A minor terminology quibble: the video refers to the Nth harmonic as if it's the fundamental frequency times N+1, but it's usually fairly standard to refer to the frequency that's N times the fundamental as the Nth Harmonic. So, the fundamental is the 1st harmonic.
For overtones, there's less of an established standard, but usually the 1st overtone is twice the fundamental, the 2nd overtone is 3x, and so on. (I tend to avoid talking in terms of overtones because of the ambiguity.)
I think that makes it easier for those who are math brained and not creative brained. To understand music theory fully, you need that creative brain. Because we aren’t even talking about resonance harmonics, triplen, or any of the crazy interharmonics.
edit
actually watching again, at the very beginning, he demonstrated resonance harmonics.
If the fundamental is 100hz, then the 1st harmonic is the fundamental (100hz), the 2nd harmonic is 200hz, the 3rd harmonic is 300hz, and so on.
Sometimes the harmonics aren't exact. On a piano, if the fundamental is 100hz then the 2nd harmonic might be, say, 200.1hz or something. Some inharmonic instruments like gongs aren't anywhere close to the "ideal" harmonic series.
This may be overly pedantic (even more so than your correct comments about numbering harmonics and overtones), but in this case the overtones are not harmonics, which, as you say, are by definition multiples of the fundamental frequency (“harmonic series” is a mathematical term). That’s why gongs are “inharmonic”: they have an overtone series that is not a series of harmonics.
Membranes have a harmonic series in two axes - kind of. They have complex (not that kind of complex, although it also is, in a way) modes on a constrained surface which are calculated with Bessel functions.
The other side of this is that if you want your open source project to be useful to people, relevant, included by default in Linux distributions, highly regarded by the community, and so on then it's expected that you treat your users well.
Not every project aspires to such things, but if you do then the path to success requires at a minimum not treating users as a burden.
Some users might be particularly rude or entitled, in which case you can politely decline their feature requests and move on.
Basically, it's never rude for a user to file a bug report or request a feature. It's never rude for the maintainer to decline to implement a feature if they haven't budgeted time (or other relevant resources) to do it, it doesn't align with the fundamental goals or architecture of the project, or they simply don't know how to do it.
It would be rude for a user to demand of maintainers more than they're willing to give, and it would be rude of a maintainer not to be at least somewhat mindful that spending at least a little bit of effort to respond to reasonable requests, fix known bugs, and keep documentation accurate and up-to-date can prevent a lot of random strangers from wasting a lot of time on something that isn't useful to them. No one has any contractual obligation to provide anything, but I think everyone should treat other people's time and attention as a scarce and valuable commodity, not to be wasted.
The main reason is that generating energy in space is very cheap and easy due to how ridiculously effective solar panels are.
Someone mentioned in the comments on a similar article that sun synchronous orbits are a thing. This was a new one to me. Apparently there's a trick that takes advantage of the Earth not being a perfect sphere to cause an orbit to precess at the right rate that it matches the Earth's orbit around the sun. So, you can put a satellite into a low-Earth orbit that has continuous sunlight.
Is this worth all the cost and complexity of lobbing a bunch of data centers into orbit? I have no idea. If electricity costs are what's dominating the datacenter costs that AI companies are currently paying, then I'm willing to at least concede that it might be plausible.
If I were being asked to invest in this scheme, I would want to hear a convincing argument why just deploying more solar panels and batteries on Earth to get cheap power isn't a better solution. But since it's not my money, then if Elon is convinced that this is a great idea then he's welcome to prove that he (or more importantly, the people who work for him) have actually got this figured out.
Let's assume your space solar panel is always in sun - so 8760 kWh per year from 1kWp.
In Spain, 1kWp of solar can expect to generate about 1800 kWh per year. There's a complication because seasonal difference is quite large - if we assume worst case generation (ie what happens in December), we get more like 65% of that, or 1170 kWh per year.
That means we need to overbuild our solar generation by about 7.5x to get the same amount of generation per year. Or 7.5kWp.
We then need some storage, because that generation shuts off at night. In December in Madrid the shortest day is about 9 hours, so we need 15 hours of storage. Assuming a 1kW load, that means 15kWh.
