Data centers in space in 30 months???!!!

[Scott Watkins]

[jyl]

He’ll say anything to keep his companies’ valuations up. With the IPO of Space X + xAI coming, expect a whole lot more of this.

[KFC911]

Sooo ... "How long will it be before we are back "up and operational?"

"About an hour"... thinking (.... now GTF outta here and go smoke another bowl of pipe-dream while we figure this out), and not having a clue either - Me

[masraum]

I didn't watch the video.

I've gotta think that a DC will need to be large, quite large, even without an enormous cooling infrastructure that terrestrial DC require. The amount of power and space required for a DC is very large. I also think that they will have to be in a geosynchronous orbit. For them to be useful, they'll also need to communicate with normal latencies. There's no way to have something in a geosynchronous orbit with latency that's not quite long. Can this be done, maybe, but I doubt it, at least not today with the tech that's available.

[sc_rufctr]

Just one question... Why Space?

Everything about data in space is difficult including security and transmission speeds.

[GH85Carrera]

Originally Posted by sc_rufctr

Just one question... Why Space?

Everything about data in space is difficult including security and transmission speeds.

The only advantage of space is the vacuum, and the possibility to radiate heat. Since it is a vacuum, there is nothing to radiate the heat into. The images of science fiction movies shows people freezing up in seconds. That is not accurate as the blood and liquid will boil, and the person will dies quickly, but they don't just freeze solid in seconds. It will take a while to cool off in Earth's shadow. In the direct sunshine it can be very hot.

It will be interesting to see if there is any advantage at all with the huge cost of getting to space.

[jyl]

He says that in space there is solar power 24H a day and low earth orbit latency is low. Actually LEOs are in shadow a quarter to a third of the time. He says they’ll use the LEOs at the shadow line. Then all his space DCs have to be in the same or similar orbit, latency will rise, the orbit will get crowded by others with the same idea. Heat dissipation in space requires radiators, a lot of heat requires larger radiators. The space DCs will have to carry not only huge solar panels (calculate panel area required per MW even with higher solar intensity) but also batteries and radiators. They’ll also be disposable, no way to upgrade as new hardware emerges. It doesn’t make sense, in practical terms. It makes sense, in marketing terms. xAI is burning a huge amount of cashflow, Space X is generating some cashflow, his ownership of xAI is higher than of Space X. So he merged Space X with xAI, the combination looks better from a cashflow perspective and he has a higher ownership interest in it than he did in Space X alone. To justify this, he is talking about how space and AI need each other and he’s the only one with all the pieces. Expect lots more of this blather leading up to the IPO. Maybe they’ll send a little demonstration DC into space - others have done it, no reason they can’t send twenty GPUs and some panels up there and get a million media stories. But it’s all a scam.

[wdfifteen]

Originally Posted by jyl

But it’s all a scam.

He could pay for it with Crypto coins. Hell, he could cool it and power it with Crypto. Crypto is magic.

[jcwade]

He has apparently given up on Mars by 2026 (I wonder how that's coming along) and now plans to build a city on the moon in 10 years.
Ten years is a long timeline for Musk. I guess that's known as the long con.

But no matter what nonsense he spouts, the Elongelicals will lap it up.

[LWJ]

I imagine it won't work. Because, as said above, it's a scam. Which I agree. EM is the world's most successful stock manipulator.

But if it did work? When we have a SKYNET situation. (oh damn. SKYNET. Arg. That sort of fits) How the holy HELL do you pull the plug on something you can't touch?

[red 928]

Originally Posted by sc_rufctr

Just one question... Why Space?

Everything about data in space is difficult including security and transmission speeds.

He already has a whole bunch of satellites up there that transmit data. That's why he put them there.
All he has to do is install into the satellites large amounts of data storage capacity, increase the data transmission capacity, and power the whole thing.
Easy peasey

[KFC911]

"Easy Buttons - 3 pack" ... only 19.95 plus s&h

- William Mays Musk Sr.

[onewhippedpuppy]

I’m not seeing the point of data centers in space, but I also didn’t anticipate Starlink to be this good either. So while it’s very possible that he’s FOS, I also wouldn’t fully bet against Elon either. Starlink shows us that latency isn’t an issue, and I suspect he could use LEO Starlink satellites for data relay purposes. I’m more curious about the practical reality of cooling and solar power generation in space compared to the logistics of more electrical power generation on Earth. If we would simply invest in more nuclear plants this entire issue would go away.

