Project Ara

~ the obituary ~

project ara was the modular phone everyone in the engineering forums wanted to believe in. you would buy a frame called an endo, then snap in modules as you needed them. a better camera when you wanted one. a fresh battery when the old one died. you would never throw away the whole phone again. the e-waste argument was strong. the right-to-repair argument was strong before that phrase had a real movement behind it.

motorola mobility, then still inside google, unveiled the project on 29 october 2013, building on dave hakkens' phonebloks concept video from september of that year. paul eremenko led the team inside google's advanced technology and projects group. regina dugan ran atap from the top. the early dev conference in april 2014 packed the computer history museum with engineers who wanted to write driver code for swappable camera modules. the second conference in january 2015 was even bigger.

the puerto rico market pilot was scheduled for late 2015. it was cancelled. google said the prototypes had failed drop tests and the magnets holding modules in place were not strong enough. the project was moved from atap into the main android division, and that was the first sign the timeline had slipped past the point of recovery. on 2 september 2016 reuters reported that google had suspended ara entirely. there was no farewell post.

i was watching this one from the engineer side. by 2014 i was already in computer engineering work, and ara was the kind of project that comes up in lunch conversation as a thought experiment about whether the bill of materials math could ever close. the answer most of us gave was no. modularity has structural overhead. the frame has to hold sockets that are stronger than the things plugged into them, the gaps for the sockets cost battery volume, the radios have to live somewhere that does not get unplugged, and the whole device ends up thicker and heavier than a non-modular one with the same specs.

~ the case file ~

~ what it was supposed to be ~

the endo was the frame. it was about the size of a normal phone, made of structural aluminum, with a grid of slots on the back held by electropermanent magnets. each slot could carry a module: a 5 megapixel rear camera, a higher-resolution rear camera, a swappable speaker, a battery, a fingerprint sensor. the modules talked to the endo over unipro, a serial protocol borrowed from the camera-interface world. google was open about wanting third-party module vendors to ship hardware the way android phone vendors shipped apks.

the bill-of-materials target for a basic starter kit was around $50. that price would have put a working smartphone into the hands of feature-phone users in markets where a $200 android was still aspirational. that target audience was the part of the pitch that was hardest to argue with, and it was the part of the pitch that the cancellation in 2016 silently retired.

~ the slide ~

the fcc problem was real. the fcc tested whole phones, not phones with arbitrary swappable radios. every new module configuration was a new fcc submission. google never had a public answer for how a modular ecosystem would survive a regulator that wanted to certify a single fixed antenna stack per device.

the drop test failures in late 2015 were also real. the puerto rico pilot was supposed to put working ara devices into the hands of consumers, with module vendors set up in carrier stores. when google admitted the magnets were not strong enough to keep modules in during normal drops, it lost the rest of the timeline. the team did rework the magnet design, and the 2016 atap demo at i/o showed a device with permanent core components and only six swappable side slots. that was already a retreat from the original endo idea, and it still did not ship.

~ what got buried with it ~

ara took a particular kind of optimism with it. you can still buy a framework laptop and you can still build a desktop pc from parts. the phone, the most personal computer most people own, went the other direction. by 2016 every flagship had glued batteries and soldered storage, with no socket bigger than a sim slot. the modular dream survived in fairphone's quieter line, which is still shipping in 2026, but as a niche product rather than the future of all phones.

the second piece that got buried was google atap as a serious hardware skunkworks. project tango was already in trouble in 2016. project soli became a feature in a pixel and then disappeared. atap kept the lights on but stopped pretending it was the lab that would deliver the next big consumer device. by 2018 the modular-phone fantasy had moved on entirely, to folding phones and to whatever samsung was promising next.

@}-->--

              )
             (_)
            .-'-.
            |   |
            |   |
            |   |
            |   |
          __|   |__   .-.
       .-'  |   |  `-:   :
      :     `---'     :-'
       `-._       _.-'
           '""""""

« previous  ·  random  ·  next »