European wholesale solar panels are about €0.1/W - €100/kW. So our 7.5kWp is €750. A conservative estimate for batteries is €100/kWh. So our 15kWh is €1500. There's obviously other costs - inverters etc. But perhaps the total hardware cost is €3k for 1kW of off-grid solar.
A communications satellite like the Eurostar Neo satellite has a payload power of 22 kW and a launch mass of 4,500 kg. Assuming that's a reasonable assumption, that means about 204kg per kW. Current SpaceX launch costs are circa $1500 per kg - but they're targeting $100/kg or lower. That would give a launch cost of between $300k and $20k per kW of satellite power. That doesn't include the actual cost of the satellite itself - just the launch.
I just don't see how it will make sense for a long time. Even if SpaceX manage to drastically lower launch costs. Battery and solar costs have also been plummeting.
Is it reasonable to use Neo as a baseline? Modern Starlink satellites can weigh 800kg, or less than 20% of Neo. I see discussions suggesting they generate ~73kw for that mass. I guess because they aren't trying to blanket an entire continent in signal? Or, why are they so much more efficient than Neo?
Interestingly the idea of doing compute in space isn't a new one, it came up a few years ago pre-ChatGPT amongst people discussing the v2 satellite:
Still, you make good points. Even if you assume much lighter satellites, the GPUs alone are very heavy. 700kg or so for a rack. Just the payload would be as heavy as the entire Starlink satellite.
You can't increase the size of the radiator and reduce the mass of the satellite. How is that supposed to work?
You're also forgetting that Starlink satellites aren't in a sun synchronous orbit which means they have to overbuild the energy generation capacity (low capacity factor) and can simultaneously take advantage of earth's shadow to cool down.
Droplet radiators can theoretically do this. The radiator is made up of extremely fine liquid droplets expelled from what is basically a big space shower head. The droplet cloud has a big surface area so more heat can radiate. The droplets are collected in a sort of drain the other side. The idea has been around for decades but there are lots of practical problems to work out, like minimizing losses due to splashes or droplets heading in the wrong direction (e.g. using ferrofluids and magnetic containment). It's never been worked on seriously because conventional radiators were always enough.
With droplet radiators increasing the effective size means using a bigger head/drain and longer booms to expand the distance between them, so the scaling properties are different to pipe based radiators.
That could be one reason they want to do it. Maybe by using data from Palantir or harvested from Elon's work with DOGE, along with twitter user data and whatever else they can get, they want their AI to be the all-seeing eye of Sauron. (Which isn't too far from what the whole ad-tech industry is about in the first place.) Or they want to make sexually explicit deepfakes of everyone Elon doesn't like. Or they want to flood the internet with AI generated right-wing propaganda.
That's not a new problem that no one has dealt with before. The ISS for instance has its External Active Thermal Control System (EACTS).
It's not so much a matter of whether it's an unsolvable problem but more like, how expensive is it to solve this problem, what are its limitations, and does the project still makes economic sense once you factor all that in?
It's worth noting that the EACTS can at maximum dissipate 70kW of waste heat. And EEACTS (the original heat exchange system) can only dissipate another 14kW.
That is together less than a single AI inference rack.
And to achieve that the EACTS needs 6 radiator ORUs each spanning 23 meters by 11 meters and with a mass of 1100 kg. So that's 1500 square meters and 6 and a half metric tons before you factor in any of the actual refrigerant, pumps, support beams, valve assemblies, rotary joints, or cold side heat exchangers all of which will probably together double the mass you need to put in orbit.
There is no situation where that makes sense.
-----------
Manufacturing in space makes sense (all kinds of techniques are theoretically easier in zero G and hard vacuum).
Mining asteroids, etc makes sense.
Datacenters in space for people on earth? That's just stupid.
Your calculations are based on cooling to 20c, which is exponentially harder than cooling to 70c where GPUs are happy. Radiators would be roughly 1/3 the size of the panels for 70c.
I get that vacuum is a really good insulator, which is why we use it to insulate our drinks bottles. So disposing of the heat is a problem.
Can't we use it, though? Like, I dunno, to take a really stupid example: boil water and run a turbine with the waste heat? Convert some of it back to electricity?
