A race of beings decides over a few pan-galactic gargleblasters one evening that rather than waste scientific minds on developing hyper-advanced technology, they should just build a huge computer to design clarkean devices for them. A computer of this size naturally requires a huge power source, so this is what they come up with:

An enormous spherical computer is built, consisting of trillions upon trillions of densely-packed processors each the size of a brain cell. This processing layer surrounds a huge reservoir of water. The computer is placed in orbit around a large body such as a gas giant, and due to tidal flexing, the fluid heats up.

The difficult part: I had originally planned for the hot water to be channeled up from the core, turned into steam and used to drive fans and so power the processing layer, until I realised that this would then involve venting the steam into space, and thus slowly depleting the core of fluid. So, is there another way to produce a giant supercomputer powered by tidal flexing?

  • 5
    $\begingroup$ Just a note that thermodynamically speaking, the energy that you get from this scheme is a tiny fraction of the energy it would have taken to put the thing in orbit in the first place - and in fact you're really just getting back a little bit of the energy you originally put in. (This doesn't apply if the machine was built in orbit though. If it was constructed from a moon that was there already then there's no problem.) $\endgroup$
    – N. Virgo
    Jan 30, 2023 at 9:34
  • 2
    $\begingroup$ What is wrong with simply using Matrioshka brain (en.wikipedia.org/wiki/Matrioshka_brain)? Your tidal energy doesn't seem sufficient for your goal. Also, your current plan has issues with COOLING of your supercomputer. Heat is not your friend in computing. $\endgroup$
    – Negdo
    Jan 30, 2023 at 12:23
  • $\begingroup$ How enormous is enormous? $\endgroup$
    – RonJohn
    Jan 30, 2023 at 18:29
  • $\begingroup$ If it has steep enough gravity well to have atmosphere then water vapor, for the most part, won't escape to space, but mass/density of the system is not given. $\endgroup$ Jan 30, 2023 at 19:11
  • 1
    $\begingroup$ Where does '… venting the steam into space' come from, please? Why can it not be condensed and dribbled back? $\endgroup$ Jan 30, 2023 at 23:48

5 Answers 5


This is straightforward. You actually want to put oceans around the outside, covered in ice so they don't evaporate. Europa, for instance has oceans a couple hundred miles deep.

We already know that water will flow to follow the tidal flux, so put generators that sap power from the tides as they rush around the planet.

Addendum: Venus has 900 mph winds that are driven by the solar wind scraping against its leading side as it orbits. The atmosphere is so thick that the wind drags the planet against the force of tidal locking to have a day that is 19 Earth days longer than its year. I'm pretty sure you could power a computer off of that.

Addendum 2: After reading Rastlin's answer, it occurs to me that the heat of computation would create convection. For something this size, you'd want subduction cracks that let the cold water rush down into the core, and have vents that allow the water to flow past the processors, drawing excess heat from them. At the processor/water interface, you could recoup some of that energy with turbines before the coolant is released back into the ocean. It wouldn't work as perpetual motion -- you'd still need tidal generators -- but why throw that energy away?

Addendum 3: If you're concerned about the tidal forces running out, you can always have other moons adding to the fun. Io has so many of it's sisters squishing and stretching it that it's perpetually molten. You probably don't want to go that far, but you probably could put electromechanical generators that convert the distortions into power.

  • 3
    $\begingroup$ Just to account for the fundamental laws of physics still working out. The energy you take is from the gravitation of the gas giant, so eventually your moon will crash into the gas giant. This could easily take dozens or hundreds of million years so probably not an issue. The energy you do take is eventually turned into heat so if you don't want to cook the moon one should check how much heat the moon can radiate away into space. That should give you the limit on your total energy use. $\endgroup$
    – quarague
    Jan 30, 2023 at 6:52
  • 2
    $\begingroup$ @quarague actually when you take energy out a planet-moon system the moon will move further away rather than closer in - Earth's moon used to be a lot closer than it is now. So the moon won't crash into the planet but get further away until the tidal heating is no longer significant. Your point about thermal radiation is spot on though. $\endgroup$
    – N. Virgo
    Jan 30, 2023 at 9:29
  • 1
    $\begingroup$ @quarague, I think you're thinking of slingshotting probes. Tidal energy is pulled from the rotation of the moon, dragging it closer to being tide-locked. $\endgroup$ Jan 30, 2023 at 16:55
  • 2
    $\begingroup$ @quarague I believe the energy is being taken from the rotation of the moon, not it's orbital velocity. Eventually the moon will stop turning, and thus stop flexing and you will have a tidal locked, but still orbiting moon. $\endgroup$
    – Nosajimiki
    Jan 30, 2023 at 17:15
  • $\begingroup$ @Nosajimiki The rotation of the planet too; the transfer doesn't truly stop until the two bodies are locked to each other. But of course, this only makes sense for a system with two bodies - add a few more moons and it's no longer quite as simple - Io being the most obvious example where most of the tidal friction actually comes from the interaction of Jupiter and the other Galilean moons. In fact, Io is in orbital resonance with Europa and Ganymede, which in turn maintains Io's eccentricity, which is where most of the heat comes from. Io already is essentially tidally locked to Jupiter. $\endgroup$
    – Luaan
    Jan 31, 2023 at 8:47

They may want to cut down on the gargle-blasters.

