Pushing the Limits of Price on Small Cheap Computers

A system suitable for embedded, educational and R&D applications has been developed based on ARM and minimal hardware (no PSU) for $15, about the price of a box of copy-paper. The idea is to have a complete stack from circuit-board layout, CPU and OS completely open and produced by cooperation with Free Software proponents and Chinese hardware design and production.

Clearly, this is at or near the theoretical minimum cost so we finally see where Moore’s Law and Free Software can take us. Already some smartphones and tablets are available around ~$100, so we are “there”, at the end of the rainbow, the pot of gold, what the material costs and a bit of labour and shipping.

This means an end to many limitations of IT:

  • the Digital Divide between rich and poor is getting narrow enough anyone who wants to can leap over,
  • within a few years no region of Earth or activity of humans need be devoid of IT because of costs,
  • Wintel need not apply, and
  • the number of intelligent devices, including computers, smartphones, tablets, netbooks will explode.

With costs of material so low and with cost of entry into industry so low, we are entering an era of enhanced innovation with the only limitation left being imagination. I can see education where costs of Wintel have widely prevented environments with 1:1 student:computer ratio, government, business and individuals sooner or later using IT everywhere daily. The technology is just too inexpensive and too productive not to use it.

Bringing a product to market for end-users is just a few months away using this technology. China already has broadly used technology developed by government for the package, keyboard and display so new products using such low-cost inputs is inevitable and sure to have a market in wide emerging markets for IT and in niches in well-developed markets. Everything from toasters to desktop PCs are within reach. Even with Wintel costing $100+ we saw netbooks at $200 so the era of complete systems costing $100 or a bit less are upon us. No ultrabooks or market distortions by M$ and Intel are going to derail this juggernaut. In 2012, small cheap computers are going to kick ass everywhere but high-performance computing and servers.

There are still bottle-necks in prices such as plastic, displays, storage and memories but they seem to be relaxing, too. The old technology was so bloated, the new technology may be able to live off discarded material. The CPU and software are no problem at all. Here’s what was on the market last year. Here’s a video of it running:

About Robert Pogson

I am a retired teacher in Canada. I taught in the subject areas where I have worked for almost forty years: maths, physics, chemistry and computers. I love hunting, fishing, picking berries and mushrooms, too.
This entry was posted in technology. Bookmark the permalink.

14 Responses to Pushing the Limits of Price on Small Cheap Computers

  1. hi folks,

    well…. the original article was mis-quoted… on almost every single aspect that i talked to the reporter about. i did *NOT* say “the sale price is $15” and i have been fielding enquiries “please sell me qty 1 (one) of your product for $15 each” ever since, thanks to the mis-reporting.

    the irony is that in the three years since, people have *genuinely* managed to hit a $15 price-tag for a quad-core board: the new Allwinner H3 quad-core 1.6ghz SoC, which has on-board HDMI, on-board SATA and on-board 10/100 ethernet as well as being able to do 4k video decode just to name a few of its features, *really is* available as a $15 single-board computer, the “orange pi plus”. pretty amazing.

    however the rhombus tech project isn’t about chasing to the bottom of the barrel on price, it’s about saving people money long-term by allowing them to upgrade robust credit-card-sized CPU Cards and/or the products that take such cards. nowadays you get hermetically-sealed products: one problem, be it hardware or software, and the *entire device* has to be thrown away, wasting perfectly good components in the process.

    by contrast, upgrading an EOMA68 CPU Card, which is as simple as pressing a button on the side, popping out the old CPU Card and pushing in the new one within a few seconds, saves 60% or more on the cost of an equivalent non-upgradeable product, whether it be a laptop, tablet, desktop or anything else.

    *that’s* what this is about…. not getting the absolute lowest-cost product (which you just know will become obselete within 6 very short months).

  2. oiaohm says:

    oe exactly is more advanced form of actions talk and bulls**t walks.

    Dr Loser not all scientist obey science based trust.

    Basic science.

    You create a theory.
    You design a test after test to try to prove theory false.
    The tests must be repeatable by anyone.
    Not a test to try to prove a theory true.

