Essay by Eric Worrall
Sadly they decided to call the project “Suncatcher” rather than “Skynet”.
Sundar Pichai says Google will start building data centers in space, powered by the sun, in 2027
- Google unveiled Project Suncatcher earlier this month.
- It aims to reduce AI’s environmental impact by relocating data centers in space, powered by the sun.
- Google CEO Sundar Pichai said the company plans to begin sending ‘machines’ to space next year.
The great AI space race has begun.
Google has been quietly working on a long-term research initiative, internally known as Project Suncatcher, to “one day scale machine learning in space.”
Google CEO Sundar Pichai told Shannon Bream on Fox News Sunday thatGoogle’s goal is to start putting data centers in space, powered by the sun, as soon as 2027.
“We are taking our first step in ’27,” he said. “We’ll send tiny, tiny racks of machines, and have them in satellites, test them out, and then start scaling from there.”
In a decade, Pichai said that it’ll be normal to build extraterrestrial data centers.
…
Google’s cosmic pivot comes amid growing global scrutiny over the power demands of data centers.
…
Read more: https://www.businessinsider.com/google-project-suncatcher-sundar-pichai-data-centers-space-solar-2027-2025-11
There might be a few issues with this plan.
Obviously there is the launch cost. Even with the insane cost of data center components, the cost of launching the whole assembly into space adds significantly to the overall cost, along with the cost of sending technicians into orbit to service the equipment.
But the biggest issue with orbital data centers might be heat dissipation. Data centers which consume hundreds of megawatts have to dissipate all of that energy as heat.
Most people think of space as cold, but for heat emitting systems like machines or humans in spacesuits, space functions more like a giant vacuum flask. The kind of vacuum which keeps your Thermos of soup hot all day on Earth completely surrounds anything you put into orbit.
On Earth, data center heat dissipation is a barely manageable problem. Data centers on Earth pump out so much heat, a sizeable portion of the cost of a constructing a data center is building the cooling system.
In space the heat dissipation problem will be far worse – you can’t just dump all the heat into a convenient large body of water located next to your data center, all the heat has to be radiated into the vacuum of space without any help from air convection.
I read the Google press release on their new orbital data center idea, and not once did they mention heat dissipation or cooling.
Why is Google advocating such unlikely sustainability ideas? First claiming data centers will be powered by nuclear fusion in the next decade, now claiming they’ll put all the data centers in space and power them with solar energy?
These grandiose plans might just be a manifestation of late stage investment bubble fever. But I wonder how bad Google’s internal divisions are, whether Google management is catching a lot of internal heat over their skyrocketing data center CO2 footprint? Perhaps these wild sustainability plans are a desperate attempt to pacify Google’s in-house climate worriers – at least until skynet suncatcher is ready to take over all the programming jobs. But that’s just my own wild theory.
Discover more from Watts Up With That?
Subscribe to get the latest posts sent to your email.
“I read the Google press release on their new orbital data center idea, and not once did they mention heat dissipation or cooling.”
Not many non-engineer types know much about conduction, convection, radiation.
Another issue is …
In order to be communicated with from the ground the satellites should be geostationary.
If they’re in Low Orbit you will lose the ability to contact them from time to time.
Also in low orbits they lose sunlight for 45 minutes every 90 minutes as they enter Earth’s shadow. However, at geostationary orbits they will be in Earth’s Shadow for over 10+ contiguous hours every 24 hour period.
It’s my understanding that low-earth orbit (LEO) satellites must have deorbiting capabilities. I’m not sure if this rule applies to the Google scheme.
Not so. At geostationary altitude (22,236 miles above Earth’s equator), the Earth’s disk subtends an angle of only 17° 12′. This, combined with the fact that Earth’s equatorial plane is inclined at about 23.4° to the ecliptic means that there is only a relatively short period of time during a full year when spacecraft in geostationary orbits experience eclipses, and then only for a maximum continuous duration of about 72 minutes any day during the “eclipse seasons”.
See https://antesky.com/how-do-sun-outage-and-eclipses-affect-communication-satellites for the full details
Thank you for the correction. I’ll update my brain accordingly.
