Each compute platform is 10x its predecessor – what’s next?


Benedict Evans of A16Z used this chart to illustrate his point that each new compute platform brings a step change in scale when compared with its predecessor. Smartphones will soon be 10x PCs, PCs were a step in scale up from mini-computers and mini-computers were a step in scale up from mainframes. In the title to this post I used 10x to describe all the changes because I think that’s approximately right and implied by Benedict’s post, but I don’t have data.

The other thing to note is that the time between each of these platforms hitting scale is falling. Mainframes came to prominence in the 1950s, mini-computers hit their stride in the 1970s, PCs were in the 1990s and smartphones were in the 2000s.

If this pattern is to be extended the next compute platform will soon be upon us.

Which leads to the question: what will the next platform be?

The only thing I can conceive of which could evolve into a compute platform with 10x more nodes than smartphones is a meshed compute network which distributes computing into the fabric of life – our watches, our white goods, our bedroom alarm clocks, as well as our thermostats, alarm systems and everything else. To me that sounds about as fanciful as an iPhone 6s would have at the peak of the PC era 15 years ago.

This vision of the future points to early investment opportunities in point products – e.g. thermostats and watches – and then into closed ecosystems, and the into infrastructure that builds an open ecosystem before a new generation of apps emerges.

As I write these words it feels highly likely to me that this sequence of events will occur, with the primary short term benefit being that we are saved from carrying computers with us the whole time. Maybe the odd inter-generational tech question my grandkids will ask my kids (their parents) will be “what, you used to carry computers with you?”.