It’s now been almost six years since Marc Andreessen coined his famous phrase, 'Software is eating the world.' What seemed prescient in 2011 is now present in our everyday lives, as software continues to turn one industry after another on its ear.
Familiar examples like Uber or Airbnb, of course, come readily to mind. But even traditional industrial companies are turning themselves into software powerhouses. Fujitsu, through its Akisai cloud, provides a SaaS-based solution for agricultural production management. GE’s digital products analyze machine data in real time to increase productivity and create new business opportunities. These days, it seems, if you’re not already a software company, or on the road to becoming one, you’re going nowhere.
Even in the world of IoT, software stands atop the technology stack, with more attention (and investment) going toward applications than the other two primary domains in the IoT ecosystem: sensors and connectivity.
Yet, that’s starting to change. The IoT industry is giving much more attention these days to the underlying hardware - the compute, connectivity and sensor technologies that together enable IoT applications. Indeed, when it comes to IoT, I believe we are at the dawn of a 'hardware renaissance.'
This trend is accelerating as IoT permeates all parts of our lives, perhaps most notably in such areas as smart home technologies and intelligent cars. Many software companies have figured out that if you want to acquire customers on a software platform, you may need to build your own hardware. This full-stack approach is evident in such hardware devices as Amazon’s Alexa and Snap’s Spectacles.
The new emphasis on IoT devices over software reflects a much larger shift in the tech world, driven by things such as systems on a chip, smartphone adoption, ubiquitous connectivity and 3D printing that make it cheaper, faster and easier to create prototypes. Easy access to low-cost computers (think Raspberry Pi at $10) and sensors only accelerate the shift. Moreover, funding is becoming more plentiful from venture capitalists, like Bolt, Lemnos Labs, SpringBoard and New Labs, that are dedicated to hardware startups. Crowdfunding is another popular source of seed money.
The hardware renaissance, like most technological shifts, is happening in stages:
Stage one: The quickest and easiest way to make a 'dumb' device 'smart' is to connect it through gateways or platforms to the internet. An example is a pump manufacturer, such as Flowserve, increasing the 'operational awareness' of its equipment in the field, by making such equipment enabled for wireless communication and monitoring.
Stage two: Making hardware 'soft' and using it as a 'platform' to generate services revenue is the next logical step. When you think of a wearable device such as a Fitbit or Oculus or a smart home device such as an Amazon Echo or a Nest, you’re not buying it for what it is, you’re buying it for what it can be. These devices are supporting rich applications the way smartphones do, unlocking a wide range of new IoT services.
Stage three: The most advanced uses of IoT involve a combination of powerful hardware and software to add real intelligence to products at the edge. Good examples are collaborative robots, or cobots, that physically interact with humans in a shared workspace. Think Marble courier robots and do-it-yourself drones, such as Flybrix. Historically used for industrial and toy applications, these intelligent devices are increasingly going mainstream as more uses are developed for them.
To be clear, not all hardware products need to go through these stages, or even be connected to the internet; each of these stages will make their specific contributions to the hardware renaissance.
Any hardware story is incomplete without talking about advances in the semiconductor market fueled by the need for onboard intelligence. To enable edge devices to make better decisions, onboard deep learning processors are required, and this, in turn, is leading to the development of tiny AI processors that deliver processing capacity equivalent to multiple mobile GPUs while, at the same time, using lesser power.
Traditional hardware vendors such as Freescale, Intel, TI, Broadcom, Qualcomm, STM and Samsung are actively promoting their own IoT hardware ecosystems. And Google, which runs the largest computer network on earth, developed its own custom silicon, the Tensor Processing Unit, specifically to run its deep neural networks. This highly efficient chip, Google believes, is the key to making AI more affordable, especially for enterprises that depend on machine learning for their success.
With the lightning pace of change, it is difficult to peer around corners to see what’s next. Yet, amid this uncertainty, we would be wise to keep in mind the words of Alan Kay, who said, 'People who are really serious about software should make their own hardware.'
Success in IoT is not just about replicating existing business models through a connectivity prism; it depends on developing robust value chains and ecosystems. Just as the cloud brought about SaaS transformation in the software products industry, hardware companies now have the potential to support rich IoT applications and build recurring revenue streams independent of cyclical product sales.
While apps have been the sexier story in recent years, hardware is still exciting because, unless we live in the Matrix, we still need physical technology to interface with the world. Hardware startups are now looking like the software startups of the previous digital age - with the power to likewise eat the world.