    If the test works then the theory is false.

    If a theory lives through a long time of this it becomes accepted theory.

    science based trust works on the same basic idea.
    I think I should trust the person as theory.
    If product/designs/code don’t exist there trust is proven false until they do.
    Quality of that decides how much I should trust them.
    If they don’t have prior works there trust don’t exist until the product is to market.
    Have they done anything suspect? Like hiding identity. So voiding trust.
    Have other people also repeated the checks if the party is trust-able.

    The step have other people checked this out bit is what brings a lot of cons undone. Most people who have been conned never checked with anyone else if the deal was above board.

    If a person does not have there trust proven false for long enough they will get trust from the people using science based trust.

    It is possible to pull a con on science based trust. But getting away without you trust being blown forever more to people using science based trust is highly unlikely.

    Yes a con man most likely will only pull a con once against science based trust. Due to the way science based trust works as a collective. Most con stunts pure fail. Even a con man do pull it off they will have enough information to warn others who use science based trust so stopping the future cons by that con man dead. Remember the collective will also work to hunt down con people in there ranks.

    Science base trust will take revenge as well. Like if they were supporting one of your older projects. You do something they don’t like or miss trust they will pull resources. Novell found this out when they did a patent deal. Lot of third party support including free hardware just disappeared.

    You watch the science based trust play out in the FOSS world over and over again. Killing projects out right. OpenOffice and Libreoffice is a real recent example of this. Once trust is broken the FOSS world will turn and fast. Time from Orcale confirming break of trust to the starting up of Libreoffice 2 weeks. Under 1 month to raise the money to start a foundation officially in Germany.

    Science based trust is something to be very aware of when running anything open source.

    Developing FOSS code QA processes create the mind set of science based trust. Locating bugs in code requires taking processes of science based trust.

    Yes it something that FOSS has created.

    Of course DR Loser you have to go general to try to save you hide. To understand the FOSS world you have to understand how there trust system works.

    Faith that MS will release a better product people in the closed source world depend on.

    In FOSS not so. You see many projects forked. Yet all don’t. The explain for why some fork and some don’t is science based trust. If a person is seeing better quality coming out of the main project they form trust in it that it will improve. The main project screws up it QA and the project gets forked.

    The project that remains a live in the end is the project that does QA properly.

    The claim about FOSS having Faith that things will get better is not backed up by the way FOSS reacts.

    Items must be getting better or FOSS will attempt to turn on it. Not if. Like with the case of X11 attempting to turn on it does not work you can end up in a worse location for a while. Until someone takes the lead of that project who can make trust and draw a workable plan forwards.

    Faith and FOSS really does not exist in the same place. Complex trust system sometimes makes people think FOSS people have faith.

    OpenOffice had been improving in features and quality in a lot of places. This provides the required feed into science based trust that they will believe it will get better because it is getting better.

    Lot of fights between people like you Dr Loser and FOSS people are caused by the different point of view science based trust gives.

    Understand science based trust and actions of FOSS make perfect sense.

  3. oe says:

    science based trust – Interesting concept there, I like it, a much more eloquent way to state the concept than the more pedestrain

    “actions talk, bulls**t walks”

  4. Dr Loser says:


    “Science based trust people don’t get ripped off by con men.”

    I think you’re sadly deluded on this one, unless you propose a circular argument via “if you get ripped off by con men, then you are not science based.”

    Whatever “Science based trust” might mean.

    As with any other knowledge domain, you’ll find that scientists are just as gullible as anybody else when taken outside that domain. (In fact, I would argue that this is a vital prop for the uptake of Linux in scientific communities. Most scientists know F-all about computers: it is not their domain.)

    Sir Arthur Conan Doyle and fairies at the bottom of the garden? Isaac Newton and alchemy (not to mention the South Sea Bubble)? Tesla trusting Edison to play fair?

    There are myriads of counter examples to your thesis, Mr Oiaohm, and I have yet to see any published study that supports it.

  5. oiaohm says:

    Not particularly Munificent bunch is not exactly true. Its a issue of trust. FOSS it is hard to win trust. It really simple to lose trust.