So they would still need to carry several GWhs worth of battery back-up and sufficient extra solar panel coverage to recharge them daily
Yes, exactly so . . . assuming there was any real feasibility to the concept in the first place.
Indeed. Even engineers can believe that heat is mainly transferred via radiation, rather than convection (convection is really conducting heat to a moving fluid). Because of this, people can erroneously believe that radiation controls our climate, and the poorly-named ‘greenhouse effect’ is more important than evaporation and convection of water.
It’s quite sad to watch.
But it is quite possible to manage heat in space with efficient radiators and a decent heat pump (like an air conditioner). This could easily cool a processor and transfer the heat to quite large surface areas for dissipation without affecting the processor. . You don’t even really need to worry about the temperature it gets to if an appropriate material is used. The hotter it gets, the better it will radiate.
Maintenance would be a serious problem, however!
The problems with practical heat dissipation in space vacuum AND with incident solar radiation being unavoidable relate to available radiation area, minimizing solar radiation heating of that area, and maximizing its surface emissivity in the direction of its view of deep space. Most operating spacecraft do that very effectively via passive thermal conduction and varying surface emissivities/reflectivities, not by using heat pumps.
A “software engineer” is not an engineer. This is why they used to be called “computer programmers,” which is what they are.
‘Software Engineering’ had the lofty goal of applying engineering principles to the process of creating software. The movement is dead but the title remains.
The original clunky implementations were engineering-like. Now the equation/algorithm bits are buried in a massive pudding of user interface and compatibility features Software engineering team management is not my thing but from nearby it seems like one of the worlds most challenging jobs – how to get a collection of very confident problem solvers with different ideas about the best way to do something to all follow the same rules while nobody will ever look unless something goes wrong? Collaborative SW development takes very particular types of personalities.
It’s not dead. SE created a three-tiered process that is now imposed on HW engineers. The need for certification is annual for all disciplines. It is like IPCC.
As mentioned in the post, people believe space is cold. They don’t think about heat dissipation because they don’t know it’s an issue.
Took a job at a data center bit over a year ago now and let me say, they put out an insane amount of waste heat as Eric Worrall mentioned will need to be efficiently dumped to space. To make all that heat as we know on WUWT they suck down an equally insane amount of power that I doubt we can power without a massive solar array. But here’s the dirty little secret, they also suck up a lot of manpower to keep those racks running and that’s without the stresses of launching a data center into space.
I’m not talking just keeping the heating and cooling running I’m talking the racks themselves need constant attention to keep running. Racks are expensive enough but to have to design them to survive launching and lasting years without human assistance? Sure we’ve been building satellites for years but not with the sort of modern computing power we put into today’s data centers.
Not to mention, they won’t get the bandwidth to and from a satellite that they do on the ground. It’s going to be a bottleneck.
They might get the necessary bandwidth using laser comm.
“Step away from the beam!”
Which of course “climate scientists” refuse to believe can possibly affect thermometers in the vicinity! “No, no!”, they cry “It’s CO2 that makes thermometers hotter! Everybody knows that!”.
Pack of ignorant and gullible fools.
I guess they have armor to stop the micro meteors along with temperature and radiation. But what about all the space junk and orbit decay?
Pass that joint again, dude.
Airbus A320 planes were grounded for software updates due to concerns that cosmic rays could interfere with their navigation systems. This follows an incident where a JetBlue flight experienced a sudden altitude drop, potentially caused by cosmic rays from a distant supernova.
Since i frequently fly w Easyjet that news was rather disconcerting..
If I understand correctly, it was an issue with the fly-by-wire flight controls, not with navigation systems per se.
https://www.flyingmag.com/us-airlines-avoid-impact-airbus-elac-issue/#:~:text=Airbus%20on%20Friday%20said%20that,of%20internal%20and%20external%20suppliers.
My impression was that the “software update” consisted of ditching a previous update.
Back 35 or 40 years ago I was at an NBS (now NIST) conference in Boulder where I met an engineer that had complete a study for NASA on the effects of cosmic rays on the non-error checked computers of that day. If I remember correctly he had found that a Dec PDP-11 would suffer around one undetected hardware error every 8 hours when flown at an altitude of 35,000 feet. I’m a bit surprised that information hasn’t filtered out into the aircraft development and upgrades of today.
