What Is This? A Computer For Ants!?

Smallest Computer

How can we be expected to teach children use a computer if they can’t even see it? I don’t wanna hear your excuses! The computer has to be at least… three times bigger than this!

Developed by the University of Michigan, the Michigan Micro Mote (M3) is quite possibly the world’s tiniest computer. It’s about the size of a grain of rice.

The multi-layered PCB (shown after the break) features 7 layers of components, surrounded in epoxy for protection. Drawing only 2 nano Amps during standby, the computer can be powered by a 1 millimeter squared solar cell. It’s designed to be glued to a window for use. It’s capable of input data via sensors, the ability to process and store the data, and then output the data wirelessly. Its range is only 2 meters at the moment, but they hope to extend it to about 20 meters.

Smallest Computer


For more information, check out the following video:

The internet of things is looking very promising. On the other hand, secret spying is even scarier.

[Thanks for the tip Ben!]

68 thoughts on “What Is This? A Computer For Ants!?

  1. Multi-PCB?
    For me it looks like a multi-die.

    How do you define “Computer”?
    Actually a CPU RAM and ROM are a “computer”. So I am afraid there are lot smaller computers in existence.
    Modern SoCs do a lot more and they are also extremely small.

          1. “A computer is a general-purpose device that can be programmed to carry out a set of arithmetic or logical operations automatically. Since a sequence of operations can be readily changed, the computer can solve more than one kind of problem.”

            Therefore-> Yes

          2. a SoC is a computer. a battery and/or power source is not a requirement for something to be a computer, just the ability to perform calculations. There is a huge difference in capabilities between a SoC and a microcontroller though. The chip in this article is technically still a SoC (System on a Chip).

          3. A practical computer need: power source and input/output. A SoC without a power source doesn’t compute anything. And a computer without any data input and output to report computations results or to act on something is of no use.

            This computer have all needed: power source and inputs/outputs.

            Sure it is a stack of silicon dies, so what.

          4. “A practical computer need: power source and input/output. A SoC without a power source doesn’t compute anything. And a computer without any data input and output to report computations results or to act on something is of no use. ”

            Not true. You wrote what makes a computer usable. That’s a moot point. Think about it: there are warehouses full of computers unplugged and turned off, packed in boxes and such. Some are in museums. That fact doesn’t make them non-computers, not does it make them “impractical”. A power source and user input/output doesn’t have anything to do with making something a computer or not, or making it practical or not. It just makes it human-useable or not. Sorry for reporting you, I meant to hit reply.

        1. The point was that a bare die SoC isn’t a computer by itself, because it can’t do anything without power and some external interface. If, by your definition, a computer is “a general-purpose device that can be programmed to carry out a set of arithmetic or logical operations automatically”, then a bare die SoC is *not* a computer, since it cannot be programmed, nor carry out any operations, without power or an external interface. It just sits there, doing nothing.

          Obviously this can carry out operations by itself, although there’s no detail as to whether or not it can be programmed/updated by itself (over radio). I’d have to believe that’s possible with this setup, though, so it, alone, meets your definition.

          1. A SoC has GPIO Pads and therefore an interface for programming and also an interface for the outer world.

            I understand that power supply problem but would you say your desktop PC is not a computer because it needs an external power plant to work?

          2. If my desktop PC didn’t have a power supply, I wouldn’t call it a computer. The power supply has a socket that you plug a cable into. That’s fine. A bare SoC die wouldn’t have a way to do that. Even if you wire-bonded wires to it, it almost certainly would need decoupling caps and a reasonable low-impedance return path in order to actually function, and doing that is challenging. Then you’d need some method of bringing signals and power to it, and all of those things would take space, limiting the space you can fit it in. None of which applies here.

          3. A SoC (as same as a µC) usually has voltage monitors, and also regulators. You only need external caps if you can not trust the power supply. Also you often have a quite big range from 3 V to 5 V . Of course the requirements are different but I can’t see a fundamental difference to an ordinary PC

          4. A SoC is a computer, plus some extra things. A power supply is not a requirement of a computer, it’s just a requirement to turn one on. Also, a SoC CAN be programmed. Programming it by ROM is still programming, there is no requirement that the program can be changed. By your logic all the “computers” sitting in warehouses unplugged are not really computers at all. This shows your logic is incorrect.

