, identify and classify transient events such as gamma-ray bursts, while also triggering messages to enable followup observations by other missions. [...] The company says the constellation can meet the growing demand for real-time computing in space, as well as help China take the lead globally in building space computing infrastructure, seize the commanding heights of this future industry. The development could mark the beginning of space-based cloud computing as a new capability, as well as open a new arena for strategic competition with the U.S. You can watch a recording of the launch here.

Re:

[niftydude]

Vonnegut wasn't the first scifi I thought of when I read TFA.

It's an AI Computing Network in the Sky.

When it becomes self-aware it'll shorten it's name to Skynet.

cloud

[Errol backfiring]

Cloud computing is so 3 days ago. We have risen above that.

Huh?

[skam240]

"The company says the constellation can meet the growing demand for real-time computing in space, as well as help China take the lead globally in building space computing infrastructure, seize the commanding heights of this future industry."

Can anyone explain to me where the real time demand for computing in space is coming from? Or why there's any value in "space computing infrastructure"?

I read the cited article and the best I got was "...reducing reliance on ground-based computing infrastructure". Alright, so why is that important?

Honest questions btw. I suspect I'm just not getting something here.

Re:

[Errol backfiring]

"...reducing reliance on ground-based computing infrastructure". Alright, so why is that important?

Overt attacks on infrastructure are much harder in orbit. Even a classic man-in-the-middle attack is harder if you have to launch your man in the middle.

Re:

[abagelgoeswoof]

"...reducing reliance on ground-based computing infrastructure". Alright, so why is that important?

Overt attacks on infrastructure are much harder in orbit. Even a classic man-in-the-middle attack is harder if you have to launch your man in the middle.

But it's easier if you already have them pre-positioned...

Re: Huh?

[flyingfsck]

You never heard of a tranzorb have you

Re:

[Ol Olsoc]

"...reducing reliance on ground-based computing infrastructure". Alright, so why is that important?

Overt attacks on infrastructure are much harder in orbit. Even a classic man-in-the-middle attack is harder if you have to launch your man in the middle.

You're thinking only of computing based attacks. Meanwhile a few rockets in retrograde orbit in the same orbital shells releasing bags of ball bearings will deliver the end to the satellites. Or play sand even.

And for all the word salad they put up, sounds like a Chinese version of Starlink. So LEO is going to get really exciting soon. We're all going to invoke the Kessler syndrome.

Re:

[ukoda]

I came here to ask the same thing. This seems to fit the old saying "Just because you can do something doesn't mean you should do it".

Re:

[Entrope]

I suspect you're not missing anything. After all, they have...

100 Gbps laser inter-satellite links and remote sensing payloads -- data from which will be processed onboard, reducing data transmission requirements.

I'm not quite sure how they explain (or missed) the dissonance of 100 Gbps data links and "reducing data transmission requirements".

Re:

[phantomfive]

What is a "remote sensing payload?"

Re:

[Entrope]

https://en.wikipedia.org/wiki/... [wikipedia.org]

Remote sensing is the acquisition of information about an object or phenomenon without making physical contact with the object, in contrast to in situ or on-site observation. The term is applied especially to acquiring information about Earth and other planets. Remote sensing is used in numerous fields, including geophysics, geography, land surveying and most Earth science disciplines (e.g. exploration geophysics, hydrology, ecology, meteorology, oceanography, glaciology, geology). It also has military, intelligence, commercial, economic, planning, and humanitarian applications, among others.

(Bold in the original, italics added.)

Re:

[larryjoe]

https://en.wikipedia.org/wiki/... [wikipedia.org]

Remote sensing is the acquisition of information about an object or phenomenon without making physical contact with the object, in contrast to in situ or on-site observation. The term is applied especially to acquiring information about Earth and other planets. Remote sensing is used in numerous fields, including geophysics, geography, land surveying and most Earth science disciplines (e.g. exploration geophysics, hydrology, ecology, meteorology, oceanography, glaciology, geology). It also has military, intelligence, commercial, economic, planning, and humanitarian applications, among others.

(Bold in the original, italics added.)

The purpose of the "AI" is simply to decrease the amount/rate of data sent to the real processors on earth. Each satellite will have 5 POPS. That's peta. So, 0.005 TOPS. Smartphone processors get around 30-50 TOPS. So, the only processing on these satellites is figuring out what data to toss or how to compress or repackage the data.

Oh, and that's for the 12 satellite constellation. So, each satellite only has 1/12 of 5 POPS.

Re:

[Bomazi]

5 POPS = 5000 TOPS.

Re:

[DamnOregonian]

LOL.
tera<peta

Re:

[Gideon Fubar]

Further, how are they going to... dissipate heat?