In the 80s and 90s I worked on a simulator which used Intel’s first memory chip, the Intel 1001. Sixteen bits of memory (it was from the 60s). The memory chip would catch errors on some random basis. Later when the sim was mothballed I went to work for Intel. There I discovered the memory problem had been a major problem for many years. Older engineers had lead blocks on their desks. Previously used to surround their memory experiments to protect corruption from cosmic rays—which was the proposed problem. The real problem turned out to be radiogenic material in the ceramic package shooting gamma particles through the circuit flipping bits.
Way back when, RCA metal-gate CMOS used to be the most radiation-resistance logic family because the thick metal gates protected the oxide layers underneath. To my knowledge it was used in the Voyager probes which are still operating after almost 50 years in space.
Don’t bogart that joint, my friend, pass it over to me.
“In space the heat dissipation problem will be far worse”
You might as well do the arithmetic. For each sq m that collects 1300 W/m2, you would need 4 sq m of radiating surface at 275K. Difficult, but not impossible. Convection etc has nothing to do with it.
It’s messier than that. The radiator would have to be in shadow, and not too exposed to heat radiating off the back of the solar panels. Possibly a long radiator perpendicular to the solar panels and station, which might make manoeuvring fun, but a large radiator in near Earth orbit would receive substantial upwelling radiation from the Earth, especially when the radiator was parallel to the Earth’s surface.
I think Jerry Pournelle or Larry Niven suggested using streams of magnetically or electrically confined dust to to radiate heat into space.
Eric,
The radiator just has to be not absorbing solar – shiny surface would do. Better still, reflecting toward the solar panels. If it picks up heat radiating off the back, that is just an internal transfer, included in this arithmetic. I think they would avoid near earth orbit.
Nick,
You are just throwing stuff on the wall to see if it sticks.
Have you studied heat sink design? Here is a question. Why do computer CPU’s use fans with heat sinks and grease between the heat sink to CPU interface?
Right ….. it really is simple. You have a flat object which is 1 sq m at the front and 4 sq m at the back. Cool.
TARDIS — Thicker At Rear Doesn’t Impact Spacing
It needn’t be flat, and even if it is, the radiating area could extend beyond the solar panel.
The TRW/Northrop Grumman engineers faced this exact problem on the design of JWST and “solved it” for all practical purposes with their innovative sun shade.
If you are to construct those radiating surfaces out of real materials, then you had better not forget conduction, and likely a coolant circulating through them!
Interestingly there may be some emerging technology which makes all this a lot easier.
There’s a lot of research into alternative semiconductors to silicon chips which can operate at high temperatures, such as Gallium Nitride which can operate up to 480C. The goal of the research is to build electronics for space probes which can operate on the surface of Venus, but semiconductors operating at such high temperatures also potentially simplify cooling problems for Earth based or orbital electronic devices.
But I don’t see any of this becoming anything more than a lab toy in the next decade.
That also assumes 100% efficiency, i.e., no waste heat at that temperature.
It also assume we no longer need those for radar systems.
Yes.
Lack of convection to assist in cooling is what was said.
As for that difficult but not impossible…
you have to keep the orientation sun side for the solar panels and outer space for the 275K panels.An equatorial orbit would put solar panels in the earth’s shadow part of the time and the cooling panels looking at Earth at 255K instead of outer space at 2.3K part of the time which increases the cooling you’ll need during that periodYou could use a polar orbit instead but then you need a lot more fuel to keep orientation facing the sun, you can’t boost or position a body in space on electricity aloneRegardless of what orbit you use, the whole thing has a very hot side and a very cold side. The heat doesn’t get to the cold side to be radiated away by magic. Fluid with a high heat capacity has to be pumped from one side to the other and back again. You could use immersion cooling I suppose which would make that easier but h*llish expensive.Pumping things around in a circle on an object in free fall has consequences, the object wants to spin the other way so you have to compensate for that to maintain positioning. Oops, more fuel.Workloads will have to be kept to latency insensitive workloads (or else you have to have a giant storage array with a copy of the data being processed which adds stupid amounts of weight to the whole thing along with more stupid amounts of cooling).You can make the data storage and servers all solid state but the pumps and/or fans are by definition mechanical devices subject to wear and increased failure without proper servicing. What’s the price of an orbital launch to change a pump motor?I’m late for dinner otherwise I would keep on going. This is an idea requiring multiple Nobel prize breakthroughs to work. I call BS, this is a marketing stunt, hey look at this shiny object just ignore our billions being poured into NG gen sets.