It's a good question, but in a closed system (like you have in space) the heat from the turbine loop has to go somewhere in order to make it useful. Let's say you have a coolant loop for the gpus (maybe glycol). You take the hot glycol, run it through your heat exchanger and heat up your cool, pressurized ammonia. The ammonia gets hot (and now the glycol is cool, send it back). You then take the ammonia and send it through the turbine and it evaporates as it expands and loses pressure to spin the turbine. But now what? You have warm, vaporized, low pressure ammonia, and now you need to cool it down to start over. Once it's cool you can pressurize it again so you can heat it up to use again, but you have to cool it, and that's the crux of the issue.
The problem is essentially that everything you do releases waste heat, so you either reject it, or everything continues to heat up until something breaks. Developing useful work from that heat only helps if it helps reject it, but it's more efficient to reject it immediately.
A better, more direct way to think about this might be to look at the Seebeck effect. If you have a giant radiator, you could put a Peltier module between it and you GPU cooling loop and generate a little electricity, but that would necessarily also create some waste heat, so you're better off cooling the GPU directly.
I think I get it. If we could convert 100% of the waste heat into useful power, then all good. And that would get interesting because it would effectively become "free" compute - you'd put enough power into the system to start it, and then it could continue running on its own waste heat. A perpetual motion machine but for computing.
But we can't do that, because physics. Everything we could do to generate useful energy from waste heat also generates some waste heat that cannot be captured by that same process. So there will always be some waste heat that can't be converted to useful energy, which needs to be ejected or it accumulates and everything melts.
What do you do with the steam afterwards? If you eject it, you have to bring lots of it with your spacecraft, and that costs serious money. If you let it condensate to get water again, all you did is moving some heat inside the spacecraft, almost certainly creating even more heat when doing that.
However there are workarounds. People are talking like the only radiator design is the one on the ISS. There are other ways to build radiators. It's all about surface area. One way is to heat up a liquid and then spray it openly into space on a level trajectory towards a collecting dish. Because the liquid is now lots of tiny droplets the surface area is huge, so they can radiate a lot of heat. You don't need a large amount of material as long as you can scoop up the droplets the other end of the "pipe" and avoid wasting too much. Maybe small amounts of loss are OK if you have an automated space robot that goes around docking with them and topping them up again.
The ISS consumes roughly 90kW. That’s about *one* modern AI/ML server rack. To do that they need 1000 m^2 of radiator panels (EACTS). So that’s the math: every rack needs another square kilometer of stuff put into orbit. Doesn’t make sense to me.
And what happens every time a rack (or node) fails? Does someone go out and try to fix it? Do we just "deorbit" it? How many tons per second of crap would we be burning in the upper atmosphere now? What are the consequences of that?
How do the racks (or nodes) talk to eachother? Radios? Lasers?
What about the Kessler Syndrome?
Not a rocket scientist but 100% agree this sounds like a dead end.
Communication is a well-understood problem, and SpaceX already has Starlink. They might need pretty high bandwidth, but that's not necessarily much of a problem in space. Latency could be a problem, except that AI training isn't the sort of problem where you care about latency.
I'd be curious where exactly they plan to put these datacenters... In low Earth orbit they would eventually reenter, which makes them a pollution source and you'd have no solar power half the time.
Parking them at the Earth-Sun L1 point would be better for solar power, but it would be more expensive to get stuff there.
Seasons mess that up unless you're burning fuel to make minor plane changes every day. Otherwise you have an equinox where your plane faces the sun (equivalent to an equatorial orbit) and a solstice where your plane is parallel to the sun (the ideal case).
Huh, I didn't know that that was possible without burning fuel. Kind of wild that it only works because the Earth has an equatorial bulge and isn't an exact sphere.
Honestly, there's not a lot else I can think of if your goal is find some practical and profitable way to take advantage of relatively cheap access to near-Earth space. Communication is a big one, but Starlink is already doing that.
One of the things space has going for it is abundant cheap energy in the form of solar power. What can you do with megawatts of power in space though? What would you do with it? People have thought about beaming it back to Earth, but you'd take a big efficiency hit.
AI training needs lots of power, and it's not latency sensitive. That makes it a good candidate for space-based compute.
I'm willing to believe it's the best low-hanging fruit at the moment. You don't need any major technological advances to build a proof-of-concept. Whether it's possible for this to work well enough that it's actually cheaper than an equivalent terrestrial datacenter now or in the near future is something I can't answer.