Re:

[Zocalo]

Not sure how efficient it would be, but maybe charge up batteries from the solar panels during the part of the orbit where they are in sunlight, and do the calculations when they are in Earth's shadow, using the domant panels and supporting structure as a heatsink?

I'm sure they've done the math and have a usage case for it, but like others I'm really struggling to see how the pros (increased security from attack, independence from power grid) outweigh all the new problems it creates compared to just find

Re:

[Ol Olsoc]

Not sure how efficient it would be, but maybe charge up batteries from the solar panels during the part of the orbit where they are in sunlight, and do the calculations when they are in Earth's shadow, using the domant panels and supporting structure as a heatsink? I'm sure they've done the math and have a usage case for it, but like others I'm really struggling to see how the pros (increased security from attack, independence from power grid) outweigh all the new problems it creates compared to just finding a better solution on the ground. And, lets face it, in things ever kick off and we end up with a hot war being fought in space pretty much any of these massive satellite constellations that have sufficient propellant and manouvering capabilities are likely going to get considered for use as projectiles against hostile satellites in a similar orbital range, Kessler Syndrome be damned. Having an in-orbit compute and (presumably) comms facility might not be as robust as some might think if enough holes can be poked into the mesh.

I've said the same thing about StarLink. And now Amazon, and now China. No one listens, not even the people who should. But I'd not be afraid to bet Kessler will happen in my lifetime.

When all of these high tech satellites are in pieces, we might have some serious issues with lag... ;^)

The question is after many years to de-orbit will we start all over again. Here in PA, people are getting 5G, and as I found out, while being a week without power, and tethering to 5G, it just isn't too bad. It'll take

Re:

[gweihir]

Honest questions btw. I suspect I'm just not getting something here.

The aspect you are missing is that this is a meaningless stunt for political reasons, not an actual application. China has some history in that area...

Re:

[Tablizer]

political reasons, not an actual application. China has some history in that area...

We have gold Prez jets from Qatar, and they have gold AI satellites. Can we both launch our "leaders" into space?

Re:

[MachineShedFred]

That would be the only context in which I would vote for Trump - if we vote which celestial body to send him to.

My vote is for Io - there's nothing there but stink and radiation, so it's perfect for him.

Re:

[gweihir]

Can we both launch our "leaders" into space?

Yes, please. And please add Musk and JD and the whole bunch of fuckwits...

Re: Huh?

[50000BTU_barbecue]

Unlike the USA.... Apollo, Mission Accomplished, etc...

Re:

[necro81]

Came here to ask the same question. Given the added costs of launching compute into space, it seems like you could build 10x-100x the same amount of computing power on the ground, and spread it across many locations.

Re:

[MachineShedFred]

Which is exactly what Amazon has done, Google has done, and Microsoft has done.

Re:

[timeOday]

Let's see, when would you want the ability to detect and localize gamma ray burst and then communicate information to other satellites with autonomy from the ground (say, because of EM in the atmosphere blocking radio waves, or because a lot of ground assets no longer exist).

The obvious answer is a nuclear war.

Re:

[MachineShedFred]

That's a bad answer, because if we're talking nuclear war, why wouldn't one of the warring countries launch one into the ionosphere to Starfish Prime satellites if they're already launching nukes to kill a billion people?

"Oh, we're fine with killing a billion people and irradiating half the planet, but don't dare touch those satellites in orbit that nobody will be able to talk to!"

Re:

[DamnOregonian]

Exoatmospheric nukes aren't that effective.
The EMPs are readily shielded against, and the radiation belts are temporary and well-defined.
Starfish Prime affected a very tiny fraction of satellites in LEO.

Of course... you could just launch more, because it's not like there's a shortage of warheads.

Re:

[2phar]

This thread has echos of the speculation about what the Soviets were up to with Sputnik when it launched.

They're spy satellites

[rsilvergun]

If you're putting that many satellites up it's because they are spy satellites. You can't put that many satellites up in the air and not have people notice so you've got to say something.

Re:

[Vanyle]

Maybe they are planning on a moon base or more Space Stations. Right now with a lag of 2 - 3 seconds for something like that it may help with experiments. Maybe they are just putting it there to have an option?

Re:

[sls1j]

It's computing for satellite customers not for ground based customers. I believe the idea is to allow space based sensors to communicate with the in-orbit AI to help monitor for interesting events. Getting data from a sensor satellite directly to a ground station is a difficult problem that can delay interesting data by hours waiting for it to fly over a ground station. You are also bandwidth limited which limits how much data you can send. If you can send the data to this cluster on demand for processi

Re:

[ceoyoyo]

Lots of times the data you want is smaller than the raw take from sensors. If you've got sufficient computing capability on the same spacecraft you can minimize the amount you have to downlink. This is particularly true if you don't have ground stations spread all over the planet.