Of course it’s nonsense. All of this tonnage has to be delivered into high orbit. At a cost of $1400 to $3000 per kilogram. And any construction crews. And any tooling needed for assembly in orbit. And it in future years, it has to be serviced, maintained and repaired as you rightly pointed out in the cost to change a pump motor.
At geostationary orbits they would likely spend 10 hours or more in the shadow of earth’s night side. They would need sufficient batteries to supply these power hours plus sufficient extra solar panels to recharge those batteries during the 12-14 hours available daily sunlight time.
I’m not buying that 10 hours in the shade thing. Geostationary orbit is 36,000km. Earth’s radius is 6,800km. Draw yourself a scale model and show me how an object at a distance 6x the radius of the Earth is in the shade for almost half the time. Excepting for the equinox periods, there is almost no shade due to the 17° tilt of Earth’s axis. The sin of 17° is about 0.3, thus at 6x the 6,800km radius, the deflection of the satellite is (6,800km * 6 * 0.3). If the shadow of Earth has a radius of 6,800km, the satellite is 12,000km from the center, or 6,000km outside the shadow at the peak. So the satellite is only ever shaded at all less than half the time, then you follow the 0.707 rule.
“You might as well do the arithmetic. For each sq m that collects 1300 W/m2, you would need 4 sq m of radiating surface at 275K.”
Did you account for radiating surfaces that are close and parallel, like aluminum plates on a conventional heat sink, recollecting any energy? I suppose the data center could have one giant shark fin, or it could be the six orthogonal surfaces of a cube. The cube would work at 50% efficiency unless you can encourage the heat to only radiate outward.
Really hot electronics cool by something-but-not-water in coppers pipes circulated by pumps.
“Water doesn’t freeze instantly in space; instead, it will first boil and freeze simultaneously. The low pressure of space causes immediate boiling, which is a form of rapid evaporation that cools the remaining liquid until it freezes into a cloud of ice crystals. This process is called sublimation once the water has turned to ice. However, if a large mass of water is released, it will take a very long time to freeze completely due to water’s high heat capacity and the limited surface area-to-volume ratio.”
Not saying it cant be done.
I’d love to be on the engineering team.
It may be technically feasible, however it is hugely expensive. The extra weight, plus the cost of the mechanical systems needed to concentrate the heat so that it can be radiated more efficiently.
Another pathetic attempt to distract and delude.
“Gotta reboot those servers. Hmm, not responding.” Sunk cost, and now space junk.
Sorry Dave, I can’t do that…
Dai-sy, Dai-sy, give me your answer true…
I blame Musk. Damn him and his cheap space access! Damn him!
You’re right.
His Mars ambitions(I hope he is aware of the 95% co2 climate catastrophe up there that turned the planet red) inspired many lunatics to do things in outer space that can be done on earth for less than 1% of the cost and with the ability to act instantly when problems occure.
IMO they could also save tons of money if they’d use a different approach,
but I won’t give them any ideas just in the highly unlikely case that I’m right. Let them burn money.
If they’re using their own money, let them do it. That’s what exploration and progress are all about.
Well,some basic common sense should always be part of the equation.
But as I wrote – let them burn money.
Meanwhile Google just announced a major data center project near Memphis, not space.
Google is building two Data Centers here in my county.
We have a coal-fired power plant located nearby and I imagine that’s why they are building here.
In space, no one can hear the AI scream.
Anyone mention radiation and bit flipping?
Will the chips be the large transistor format, being less susceptible to high energy particles, or will they be the latest technology, all so tiny that one high speed electron could change your climate model into garbled noise.
Maybe they are already doing this and nobody noticed?
Mike’s Cosmic Nature Trick?
That deserves many upvotes!
There are processes – look up Single Event Effects. Satellites have to deal with lonely protons flying through space at near light speed.
