The Internet of Things will need a whole new breed of chip — here’s what they’ll look like

First connected computers. Then connected devices. Now connected ‘Things’. The technology industry is gearing up for a major transformation, one that may end up having an even greater effect on our lives than the PC and smartphone revolutions.

The end uses are exciting — both those we’re anticipating and those we haven’t thought of yet. But the era of Things will also drive a revolution in the backend of technology. For hardware manufacturers, it will bring a radically new set of expectations and challenges.

We are starting to understand the devices that will underpin this vast movement. At the most basic level, intelligent things will need to contain a certain kind of chip. On the surface, there’s nothing particularly special about these chips; they will contain a number of components (transmitter, sensor, memory, etc.).

All of these components exist in some form in electronic devices today. The problem is that none of the components are good enough to build next-generation Internet of Things (IoT) chips. So we will have to completely re-engineer these basic building blocks of computing and then fit them together to make a chip that can perform under a radically new set of expectations.

What are those expectations? Although the specifics will vary widely by application, manufacturers across the IoT industry will be designing components that can fulfill four basic requirements: power, value, security, and integration.

Power

Today, we don’t think too much about the power our technology consumes. It’s annoying that most smartphones run through their charge in a day, but consumers are generally demanding bigger screens over longer battery life. In the large technology company category, powering data centers is the issue — with many companies building their own power plants to supply the massive amounts of energy needed. This is a long term problem for the industry, but it does not demand immediate innovation.

Not so with IoT ecosystems. Imagine hundreds upon hundreds of sensors spread across acres of farmland. Or a device that spends months inside a human body tracking the effectiveness of disease treatment. We can’t plug these things in, and we certainly can’t be changing the batteries every few days.

IoT chips will have to operate at low, low, low power. Some will be able to run for months without a recharge. Some will harvest energy from body motion, from the sun, or from temperature changes in the environment.

Value

Engineering quality components at reasonable prices is nothing new. But the model for delivering value in an IoT world will be quite different. Many smartphone components, for example, are approaching commodity status. The goal of the industry for years has been to engineer them smaller, faster, and cheaper.

IoT is different, partly because so many applications will deliver cost-savings by optimizing performance and identifying problems before they arise. The long-term value of IoT systems may be even more significant and harder to quantify. Businesses will use huge new troves of data to inform strategic decisions and long-term investments. Doctors will have new insights into patient health. And consumers will use new products and services to improve their lives in ways we probably haven’t even thought of yet.

Adding to this uncertainty are the upfront costs of implementing numerous IoT systems. Some will comprise hundreds if not thousands of chips, which creates a more complicated picture. As before, manufacturers will constantly be challenged to achieve high functionality at reasonable prices, but in a new and unfamiliar landscape.

Security

Devices everywhere, with little capacity to run security software, private data transmitted constantly over Wi-Fi, highly sensitive systems connected to the internet — what could go wrong?

As Emil Sturniolo of InStep Group points out, the Internet was built without much thought to security — once problems began arising, we had to go back and try to fix the issues that were being exploited.

That is simply not an option with IoT. There will be too many vulnerabilities, too much sensitive data, and too many accomplished cyberattackers looking for an opening; we must design systems that are fundamentally centered on security. One promising solution involves encrypting data at the microcontroller level. As with today’s cyber-security breaches, we may not recognize every vulnerability until after the fact. But on the whole, this is too important to get wrong and will be a major discussion point in the industry for years to come.

Integration

Hardware devices today usually include a grab bag of sourced parts. Even the iPhone, with its meticulous end-to-end supply chain management, includes many components that are developed independently or lightly modified.

With the Internet of Things, we no longer have the luxury of simply buying various components and making them work together. These devices will have extraordinary requirements placed on them and, for the most part, must fit a specific environment. A connected home, office, or factory cannot have dozens of bulky, ugly sensors lying around. Many devices will have to fit in tiny spaces, like the insides of water pipes, engines, or even human organs.

For an IoT system to deliver value, the device components need to work together as a single entity; they must be perfectly integrated for the device to meet its requirements. And in most cases, the device itself will need to be smoothly integrated into its environment.

All this points to a more unified manufacturing process, which will mark an important industry shift for hardware manufacturers. Many of today’s computing components, such as memory, are so easily reproducible that they have reached commodity status. By contrast, the IoT industry is likely to see a sharp rise in “application-specific” hardware manufacturing, where the entire design and development process is customized for one particular use-case. This may lead to a more diffuse hardware ecosystem, with specialized manufacturers designing end-to-end solutions for various IoT applications.

It’s an exciting time for the technology industry as we set out to reengineer the basic building blocks of electronics for a new era of computing. The companies who do so successfully will be big winners in an emerging multi-trillion dollar industry.

Narbeh Derhacobian is a co-founder, president and CEO of Adesto Technologies. He has over 17 years of industry experience working on discrete and embedded memory technologies. He has held technical and managerial roles at SST, AMD, Virage Logic and Cswitch Corporations. Narbeh’s industrial experiences cover development and commercialization of split-gate NOR flash, ETOX NOR, and NAND flash as well as SONOS based MirrorBit NOR flash technologies. He has over 40 granted patents in various areas of semiconductor memories.