  1. They appear to be heating the computer's coolant to the boiling point to produce steam to power the computer. The computer probably won't like this.
  2. As pointed out by other answers, tides already move the water. You could convert this motion directly to power, which is actually easier if the water is on top and the computer in the middle. This also allows your power generation equipment to dissipate waste heat directly to space, and natural convection constantly delivers cool water to the computer. You could also restrict the motion of the water to reduce the amount of energy dissipated when the computer isn't doing anything, which brings us to:
  3. Tides aren't the energy source here, they're just the means of coupling it to your generators. The energy source is the computer's orbital (and rotational) energy. In placing the computer into orbit around the gas giant, you were just storing energy that would later be released slowly via tides as the computer's orbit decayed (circularizing if there's no third-body interactions keeping it in an eccentric orbit, as is the case for most moons experiencing much tidal heating in our own system). Much of that energy is in fact being released as heat in the gas giant.

If the computer is being built in an existing moon, harnessing tides for power might make sense: that energy's already there to begin with, tides are just slowly releasing it. If they are constructing the computer as an artificial moon large enough to experience significant tidal heating, they obviously already have other, better ways to power things.

  • $\begingroup$ I believe the energy is being taken from the rotation of the moon, not it's orbital velocity. Eventually the moon will stop turning, and thus stop flexing and you will have a tidal locked, but still orbiting moon. $\endgroup$
    – Nosajimiki
    Jan 30, 2023 at 17:19
  • $\begingroup$ @Nosajimiki That's possible, though the biggest examples of tidal heating in our own solar system involve moons that tidally locked long ago, but are in eccentric orbits. The point remains, it's only releasing energy they put there initially. And there is still likely tidal drag causing orbital decay, since the moon will be in a low orbit to cause that kind of tidal heating. $\endgroup$ Jan 30, 2023 at 18:31
  • $\begingroup$ Hmm... that makes since, but it does make me wonder... does the planet lose orbital distance proportionally or does the orbit become more or less eccentric over time. $\endgroup$
    – Nosajimiki
    Jan 30, 2023 at 19:23
  • $\begingroup$ @Nosajimiki tidal drag is strongest when the planet is closest, so it tends to circularize the orbit unless a resonance with a third body is maintaining it. In which case it drags both bodies in. An example from our own system: en.wikipedia.org/wiki/Io_(moon)#Orbit_and_rotation $\endgroup$ Jan 30, 2023 at 19:53
  • $\begingroup$ Ah, so barring a multi-body system, the orbit would not necessarily fall into the planet, but lose eccentricity until finally settling into a circular, orbit (at which point tidal heating would cease)? $\endgroup$
    – Nosajimiki
    Jan 30, 2023 at 20:11

I don't see the need for heating. If you have tides, you can set up tidal generators. Of course that only works if your machine is large enough, which is probably larger than moon size. But it would probably have to be at least that big to experience tidal

An enormous spherical computer is built, consisting of billions of densely-packed processors each the size of a brain cell. This processing layer surrounds a huge reservoir of water. The computer is placed in orbit around a large body such as a gas giant, and due to tidal flexing, the fluid heats up.

So I'd do it the other way around, find a moon, put the computer on the floor or the core, and dump enough water to cover the surface. Then set up tidal power generators to power the core.

Also, a human brain has 100 billion cells in it, so depending on how many billions of processors each the size of a neuron you have, your spherical computer would not be large enough for tidal flexing.

Now, here is the difficult part. I had originally planned for the hot water to be channeled up from the core, turned into steam and used to drive fans and so power the processing layer, until I realised that this would then involve venting the steam into space, and thus slowly depleting the core of fluid. So, is there another way to produce a giant supercomputer powered by tidal flexing?

You wouldn't have to vent the steam, since it would naturally cool through black-body radiation once it's on the top layer. At that point it could filter back towards the center, though you'd want to separate the inputs from the outputs somehow, possibly by having them jut above the surface, then having condenser coils spiraling downwards, which will cause steam to turn into water while traveling back to the surface.


Your computer will be powered by electrical current generated by the motion of the moon through the magnetic field of the gas giant.

If a conducting object is in motion through an external magnetic field, an electrical current is produced in that object. The current produced gives rise to a magnetic field that opposes the external magnetic field. This is electromagnetic induction.