“Single event effects (SEEs) in aerospace are the result of a single high-energy particle striking an electronic component, which can cause either temporary (soft) or permanent (hard) malfunctions, such as data corruption, functional interrupts, or destructive failures like latch-up or burnout. These effects are a major concern for electronic reliability in space and at high altitudes, necessitating rigorous radiation testing and mitigation strategies to ensure mission success.”
Handled with parity bits. With a single parity bit, any single bit error can be detected and flagged. With two parity bits, any single error bit can be detected and corrected, 2 bit errors can be detected and flagged.
https://www.numberanalytics.com/blog/ultimate-guide-to-parity-bits-in-digital-logic
Yes. The point was… someone thought of it a long time ago, gave it a name, came up with a workable solution then retired to golf in Florida. (The solution usually involves incorporating a metal layer in a semiconductor package to take the bullets plus using statistics to say lightning won’t strike twice in the same place)
Are you positive?
As positive as a Positronic Brain
proton… positive… then an Asimov Joke … haha/groan etc
Don’t be so negative.
Did anyone mention ECC?
“An error correcting code (ECC) is an encoding scheme that transmits messages as binary numbers, in such a way that the message can be recovered even if some bits are erroneously flipped. They are used in practically all cases of message transmission, especially in data storage where ECCs defend against data corruption.“
— https://brilliant.org/wiki/error-correcting-codes (my bold emphasis added)
In general, all semiconductor devices degrade with radiation exposure, to include in this case the solar cells. Lots of engineering can be/has been done to reduce degradation rates, but it can’t be reduced to zero.
In addition to the problems others have raised they must rotate the orbit of this monsterous array a little daily to keep the orbit perpendicular to the sun.
Wouldn’t it be easier to just commit to nuclear fission. It has been incrediably reliable for 70 years. It would certainly be more reliable and lower cost than this and other crazy schemes.
Agree. However if you asked me in 1990ish whether I wanted to blow a few trillion in borrowed government money on slowly rotating space computers or on super fast ground computers that make nothing but bad weather forecasts…
(intended-to-be-obvious ending: I’d do the space computers)
Let’s see, launching the mass of a typical 1 GW nuclear power plant downsized to just 200 MWe but then upsized since water cooling is not a option into space would be . . . ummm, carry the seven, watch the decimal point . . . yeah, just not doable. /sarc
If instead you meant “just commit to building a nuclear fission power plant dedicated to a AI data center or two” on land, then you have to first find an acceptable site for such, do the necessary EIS work and filing, amass the upfront $billions in funding (preferably not from taxpayers), and then wait some 8-10 years for it to get built, inspected and thoroughly tested prior to bringing it into commercial service.
In terms of “easy”, nothing is so if it involves politics and a new nuclear power plant dedicated solely for AI use will stink of such.
Please delete this. It wasn’t helpful.
Just build an SMR – it can’t possibly be more expensive than this nutty notion.
It seems that google never heard of the supercheap renewable energy we have down here on earth.
If they had they’d never do such a gigantic carbon footprint thing to launch a few tons of future trash into space.
And how sustainable is something where 100% of the materials get lost foreever?
The whole process of launching all this stuff into space will probably need more energy than they’ll save in those years before the system breaks down.
And whenever there is a physical problem you need to waste tons of energy to sent someone up there.
That’s a bit less sustainable than letting someone walk 5 meters from his chair to the server.
Therefore my guess is that this is a stealth military thing.
“It seems that google never heard of the supercheap renewable energy we have down here on earth.”
That made me laugh! 🙂
Maybe Pichai is indirectly saying putting
data centers in orbit is more soluble a problem than making wind and terrestrial solar practical as a grid level power source?
It is just about virtue signalling under the header: innovation. That usually involves a lot of silly ideas. A lot of ideas stay at the table where they were constructed..
Thousands of square meters of radiators in space is even dumber than thinking sunshine and breezes can cost-effectively power a modern society, much less even a fourth of primary energy.
I’m sure the Chinese wouldn’t jeopardise this multi-billion-dollar project once it’s up and running by “accidentally” losing control of a satellite and causing a collision. And if we’re talking a data-centre-sized data centre, it’ll be Gravity only a hundred times worse. Goodbye GPS, Starlink, and everything else.
Eric, a vacuum keeps nothing hot – or cold. No offence intended, but 300,000,000 km of vacuum has no insulating effect between the Earth and the Sun. Nor does the vacuum surrounding the Earth prevent the surface cooling each night, or getting hot during the day.
The vacuum flask has two maximally reflective surface separated by a vacuum. The vacuum is just to prevent convective and conductive heat loss.
The physics is pretty simple, but widely misunderstood. I guess the name “vacuum flask” leads people to think that the “vacuum” insulates, when it doesn’t.
Unfortunately, Google might believe its AI which hallucinates “a vacuum is an excellent insulator.” No it’s not. The speed of light is specified in a vacuum – the absolute least impediment to energy transfer. Google AI even says “an insulator is a material that resists or blocks the flow of energy . . . “.
Which Google AI statement do you prefer?
I believe the 1st version – the one that keeps the coffee in my thermos hot for 12 hours.
While we have here a bit of greenhouse logic ( It’s not the co2 gas that keeps the greenhouse warm) vacuum seems to work fine as insulator (assuming thermos has no secret heating system inside) in combination with other stuff.
On the other hand – as vacuum is kind of nothing but empty space(quantum yadda aside)
and space is said to have no properties(like being bend by gravity),vacuum does exactly… nothing
Exactly. Is nothing, does nothing, and Nature doesn’t abhor it.
Absolutely no conduction in a vacuum. That’s definitely resisting the flow of energy in my opinion.
Radiation in a vacuum flask is miniscule compared to conduction & more importantly convection, just as it is in the climate.
There’s also no convection in a vacuum, nor in a zero gravitational field…
One of my genius friends in grad school was conducting crystallization experiments on the Space Shuttle. No gravity, no convection in the liquid phase, more perfect crystals.
A common misperception.
The ISS, at its nominal altitude of 400 km above Earth’s surface, and its inhabitants experience a downward gravitational acceleration of about 0.89 g, but the station itself is simultaneously accelerating downward at 0.89 g as a result of it orbiting in continuous “free fall”. Everyone and everything inside the station experiences the same gravity and “free fall”, with the combined effect being effective weightlessness.
Indeed. I suppose I should have written “no gravitational force”, due to being in free fall. Same as the trip down from the leaning tower of Pisa.
Even a free-falling object—be a ball dropped off the Tower of Pisa or the ISS—experiences continuous gravitational force. That is the reason that they accelerate continuously.
Keep in mind that acceleration is actually a vector (magnitude and direction). In the case of a ball dropped from a tower, that vector is normal to the flat ground that the ball eventually hits. However, in the case of ISS (and all other orbiting satellites), the gravity vector is continuously rotating about the inertial frame of reference of the satellite . . . rotating through a full 360 degrees over one orbital period, without the satellite ever hitting something to stop its free fall.
Just a slight correction. There is no conduction across a vacuum. Conduction within a homogeneous material can occur in a vacuum.
Jim,
Yes, the Earth floats in a vacuum. So does the Sun, I guess.
Ah, the wonder of words! I love it!
And your opinion is as worthless as mine – in my opinion of course. In fact, a vacuum does not impede the flow of energy at all – a vacuum is literally nothing.
The terms conduction and convection are terms originally coined by people who were ignorant of the laws of physics as currently understood. Good for practical use, but provide precisely no understanding of the mechanism which leads to their observation. Sorry, sounds a bit complicated – and it is.
If you could find reproducible experimental support for your speculation, I would accept it, of course. There is none, as far as I know.
Similarly, the opinion that adding CO2 to air makes thermometers hotter is totally unsupported by reproducible experiment. Nonsense, in other words. Delusion.
All good fun,<g>
The thermal principle of the Dewar flask is vacuum. Most Dewar flasks have metal linings metal but are not “maximally reflective”. You can look this up.
Randle, a vacuum is not a “thermal principle”.
You say –
More word salad. You, like other ignorant and gullible people, seem to believe that the Earth is insulated from the effects of sunlight by 300,000,000 kms of fairly hard vacuum.
I suppose you are going to claim that you were talking about a specific type of construction which uses aluminium foil in lieu of the maximally reflective surface obtained by “silvering” an internal glass surface.
Just because you can read nonsense on the internet does not necessarily mean that it is true. However, Google AI grudgingly admits –
Rather dodges the question, as manufacturers could intentionally make the silvering less than maximally effective, I suppose.
Maybe you could “look this up”, and let me know. Feel free to waste as much time as you like.<g>
Michael – you can stuff it.
Out of this world!
From the article: “Sundar Pichai says Google will start building data centers in space, powered by the sun, in 2027”
Good! That means the U.S. will beat the Chicoms by three years. The Chicoms claim they are going to have a working, demonstration solar power satellite in orbit by 2030.
I think Solar Power in space is the future for the human race. I think humans living in space is the future of the human race.
But Eric does have a point about being skeptical about the Data Centers dissipating their heat in space. I would be curious to see how they intend to solve this problem.
That’s good. Does the US give itself a certificate and a handsome trophy for winning the race?
Only joking, but seriously, I fail to see the point. In my worthless opinion, the preoccupation with “winning” has been responsible for more avoidable death and misery than anything else I can think of.
It all seems a bit silly when you buy an article labelled “Proudly American since 1927”, and find it’s “Made in China”. Who won? The US company, or the Chinese company that produced the goods? Should I care?
There is one option Google is not mentioning.
They could leave the Data Centers on Earth, and use Solar Power Satellites to beam power down to them.
Past studies have said the microwave beam coming through the atmosphere from space would not be harmful to living creatures. Although those studies were done some years ago. I haven’t seen anything lately on it.
It would definitely be the way smarter option to keep the supercomplex stuff that works on a nanolevel,and therefore needs high levels of maintenance, down on earth.
But imo the whole idea is totally obsolete if Russias nuclear rocket is a real thing.
As this would mean that the energy problem is kind of solved.
The practical problems of solar power satellites are manifold and have been well documented.
Fundamentally, an annual average solar power flux of 1361 W/m^2 is just too low to provide many terawatts of power.
Global electricity generation in 2024 was 30,850 terawatt-hour (TWh).
— https://en.wikipedia.org/wiki/Template:Latest_pie_chart_of_world_power_by_source
Averaged over the course of a year, that’s equivalent to a continuous use of 3.52e12 watts.
Let’s assume that we can convert sunlight to DC electricity to satellite-emitted microwave power to ground-received microwave power to ground received DC electricity to grid-compatible AC electricity at an estimated efficiency train of (0.5*0.9*0.9*0.9*0.85) = 31%. To supply Earth with just 1% of its current demand for electricity using SPS based on the most advanced solar cell technology would require a total in-space solar array area of (3.52e12/(0.31*1361)) = 8.3e9 m^2, or about 8300 km^2 (about 3,220 square miles). YIKES!
Even SpaceX with their massive Starship launch system can’t help out here.
You hit that nail squarely on the head with a large hammer!
My apologies to all . . . I made an inadvertent math mistake in my above calculation for required solar array size.
I stated the calculation of solar array area was to supply just 1% of Earth’s current demand for electricity (1% of an average steady-state power of about 3.52e12 watts) but I neglected to multiply the total demand by 0.01 in my shown calculation to reflect that 1%.
Therefore, with stated assumptions and efficiencies, the total in-space solar array size to supply just 1% of today’s worldwide electrical demand would be around 83 km^2 or about 32 square miles.
So, “Yikes” instead of “YIKES!”. I still stand by my bottom line.
Does this fit in the class of Google claims that go exactly nowhere, like Google Wave and Buzz and various hardware claims like cars.
Just because one is very knowledgeable about AI it doesn’t mean he/her is very smart in the common sense of that term.
Please allow me to add to Eric’s spot-on comment given in the above article:
“But the biggest issue with orbital data centers might be heat dissipation. Data centers which consume hundreds of megawatts have to dissipate all of that energy as heat.”
But before those data centers can generate that steady-state operational waste heat, they first have to receive at least that much power. And for technology maturity, reliability and convenience reasons that will most certainly have come from PV conversion of solar radiation at the design orbiting altitude(s).
What’s wrong with this?