    1. IMHO, CPU + Memory (RAM/ROM) doesn’t make a computer. You need a way to put things on memory and a way retrieve the results of the computation. A computer is defined by CPU + Memory + I/O and SoCs doesn’t have I/O by themselves.

      1. Huh? Plenty of chips have onboard I/O. The ESP8266, for instance, has everything onboard to be a computer, without much of anything else (it’d have pretty crap range without an off-chip antenna, though, if it’s even possible). It would just need power, which is the complicated part to make tiny, which is what the article actually says. Note here how many layers are dedicated to power: the battery, stabilization, charging and control, and the PV cells.

    2. Before the term “computer” was taken and redefined in the 1940’s, “computer” was a job title given to any person who’s job function was processing numbers using math.

      Even still today, the most basic definition of “computer” is a thing that takes energy and converts it to a subset of “useful work” called computation (manipulating numbers, though “math” is a more common if not overly-simple name for it)

      Just like “motor” is any thing that converts energy into the subset of useful work called “force”, and does not require electricity to be involved despite that is a very common form of energy used, a computer doesn’t actually require electricity as its energy either.

      Just because the mote computers have very limited resources does not mean they are not computers.
      Many AVR and PIC chips have very limited resources compared to something like my gaming PC – but both at their core are computers – and the mote computers are no different in that aspect.

        1. I looked it up over all it’s much less powerful then lets say a C64 or even a ZX81 but it is still a Turing complete computer.
          You don’t need a lot of processing power for a wireless sensor instead energy efficiency is more important.

  2. Being pedantic here but is it the M3 or the M^3? I suspect the marketing department went for M^3 because it looks cooler as a logo but they probably still call it the M3 unaware that their logo now say’s M Cubed. Petty I know, but it’s a bug of mine, bit like when companies put the word ‘Quantum’ in front of their products to make them sound more impressive.

    1. This isn’t a mathematical formula. Their usage is a commonly accepted usage where the exponent is to indicate the repetitive initials of an alliterative name. In this case, “Michigan Micro Mote”.
      That and “3M” was already taken by the Minnesota Mining and Manufacturing Company.

  3. Bullshit ! Okay, yes it can compute things, but to compare this thing with a desktop pc is totally bullshit! it has no monitor, no input device, ooooh, but it has wireless connections.
    Yes, its nice that they can stack some chips and bond it to a microcontroller without external components. But to advertise with a compare with a desktop PC ??? WTF!?!?!

  4. It’s a pity that these things are put in the hands of PR functionaries and newspeople. The really interesting part of this isn’t that they’ve made a programmable device that’s today’s tiniest (nor is it the super smart lurking nano-bot swarm of Crichton’s bad scifi) but that they’re pushing the boundaries of power consumption.

    The raison d’être of this is building down to the smallest device that can be “self-powered” rather than actuated by RF pulse (RFID) or some other device. Will we have solar-powered nano-Arduinos to gripe about here? Probably not. Will we be injected with nano-bots to assay our circulatory system? Possibly (our poor kidneys). Will the battery fabrication technology and low power chip design filter into our kids’ iPhone 11? Very likely.

    1. “Nanobots” are really, really far away, if they’re even possible (which I doubt). Even if we relax the idea that a ‘bot’ has to be able to move, this would be a ‘millibot’: it’s still millimeter-scale. That just really stresses how far you would have to go to get to ‘nano-scale’, and the power problems just continue. I think even getting to micro-scale would be really, really challenging.

      Also, it’d be much cooler if this *was* actuated by an RF pulse, rather than by ambient light, since an RF pulse is completely nonintrusive to the people in the room (rather than ‘leave the lights on so the computers have power’). Sadly, basic physics says that when you get smaller and smaller, photovoltaics will always win out.

      1. Hitachi and others already have RFID “dust” but, as always, the power-storage (and therefore sustained operation rather than just momentary response) is likely a major limiting factor.