    Groups that the FOSS world trust get very well treated. Sourceforge and kernel.org basically never needs to buy servers. They just mention they need some and they get some free delivery and all.

    Blender for the movie projects is trusted so gets a stack of free resources.

    If you can win the FOSS world trust they a very munificent bunch. Take raspberry pi the first units sold for thousands of dollars each and most have been given to museums and other places.

    When raspberry pi had produced the first units they won the trust of the FOSS world.

    One Laptop per Child. Again the FOSS world will pay up for existing product.

    Its all trust. Paranoia is not what FOSS exactly suffers from. FOSS requires trust to be proven-able and testable. FOSS is a method of science when it comes to how it trusts.

    Science based trust people don’t get ripped off by con men. Of course those who like don’t like science based trust like to call it paranoia. So they don’t have to accept the fact they are running on faith not proven trust.

    Faith is really not part of the FOSS world that much.

    Lack of faith in closed source is a common item in the FOSS world. Distrust of companies like Microsoft who have repeatedly breached trust is to be expected in a world running on science based trust.

  6. Dr Loser says:


    “FOSS people are not a trusting bunch Clarence Moon when it comes to money. If you don’t have a trust-able ID you don’t get paid.”

    I am irresistibly reminded of the ‘=>’ operator in first-order predicate logic here (“implies”).

    Not only are FOSS people not a trusting bunch (unsurprising, given the high incidence of paranoia amogst them), but they’re not a particularly munificent bunch, either.

    Have you any evidence that “FOSS people” (I like to think of them as similar to the Beaker People, except lacking in a suitable beaker that will hold water) will actually pay for anything at all?

  7. oiaohm says:

    http://git.rhombus-tech.net/eoma So far its giving more than most scams do. Clarence Moon.

    Also the Allwinner A10 is a common prototyping chip.

    Not like I could expect you do any homework to check if the require paper work existed. For this stage of development everything is in order.

    Of course I would not pre pay myself until the first units are produced. In case of bugs. Its so simple to stuff a board design up.

    Of course this is not the first item this guy has designed either.


    The developer and the other embedded developers helping are not unknown to the FOSS world. Clarence Moon. Quite a well known person and if he pulls a fly by night he will be quite catch-able.

    FOSS people are not a trusting bunch Clarence Moon when it comes to money. If you don’t have a trust-able ID you don’t get paid.

    Yes his pricing will be fairly close. He has experience at doing this stuff.

    This is not like the Nigerian stuff at all we have ID of who we are really dealing with.

    Scam you normally don’t have the people exact ID’s to hunt them down.

  8. Clarence Moon says:

    I am sure that this is a well-intentioned sort of love-in opportunity for the open world set, but it has all the earmarks of a scam. Commit to some purchase at a price to be set in the future by some anonymous source hiding somewhere? I might be equally profitable to invest in the futures offered by the widows of Nigerian colonels.

  9. oiaohm says:

    Ivan really less humans are being used to make stuff. Human make too many errors. Also robots work without sleep. Even if you pay a human -20 dollars a day it still cheaper to have a robot make the devices. Reason mistakes are expensive. So if you want to make bucks you don’t want humans making devices.

    If you look closely at http://rhombus-tech.net/ it recycling preexisting robot plants. So I don’t have to make a new factory. Alter a robot factory that has been making things like PCMCIA network cards now to make plugable processor boards. The old PCMCIA cards were never assembled by humans. There a job of a pick and place machine.

    1991 is the PCMCIA port no more patents on that bugger. Its also a well tested port design so dependable. So some of the FOSS designed stuff is going to be highly dependable.

    Think about it how many laptop plants have to exist that place a PCMCIA port on the side that is a ideal frame to have a new motherboard design placed inside supporting this. Smaller and lighter motherboard more room for battery.

    Same is true for the Raspberry pi. Mind you broadcom makes a profit from every Raspberry pi sold they make IP rights on the chip used. So all the development time written of by a non for profit they will recover.