1) In geostationary orbits, Earth eclipses the Sun at certain times during a full year. Such “eclipses usually occur around the spring and autumnal equinoxes, and each time they occur continuously for 45 days, a total of 90 days, and the 2 days of spring and autumnal equinoxes have the longest eclipse duration of 72 minutes.” — source: https://antesky.com/how-do-sun-outage-and-eclipses-affect-communication-satellites . So, how will any orbiting data center keep operating during those power blackouts . . . anyone think they’ll be flying batteries sized for several hundred MWh capacity? Or maybe Google CEO Sundar Pichai believes that there will be SMRs or even compact fusion reactors—qualified and approved for launch and space use and with operating fuel loads sufficient to last, say, 40 years—within his stated timeline of “in a decade”. The alternative is to not place the orbiting data centers in a geostationary orbit but to instead use essentially Earth polar orbits to avoid eclipses . . . but then all ground stations communicating with those data centers will need to have steerable RF antenna arrays or steerable laser receivers to point to and track the movement of those spacecraft across the sky to send/receive data . . . KA-CHING! Good grief!
2) Then there are the fundamentals of available solar energy flux. Solar insolation at Earth’s average distance from the Sun is about 1361 W/m^2. And currently the most efficient solar cells—achieved and demonstrated in laboratories but not in the field—have only about 40% efficiency under standard sunlight . . . see https://commercialsolarguy.com/50-efficient-solar-cell . Let’s assume that within the next decade we can produce and fly in space reliable, long life solar cells with 50% conversion efficiency. In such a case, to produce 200 MW of electricity one would a solar cell area of (200e^6/(0.5*1361)) = 294,000 m^2, equivalent to a single square array of solar cells of 542 m (about 1/3 mile) length on a side. Good grief^2!
3) If we wire the solar cells in such arrays to provide a “reasonable” working DC voltage of, say, 700 vdc and limit the size of the wires to that used commonly in household electric wiring to wall receptacles (14 gauge, rated for 15 amp max load), we are talking about using (200e^6/(700*15)) = 19,050 such wires to carry 200 MW of electricity steady state. Wow . . . that’s many tons of copper! And there are all the DC-DC converters (and their own conversion inefficiencies, not considered above) with down converting the (presumed) 700 Vdc transmission down to the +3.3 Vdc, +5 Vdc, +12 Vdc, and -12 Vdc voltages typically used for computer motherboards (ASIC chips use power provided at +5 Vdc). Why not just wire the solar arrays to provide 5 Vdc directly? Well to transmit 200 MW of electricity at 5 Vdc would mean transferring a total of 40,000,000 amps over an average distance of about ((542/2)/2) = 135 m = 38 feet, assuming two equal size square solar panels were used on opposite sides of the orbiting data center.
So how are those orbiting data centers going to be able to steer (i.e., maintain Sun pointing) of such massive solar arrays without consuming massive amounts of propellant for reaction mass via rockets to provide the required torque? Oh, yeah, the launching of autonomous on-orbit-refueling tanker spacecraft, also to be developed and proven “in a decade” . . . /sarc
Despite Sundar Pichai’s beliefs and optimism (or is it stock pumping?), I can confidently predict there’ll be no orbiting data centers (of even the tens of megawatts size) in a decade.
Even according to Google AI –
The devil in in the details – thermal management is a major detail.
I was involved with several HEL (High Energy Laser) tactical level demonstration programs; thermal management was always a day late and many dollars short.
Google has a corporate tradition of Tesla-like claims that never actually happen.
. . . and Tesla and SpaceX are among the worst in their predictions of the future.
Anyone really believe the Tesla Cybertruck is an outstanding success or that SpaceX Starship HLS will be ready to land humans on the Moon before end-CY2027 (the currently planned Artemis 3 mission)?
ROTFL!
Does this make sense to anyone? I can’t imagine anything being cheaper to operate in space. There are some things that can only be done in space but that is a different issue. At some point we need to think about all the crap we are putting in orbit around the Earth, there has got to be a limit you would think.
When you look up, and don’t know whether it’s day or night because the sky is totally obscured by crap, the limit must be getting close!
I’m optimistic. Space is big, and anything we hurl into space will either fall back to Earth, or zip off into the void, eventually falling on something else.
A perpetually orbiting body is physically impossible, to the best of my knowledge.
No need to panic – yet!