        1. Visible light is a slice of the electromagnetic spectrum, it is not part of the longer wave radio spectrum. We have defined loose boundaries for how electromagnetic radiation interacts with matter at specific frequencies. If you could see FM radio broadcasts, well you’d be super human.

      2. Nano robots are likely going to be biologically created, possibly from modified virus species. In fact, they have specifically succeeded in creating a virus that targets cancer cells and attacks them like a normal virus would attack healthy cells. A virus capable of finding and attacking specific cells, entering and rather than replicate, it could die, leaving modified DNA strands would mean you could modify many aspects of your body very quickly.

      3. I have to disagree with your claim on being possible.
        I offer as evidence all the ribosomes in your bodies cells, since ribosomes both exist and are a pretty power efficient machine that already exists and meets the full definition of a nano machine.

        I unfortunately do however agree that we are still very far off from the technological level to design and create a similar machine from scratch.
        (Unfortunate only in that I would love to see such a creation within my life time, but alas it seems very unlikely that will be the case)

  5. Embed these in a floor of a building, making sure they can get light. Mesh network them. equip them with siesmic or sound sensors. The mesh would be able to track movement across the floor. Use as an alarm sensor or to record movement patterns over time.

    1. You are describing the localisers in Vernor Vinge’s Deepness in the Sky. Small ubiquitous dust-mote computers for sensing. Meshed together they have enormous parallel processing ability and they’re the ultimate tool of a totalitarian government.

      1. I was wondering when someone would mention Vinge. Vinge also wrote about having AT&T drop a bunch of these things with IR laser comms by helicopter to make an instant high bandwidth communication network. Oh and because each one of these things has a laser you can do time of flight to figure out where it is in relation to it’s neighbors and with all this you can figure out the position and orientation of stuff easily.

        This way you don’t need crazy computer vision stuff to solve problems like delivery drones, augmented reality, or getting robots to pick up your dirty clothes. Who needs SLAM when you can just ‘ping’ a whole bunch of nodes in the ambient environment.

        And this was in 2000!

        He also wrote about them in Rainbow’s End too. He didn’t really mention the whole government surveillance angle much though.

  6. A computer computes. As in does boolean math. Which is what all computers do. What all of you are hung up on is the complexity of a computer. It doesn’t have to have physical inputs or outputs to compute. It’s performing computations for logic, and this defines it. Doing useful work, on the other hand, is where people have trouble defining a computer. You may not call this a non-computer simply because it lacks common appearance to a laptop or desktop. Even an atmel ATTINY85 is a computer. It does little in terms to a laptop, yet it can do things a laptop can not easily. By easy I mean what a microcontroller can interface to without complicated work to develop the drivers necessary to command vast networks of logic circuits. It has the logic and basic necessities to work with low level circuits. A laptop can not do this in any simple terms. It is basically too complex, layers of control that are too extensive to turn relays on and off, or control power circuits. Yet both perform boolean computations. At their own designed speed. This is what makes them computers.

    1. I kind of agree with you, except for the boolean math part. There have certainly been analogue computers which have no requirement for boolean logic, some examples of which I think have been posted on HaD.

  7. A computer has to run doom possibly doom 2 or it is NOT a computer….thats right VIC20 you are now a big complicated calculator with detachable screen. Commodore 64? Calculator with disk drive.

  8. You all need to read this old essay by Charlie Stross. Consider what happens when someone manufactures these in bulk and air-drops them over a city at one-per-square-metre density then hooks them all up into an ambient-light-powered wireless mesh. It’s world-changing, and probably not in a good way.

  9. I am impressed that no one has mentioned the one thing this needs to become a MASSIVE spy tool: put a vibration sensor (sensitive in the range of ~30 to 18k Hz) along with the camera, and you have a superb surveillance device. You glue this to an external window. The window pane itself becomes your microphone, sunlight through the window supplies power for the device. Two meter transmission range does suck, but they expect to extend it to 20m, so wait a year or two, and the NSA will be producing these things in bulk. Actually, I would be surprised if No Such Agency doesn’t already have something similar.

Leave a Reply

Please be kind and respectful to help make the comments section excellent. (Comment Policy)

This site uses Akismet to reduce spam. Learn how your comment data is processed.