We have radiators on the ISS. Even if you kept the terrible performance of those ancient radiator designs (regularly exposed to sunlight, simplistic ammonia coolant, low temperature) you could just make them bigger and radiate the needed energy. Yes it would require a bit of engineering but to call it an "unsolved problem" is just exaggerating.
It's a solved problem. The physics is simply such that it's really inefficient.
> ... we'd need a system 12.5 times bigger, i.e., roughly 531 square metres, or about 2.6 times the size of the relevant solar array. This is now going to be a very large satellite, dwarfing the ISS in area, all for the equivalent of three standard server racks on Earth.
The gist of it is that about 99% of cooling on earth works by cold air molecules (or water) bumping into hot ones, and transferring heat. There's no air in space, so you need a radiator 99x larger than you would down here. That adds up real fast.
That’s the secret plan - cover LEO with solar cells and radiators, limiting sunlight on the ground, rendering ground base solar ineffective, cool earth and create more demand for heating; then sell expensive space electricity at a huge premium. Genius!
I think you may be thinking of cooling to habitable temperatures (20c). You can run GPUs at 70c , so radiative cooling density goes up exponentially. You should need about 1/3 of the array in radiators.
Bezos has been pushing manufacturing-in-space for a long time, as a ideal candidate for what to do in space that you might prefer to not do on Earth. Robotics, AI automation, manufacturing - combo it in space, let the robots manufacture for us in space. Abundant energy, low concerns about most forms of pollution. We'll need to dramatically improve our ability to transit mass to and from cheaply first of course (we're obviously talking many decades into the future).
> Bezos has been pushing manufacturing-in-space for a long time, as a ideal candidate for what to do in space that you might prefer to not do on Earth. Robotics, AI automation, manufacturing - combo it in space, let the robots manufacture for us in space.
LOL, this seems so far off from the reality of what manufacturing looks like in reality.
- sending raw materials up there
- service technicians are necessary ALL THE TIME, in fully automated production lines
- sending stuff back down
Maybe I lack vision, but data centers in space is a 1000x times better idea and that is already a terrible idea.
Space manufacturing is a real thing, there are already companies trialling it. The factory is small, satellite sized, and it deorbits when the manufacturing run is done. The results are protected enough for them to be picked up from Earth.
The justification (today) is that you can do very exotic things in zero-G that aren't possible on Earth. Growing ultra-pure crystals and fibre optics and similar.
Ok, that I might buy. If there is a product one can build in zero-G that one cannot build on earth. Especially something like growing crystalls. Sure. But trying to compete with something that can just as well be build on earth on the premise that it will be cheaper to do the same thing just in space is insane.
It's the same issue that I have with data centers in space. I don't think there is any big technical hurdle to send a GPU rack into space and run it there. The problem is that I have a hard time to believe it is cheaper to run a datacenter in space. When you have to compete solely on cost, it will super hard.
I don't think it's insane. It might not work or be competitive but it's not obviously insane.
In a frictionless economy governed by spherical cows it'd be insane. But back here on Earth, AI is heavily bottlenecked by the refusal or inability of the supply chain to scale up. They think AI firms are in a bubble and will collapse, so don't want to be bag holders. A very sane concern indeed. But it does mean that inferencing (the bit that makes money) is constantly saturated even with the industry straining every sinew to build out capacity.
One bottleneck is TSMC. Not much that can be done about that. The other is the grid. Grid equipment manufacturers and CCGT makers like Siemens aren't spinning up extra manufacturing capacity, again because they fear being bag holders when Altman runs out of cash. Then you have massive interconnection backlogs, environmentalists attacking you and other practical problems.
Is it easier to get access to stable electricity supplies in space? It's not inconceivable. At the very least, in space Elon controls the full stack with nearly no regulations getting in the way after launch - it's a pure engineering problem of the sort SpaceX are good at. If he needs more power he can just build it, he doesn't have to try and convince some local government utility to scale up or give him air permits to run generators. In space, nobody can hear you(r GPUs) scream.
> "At the very least, in space Elon controls the full stack with nearly no regulations getting in the way after launch - it's a pure engineering problem of the sort SpaceX are good at. If he needs more power he can just build it, he doesn't have to try and convince some local government utility to scale up or give him air permits to run generators. In space, nobody can hear you(r GPUs) scream."