    FOSS is recycling. This is the problem there are more and more plants we have evolved to obsoleted than with a few minor changes can come back on-line. Cost the plant is basically paid for. The plant runs dependably. There is proft to be made. The plants are more than able to spit out many millions of units per day. Since that is what they are designed todo.

    Recycling plants for plastic has come on line in china no human required to sort it feed devices in one end raw materials to make more devices come out the other end. China has closed the material loop.

    This means the volume of new plastic required will reduce over the next few years. Faster than oil supplies reduce.

    Bioplastics have not moved up in price and china only has quite a few billion tons of devices to recycle for plastic. Mostly caused by having humans in production lines.

  10. Ivan says:

    “And then of course there’s all that plastic and the other downward-trending commodities involved.”

    Plastic prices are going up. Thankfully most plastic production has moved to third world countries where labor costs are kept incredibly low by the unhealthy plant conditions, so that isn’t really an issue.

    It should be possible to make a cheap ‘set top box’ PVR/video streamer using linux. You could even sell them at a nice markup. Just remember to lock them down so the vocal minority can’t install any of their software, it’s quite possible and legal.

    It’s a win-win situation, really, you’d upset the high hippy and make an assload of money off the backs of others. Let’s face it, that is what open source is all about.

  11. oiaohm says:

    Dr Loser http://rhombus-tech.net/ and the tablet show the PSU for those devices is EU standard for phones. Basically a high power USB port that you might have on your TV/Computer/……..

    Good ones of those stand alone are about 10 dollars. Reason the complete board when running contained only pulled 1.5 watts. So using it solid state drive on board yes it can run. In charging mode the specification of usb allows for 1.5 amps at 5 volts or nice 6.5 watts of power. So you have 5 watts left over to drive stuff like a small screen.

    If you want to connect on a hard-drive. Laptop hard-drive. You may pull that off still in the same power-supply.

    Full size 3.5 hard drive a little bigger power supply unit required. 50 watt about 15 dollars for a good quality one. And that is mostly to power the harddrive. You technically could daisy chain the rhombus in. Even with one of the rhombus do everything boards ie development platform the 50 watt would do.

    15 dollars powersupply. 15 dollar cpu board with extras. 15 dollars motherboard and 15 dollars case. Its a max of 60 dollars. The most expensive parts is going to be the screen and harddrive. You will get decent keyboard and mouse pair for 15 dollars. Basically light weight computer tower for under 200 dollars with this hardware.

    http://elinux.org/Embedded_Open_Modular_Architecture/PCMCIA rhombus is first generation of this tech.

    Basically upgradable tablets.

    First of this is a 1.5 ghz processor with up to 1G of ram. That is decent to run light weight Linux desktop and thin terminal clients.

    Rhombus is looking to be second generation of this tech as well. http://beagleboard.org/bone

    Again these are very small computers. That are getting down right powerful. They look nothing like the PC motherboard designs of old.

    Beagle bone is more targeted at robotics.

    Most of what going on is reusing existing socket designs.

    Question is what tech will follow Rhombus duel and quad core arm chips? With more ram?

    scientific calculator is no good for robotics or a lot of the other roles.

  12. Dr Loser wrote, “Would it do more than a decent scientific calculator?”

    You betcha! ARM can run GNU/Linux. Most scientific calculators have a much narrower range of applications (mathematical and statistcal functions, perhaps graphing). You could make a pretty powerful calculator with the gadget though.

  13. oe says:

    “Ultabooks”, funny how these are bandied about on a the morning mainstream (read: PAID) media on the morning shows…..I don’t think consumers will fall for these boat-anchors, when it comes to portable always-on computing MS brand is a liability. Better to buy an iMac if you’ve got the cash to burn, otherwise its a GNU/Linux EEE, ACER, or Samsung ARMed tablet for 1/5th to 1/10th the price…..

  14. Dr Loser says:

    Any idea how much the PSU would cost, Robert?

    And then of course there’s all that plastic and the other downward-trending commodities involved.

    And what, precisely, is it good for? Would it do more than a decent scientific calculator?

    Because I can pick one of those up round the corner for far less than $15, you know.

Leave a Reply