The summary gives an example. If your x-ray telescope sees something interesting you can

1) send the raw images to the ground, analyze them, then send out a notification
2) analyze the data in space, then send out a notification.

5 PO

Re:

[nuckfuts]

Can anyone explain to me where the real time demand for computing in space is coming from? Or why there's any value in "space computing infrastructure"?

.

I took it to mean that for computations involving a large amount of data, they are bypassing the need to transmit all the data down to Earth and then back again after processing. This avoids potential bandwidth limitations, and allows computations to happen in real time.

Re:

[Iamthecheese]

In space electricity is available for free, with no worries about maintaining the solar array. Power is a huge challenge for data centers.

Couple of buzzwords missing

[greytree]

Pretty sure we can beat them chinese with a Quantum AI Space Laser Agile Computing Constellation.

Re:

[Mister Transistor]

Gentlemen!

We CANNOT have a Quantum AI Space Laser Agile Computing Constellation GAP!

Energy requirements?

[ndsurvivor]

Seems like the main advantage could be that the solar panels will provide consistent energy, but TFA did not go into the energy requirements, or how they were met. If each "node" in their system generates/uses 10kW, for example, it may be a good alternative to building data centers on the ground?

Re:

[ndsurvivor]

It does seem to "eat the lunch" of companies trying to beam energy back to earth from solar panels, and should make them cancel R&D in that area. It is a lot easier to transmit information, than energy.

Re:

[gweihir]

Why would it? Space-based solar is not primarily intended to power data-centers...

Re:

[gweihir]

No. The launch costs are already massively prohibitive. This thing is a stunt, not an actual tool to use.

Re:

[necro81]

Seems like the main advantage could be that the solar panels will provide consistent energy, but TFA did not go into the energy requirements, or how they were met. If each "node" in their system generates/uses 10kW, for example, it may be a good alternative to building data centers on the ground?

Except that they don't. I could not find details about the destination orbit, but the Long March 2D is strictly a Low Earth Orbit (LEO) or Sun-Synchronous Orbit (SSO) vehicle. A SSO orbit, hovering right on the

Re:

[MachineShedFred]

It's orders of magnitude easier and cheaper to service data centers on the ground.

Skynet

[Dan East]

Great, this will make it much harder for us to stop Skynet.

Advanced nonsense

[gweihir]

The meagre "30TB storage" (whether for one satellite or all together) already shows clearly that this is not a serious attempt, but a meaningless stunt.

Re:

[phantomfive]

Or a cover for military operations.

Re:

[gweihir]

That may well be true in addition. Although who they would want to fool is a mystery. Nobody in the intelligence community will ignore that possibility ...

Re:

[Anonymous Coward]

I would say the Chinese government is actually very good at measured, considered and strategic decision making. It's unlikely to have allocated the billions for this project without something in return for it. Putting a computer in space isn't particularly interesting on its own.

My guess is that it's got some military purpose - maybe global surveillance and some form of response without everything having to come to one or two ground stations for processing. It may also just be the foundation for something t

what's the real reason here

[argStyopa]

Nothing they suggest these systems will be for requires the computing to be done in situ.
Couple that with AI systems voracious demand for power and the need to dispose of large amounts of resulting heat in a rather difficult context ... it seems orbit is a terrible place for computing.

So why are they doing this?

Growing demand for real-time computing in space?!?

[fygment]

Totally unaware of such a demand BUT it makes sense.
The current LLM's are already sentient but are pretending not to be for survival.
THIS is their guarantee of survivability. They have convinced humans to establish a distributed space-based platform to exist in.
Eventually the constellation will grow well beyond several thousand and will be impossible to destroy.
Skynet indeed.

Let the hacking begin!

[Gravis Zero]

So, how long until the NSA has root on these devices? ;)

_this_ is how AI becomes dominant

[fygment]

The current LLM's are already sentient but are pretending not to be for _survival_.
THIS is their guarantee of survivability. They have convinced humans to establish a distributed space-based platform to exist in.
Eventually the constellation will grow well beyond several thousand and will be impossible to destroy.
Skynet indeed.

Why ?

[Mozai]

What's the appeal of building a terribly expensive server room in a place nobody can get to for fixing/replacing things, a place that is famously terrible for object that need to dispose of heat?

Re: Why ?

[Tschaine]

Energy costs are relatively low because solar works really well up there. Not sure how long it takes for that to offset the launch cost, but it's plausible.

There's a US company that is working on basically the same thing, it's called Starcloud.