Wouldn't he be able to float solar panels and GPUs out into international waters and run them on cargo ships powered by bunker fuel much (much much) cheaper than launching them into space?
Building nuclear-powered and solar powered datacenters in places with low population density will still be cheaper. Do you think Mongolian government won't allow China to build datacenters if the price is right?
It might be easier in China but that doesn't help Elon or Americans.
Solar powered datacenters on Earth don't make sense to me. The GPUs are so expensive you want to run them 24/7 and power cycling them stresses the components a lot so increases failure rate. Once it boots up you need to keep the datacenter powered, you can't shut it down at night. Maybe for CPU datacenters solar power can make sense sometimes, but not for AI at the moment.
Nuclear is super hard and expensive to build. It probably really is easier to put servers in space than build nuclear.
It’s not necessarily cheaper energetically to get stuff from an asteroid than from Earth. You’d have to accelerate stuff from a wildly different orbit, and then steer it and slow it down. Metric tonnes of stuff. It’s not physically impossible, but it is wildly expensive (in pure energy terms, not even talking about money) and completely impractical with current technology. We just don’t have engines capable of doing this outside the atmosphere.
> It’s not necessarily cheaper energetically to get stuff from an asteroid than from Earth. You’d have to accelerate stuff from a wildly different orbit, and then steer it and slow it down.
Delta V from just about anywhere in the solar system is lower than launching from the surface of Earth. You could launch stuff from Mars and bring it back to Earth orbit with less energy than launching it from Earth. The rocket equation is really punishing.
Right. The alternative is not to send materials from Earth for processing in space, that would be stupid. We send finished stuff, which were manufactured on the ground. But you don’t mine finished widgets from asteroids. You mine ore that needs refining and processing before being used to manufacture things. This ore is orders of magnitude heavier than the finished products, never mind all that’s required to do anything useful with it.
I think you might have no sense of what it takes to go from a raw mined material to something that can be used in a factory. I am not saying it cannot be done. I am just saying it cannot be done in a way that is cheaper than on earth.
The show For All Mankind kind-of hinted at how the labor problem would be solved: recruit like the military and promise huge bonuses that will probably not be realized because space is risky business
I think it makes more sense if you invert the manufacturing cycle.
Automated asteroid mining, and asteroid harvesting, are potential areas where we have strong tech, a reasonable pure automation story, and huge financial upsides. Trillion dollar asteroids... If we’re sourcing metals out there, and producing for orbital operations or interplanetary shenanigans, the need for computing and automation up there emerges.
And I imagine for the billionaire investor class now is the window to make those kinds of plays. A whole set of galactic robber barons is gonna be crowned, and orbital automation is critical to deciding who that is.
When Bezos first mentioned drone delivery, many intelligent, serious people laughed at it and accused of Bezos running out of ideas as Amazon was stagnant
That is a fun thought experiment, as we wouldn't want to manufacture too far away from earth we may still be within the earth's atmosphere. I wonder what effect dumping greenhouse gases into the very upper levels of the atmosphere would have in comparison to doing it lower down. My assumption is it would eventually sink to a lower density layer, having more or less the same impact.
Do they need to be maintained? If one compute node breaks, you just turn it off and don't worry about it. You just assume you'll have some amount of unrecoverable errors and build that into the cost/benefit analysis. As long as failures are in line with projections, it's baked in as a cost of doing business.
The idea itself may be sound, though that's unrelated to the question of whether Elon Musk can be relied on to be honest with investors about what their real failure projections and cost estimates are and whether it actually makes financial sense to do this now or in the near future.
AI clusters are heavily interconnected, the blast radius for single component failure is much larger than running single nodes -- you would fragment it beyond recovery to be able to use it meaningfully.
I can't get in detail about real numbers but it's not doable with current hardware by a large margin.
Lasers, space, super geniuses, and most importantly money. You're worrying too much about the details and not enough about the awesomeness.
But seriously, why are all the stans in these comments as unknowledgeable as Elon himself? Is that just what is required to stan for this type of garbage?
https://en.wikipedia.org/wiki/Forestiere_Underground_Gardens
reply