Creeping Towards Hardware As A Service

There seems to be cool new hardware everywhere I look these days. Swimming drones. Crash-proof flying drones. Creepy robots. Kickstarters like Hackaball, “a smart and responsive ball children can program to invent and play their own games.” Startups like Orion Labs, with their Star-trek-esque Onyx communicator. Last month I went for a run along San Francisco’s Ocean Beach and was promptly buzzed by a drone. This did not seem particularly unusual.

Then I came across the fascinating Hardware Trends 2015 deck from Haxlr8r:

[slideshare id=45178242&doc=hardwaretrends2015-150226090440-conversion-gate02]

A few lines in it which particularly struck me, though you should really read the whole thing if you’re at all interested in hardware and/or startups:

components are getting commoditized and products are global from day one … prototyping is easier and cheaper … low-cost automation, 3D printers and robots are expanding to new industries and entering workshops, labs, and homes.

Indeed. I’m especially interested in hardware that makes other hardware development easier. Othermill. Hardware Battlefield winner Voltera. (Founded by a trio from my alma mater in Canada, he observed with some pleasure.) Raspberry Pi HATS.

It’s a truism that the current software boom owes a great deal to software becoming much easier to develop. Write code on an off-the-shelf laptop, push to the cloud, distribute to a billion smartphones. No need to install and configure your own servers, or mess with device drivers; no need for a million dollars in capital just to get a reasonable web service up and running. It seems to me that, tiny bit by tiny bit, one miniscule jigsaw piece at at time, we are–slowly–approaching a similar inflection point in the world of hardware — one which will inevitably be followed by a similar Cambrian explosion in hardware development, hardware startups, custom hardware art projects, etc., just as we’ve seen in software.

I don’t mean to go all blithe Pollyanna on you here. Hardware is hard, and our nonlinear or messily-nonlinear physical world tends to be a giant damper on any aspirations for exponential progress a la Moore’s Law. But at the same time, it’s difficult to deny that the tools available for hardware hackers are getting better in somewhat the same way that they did for software developers a decade or so ago, albeit in a slower, more disconnected, more fitful way.

We have years and years to go before anyone anywhere can write up a prototype specification in a hardware description language, one which is to Verilog as Python is to C; upload it to some centralized manufacturing mill; and have it etched, printed, assembled, and delivered by the close of the next business day. (Which is pretty much what we all take for granted with software nowadays.) But I can at least envision that happy day, and see that we’re moving towards it, one slow shuffling step at a time.