Hardware not Software is Eating the World: Len was right.

In 1996, Len Widra, a Principal Engineer at SGI, and I got into a heated argument over the importance of software vs. hardware.

I was a kid out of school with an ego to match. Of course, I thought I knew everything. The crux of our debate was whether the software was a relevant technology or important technology.

Len’s observation was that software was irrelevant or something like that. Hardware, he observed, was the important technology.

As a software engineer, this was infuriating. As a computer scientist, this wasn’t very kind. How dare he say that a bunch of silicon was more important than my code?

It’s been almost 18 years, and I’ve learned more.

What I have learned is that new software rarely, if ever, displaces old software unless some new hardware shows up. New hardware shows up, and that new hardware makes the old software irrelevant or obsolete.

There is one interesting caveat. Some software applications are really dependent on the quality of the algorithms, and as the algorithms improve, the software gets obsoleted regardless of the underlying hardware changes. In many cases, the emergence of new algorithms creates new hardware that helps obsolete the old software.

For the vast majority of software systems, however, that’s not the case.

When you are looking for a new opportunity in the technology space, what you need to look for is where new hardware is emerging. If it is sufficiently different, that new hardware will obsolete the old software that was tied to the new hardware creating new opportunities for new software.

A mouthful.

A few examples:

(1) the emergence of x86 servers created the opening for Linux. Before x86 servers were a reality, the UNIX vendors owned the entire software and hardware stack. When x86 became good enough, a new software stack could win because the software used new hardware.

(2) Flash in the storage industry has truly created a massive disruption, enabling many different kinds of software stacks.

(3) Merchant (aka Broadcomm) Silicon is disrupting the networking space that reminds me of the x86 disruption.

(4) ARM processors made mobile computing plausible.

Maybe my favorite example is this picture from TIOBE Software that measures the popularity of programming languages. TIOBE measures popularity – not use or lines of code – and has been doing that analysis for many years:

You look at the chart, and you realize how slowly programming language popularity changes except for one programming language: Objective-C. The popularity of a single programming language changed dramatically not because it was good or bad but because of a single new hardware platform that enabled new software.

The hardware disrupts because it enables software that was impossible before. The carefully calibrated trade-offs that are baked into a system are tossed into the sea with new hardware. When you want to look for disruptions to your business, never look at software; software is irrelevant; look at the hardware …