Like clockwork, apple introduces an upgrade to its iconic iPhone each year. Sometimes it's thinner or less curved, but much like a new Porsche, you can look at it and think nothing's changed. And yet, like Porsche, a lot is different—if you peer under the hood.
When I got my hands on the iPhone 5S, I marveled at the material science behind that bit of glass and metal. The big surprise was the A7 processor—a 64-bit chip—powering that tiny device. And perhaps no one was more shocked than Qualcomm, whose main gig is to make chips for mobile devices.
The A7 neatly sums up the new reality of our always-connected world: Every market is ripe for upheaval, competition can come from anywhere, and today's customer could very well be the one who knocks you from your perch.
Think back to just 15 years ago when the likes of Cisco, Dell, HP, Intel, and Oracle got big supplying the backbone of Internet computing. Their incumbency meant they had to serve the needs of a lot of customers, which slowed them down. And as the metabolism of change in the Internet era has sped up, they couldn't keep pace.
When Facebook was in its megagrowth phase a few years ago, it realized that the big server companies couldn't make what it needed to serve up all those cute baby pictures and endless event requests. So Facebook set up a skunk-works project and designed its own servers specifically to make Facebook services zoom at the lowest possible cost. It sent the plans to an Asian server maker named Quanta to build these streamlined boxes cheaply.
Facebook didn't have designs on the $55 billion server market. It took matters in its own hands and ended up creating a competitor that Dell, HP, and IBM didn't see coming. Facebook, thanks to its sheer size and complexity, is the standard-bearer for the data-rich, highly networked future of information. When it designs machines to handle its workloads, it's creating the next-generation server. The big-hardware makers let the tail wag the dog.
As with Facebook, Apple's main business isn't making chips. But it had to craft its own processors once it realized that the major semiconductor manufacturers simply weren't going to push the envelope on performance fast enough to meet the company's development timeline. Setting up an engineering team and paying $278 million in 2008 to buy a microprocessor design firm is a bargain if it helps you produce quarterly profits in the billions.
Unsurprisingly, both Google and Amazon, the other two dominant companies of our day, also have stories of taking their technology destiny in their own hands. Amazon Web Services is a classic example, and Google invented something called MapReduce Framework, which allowed it to process many terabytes of data across a cluster of thousands of low-end computers.
That work eventually led to the creation of the open-source Apache Hadoop project at Yahoo, which Yahoo released free to the world for anyone who needs to munch massive amounts of data. (Sorry, Oracle.) Facebook, too, open-sourced its server designs so anyone else could copy and adapt them for their needs.
Why would Yahoo and Facebook take what looked like a competitive edge and set it free? The hacker ethos is part of their business thinking: Sharing is in, and, hey, if people tinker with your idea, that's free R&D. In the case of Facebook's web-server design, as more companies adopt its plans and enlist Quanta and other Asian manufacturers to build them, economies of scale kick in and Facebook's costs to buy new servers only come down further.
The strategy today is simple: In order to move fast, build what you can't buy or risk losing control of your fate and becoming the next Palm, Motorola, or HTC. And if, in the process, you disrupt an Oracle or a Qualcomm? So be it.
Om Malik is the founder of Gigaom, a Silicon Valley–based digital publisher, events producer, and research firm. He's been writing about technology for more than two decades.
A version of this article appeared in the March 2014 issue of Fast Company magazine.