[KFC911]

I dabbled in network (Hughs satellite VSAT) with geocentric latency way before Wayne was there ... and data centers too ... just buttons....

19.95 plus bs & handling

Cybertrucks will power them!

[jyl]

Originally Posted by jcwade

He has apparently given up on Mars by 2026 (I wonder how that's coming along) and now plans to build a city on the moon in 10 years.
Ten years is a long timeline for Musk. I guess that's known as the long con.

But no matter what nonsense he spouts, the Elongelicals will lap it up.

He gave up on Mars because the US govt is focused on the moon, so moon gets more government contracts.

Everything he says is intended to get him something.

[masraum]

Originally Posted by jyl

He says that in space there is solar power 24H a day and low earth orbit latency is low.

I wonder if he's planning to put entire city blocks (or larger) in LEO, because Data Centers are usually BIG.

Actually LEOs are in shadow a quarter to a third of the time. He says they’ll use the LEOs at the shadow line.

Huh? Don't they all hit shadow at times. I would be surprised if he had any that were never in shadow.

Then all his space DCs have to be in the same or similar orbit, latency will rise, the orbit will get crowded by others with the same idea. Heat dissipation in space requires radiators, a lot of heat requires larger radiators. The space DCs will have to carry not only huge solar panels (calculate panel area required per MW even with higher solar intensity) but also batteries and radiators.

Right, so we're going to put satellites the size of aircraft carriers in LEO. No problem. Sounds good.

They’ll also be disposable, no way to upgrade as new hardware emerges. It doesn’t make sense, in practical terms. It makes sense, in marketing terms. xAI is burning a huge amount of cashflow, Space X is generating some cashflow, his ownership of xAI is higher than of Space X. So he merged Space X with xAI, the combination looks better from a cashflow perspective and he has a higher ownership interest in it than he did in Space X alone. To justify this, he is talking about how space and AI need each other and he’s the only one with all the pieces. Expect lots more of this blather leading up to the IPO. Maybe they’ll send a little demonstration DC into space - others have done it, no reason they can’t send twenty GPUs and some panels up there and get a million media stories. But it’s all a scam.

20 GPUs is NOT a data center. LOL! Maybe I should invest now with the idea of pulling out later.

[KFC911]

Steve ... yer on call for network problems ... I'll let Elon push the easy button for latency ...

Mebbe 12" ... that's what Dr. Wirth displayed by holding his hands apart .... methinks it was "Switching Theory 101" ... how far light travels in a nanosecond ... give or take

The devil is always in the details...

Ea$y Pea$y buttons ... prolly take a few bazillion bucks and back ordered

[Dixie]

That going take a LONG power cord

[KFC911]

Originally Posted by Dixie

That going take a LONG power cord

Cynicism will not get you brownie no$ing point$ ... yer NOT a team player!

"Our be$t people are working on thi$"

[masraum]

Originally Posted by KFC911

Steve ... yer on call for network problems ... I'll let Elon push the easy button for latency ...

Mebbe 12" ... that's what Dr. Wirth displayed by holding his hands apart .... methinks it was "Switching Theory 101" ... how far light travels in a nanosecond ... give or take

The devil is always in the details...

Ea$y Pea$y buttons ... prolly take a few bazillion bucks and back ordered

Yes and no. I'm a point of escalation for really big problems that the ops guys can't handle. Fortunately, those are few and far between (1-2 /yr, max). In years past, I was on call a lot more.

I actually used to work for a place whose primary business was providing voip and data over GEO satellite (primarily oil drilling in the Gulf of Mexico, North Sea, Canada, but also Halliburton/KBR in Afganistan and Iraq. In that business the latencies were 600-800+. You had to get used to talking on the phone. We didn't say "over", but the delay was just about like that.

Now I support "Low Latency" or "High Frequency" trading. So we do deal ms, µs, and ns. We have switches that switch at layer 1 that, depending upon the configuration, have single digit nanosecond port to port latency https://www.arista.com/en/products/7130l-series.