This is how Jupiter's moon Europa is getting its magnetic field - it is being induced. It is not a static field generated internally like that of the Earth. Electrical currents are being induced inside Europa as it traverses Jupiter's magnetic field, and they betray their presence by changing to generate a magnetic field which opposes that of Jupiter.

europa induced magnetic field


Like Europa your computer-containing moon has salty oceans and within these, electrical currents are produced. Your computers energy harvest apparatus offers a more conductive path for these currents to equilibrate and so they take a shortcut through its wires, doing some work along the way. The energy is generated from the gas giant producing its magnetic field and the motion of your moon through that field. This method of generating electricity will slow the moon very gradually and so might cause its orbit to decay over many millions of years.

It is not exactly tidal flexing but similar in that the energy is derived from the relative motion of the two bodies.

Prior art:

Is it possible for an organism to convert magnetic radiation into a sustainable energy source?


Hmm... Actually the right answer depends a lot on the material base on which your processors are built.

Plus, in the aforementioned case of a bunch of cores, there's the question of exchanging data between them. Which, in fact, is more complicated than just making a core. This also raises the question about the wave - how exactly will the wave pass through the computronium so that it transmits energy to the whole depth? I assume you've got a bunch of channels in there that are parallel to the core and the water flowing through them can do useful work. Otherwise, if the water is warming at the top, you're going to run into architectural problems of transmitting energy from top to bottom, to the depth.

Also, a computer this big has to have storage, possibly RAM, and interfaces for I/O.

Let's go point by point:

  1. it makes more sense to use optics for connectivity rather than current. This will greatly reduce heat dissipation, and will also be less expensive in terms of resources, wires have to be made of something.

  2. If you do coupling to optics, then converting light to current is not a difficult task. But computing on the core will be a bottle neck, since purely optical computers are 3-5 orders of magnitude faster. So it is reasonable to abandon silicon as well. By the way, this could be a good idea because it greatly reduces the requirements on the manufacturing process - no need to chase after small nanometers with their absurdly low yield of good product, you can do for optics in the hundreds of nanometers processes. And they do not require many rare elements.

  3. At this size, you can think about the specialization of individual cores. Some of them stream data packets, some of them multiply matrices, some of them look for the biggest value in an array of numbers. In my opinion for those, who are ready to build a computer about the size of the planet, the market logic of the architecture of processors and their specialization rules are not valid. It is utilitarian utility that dictates the profitability of a hybrid architecture. Two examples are Heinlein's Moon is Harsh Mistress and Groq company, who produces hardware accelerators for ML

  4. Interconnecting everything-to-everything is going to be a very complicated affair. Either we make a separate network layer and dedicated cores for it (which creates practically an internet within cores), or we have to accept that cores communicate only with neighboring cores in some range (100-1000-1000000 cores). This is reminiscent of neurons in the brain. Then, either the cores transmit data in a chain (which is time consuming) or we switch to a in-memory computation approach. This is reminiscent of the game of life - each nucleus counts something and passes it on to the next step

  5. Reversible computing, because of the magic scientific stuff, allows no heat to be generated during computation, as there is no loss of information. So there is practically no heat problem then. And this approach can still be done today, using ordinary silicon

  6. I like your original idea with water evaporation. That's basically how the planet works - water evaporates, goes to the sides and falls down at the poles or in the mountains. If you run thermally insulated capillaries from the poles, the melted water will go to the right cores even to the equator, and from there it will go up, evaporating in the process. By the way, maybe you don't need water, but organic liquids, at least methane. Such an atmosphere is even more likely. Look in the direction of Rankin's organic cycle. And google MEMS heat exchangers - turbines can be made very small. If the fluid on your planet consists of a ferrofluid or liquid metal, the thermal exchange would be more efficient, and fluid transfer could be done by means of a magnetic field. The fluid flowing from the bottom through the microchannels in the core would simultaneously draw heat and, passing through the magnetic coils, generate current in the processors. Look at MHD

  7. The cooling method you choose dictates the temperature range. But in fact - the technology you choose to make the processor is more important because it dictates the temperature range, and the cooling only has to adjust to that. Example - in p6 I talked about liquid metal cooling. If you do a bunch of radio tubes instead of cores (and darpa has done such projects at the current level of technology), if everything is heated to 400-600 degrees Celsius, the radio tubes can work without heat. And the performance of radio tubes, for reference, is 2-3 orders of magnitude higher than processors. And it is resistant to the magnetic field. That is, the free heating of the planet makes it possible not to melt, but to work quite well.

On the other hand, you can go down - at cryogenic temperatures you'll be running everything with superconductors. The heat output will go down, it will be more important for you to protect yourself from external heat from the gas giant and the sun. You can make a bunch of re-reflectors of ice on the surface to shine like a mirror.

You could go the way of a liquid ocean in which there would be individual cores in the form of organic cells. They do calculations and exchange their photoluminescence with each other, transmitting data. Bacteria computers are also made - you can google it.

And the bottom line is you don't need cores if you're making ASICs. You can build huge hybrids of waveguides and photonic crystals that will do the computing. You feed an electromagnetic wave to the input, and you take the results at the output. It would be fun to link this to the planet's magnetic field to feed the signal amplifier repeaters inside the device.


You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .