Contextual Intelligence and the IoT Revolution: Sense, Connect, Learn and Respond
The pace of change continues to accelerate in both the semiconductor industry and the broader high-tech market space. We’ve crossed over from the Mobile Era—which put computing power and wireless connectivity at consumers’ fingertips—to the IoT Era, which is rapidly embedding these capabilities in a range of smart products, from thermostats, home appliances, and connected cars for consumers, to Industry 4.0 smart factories.
From where I sit in Silicon Valley—Ground Zero for the IoT proliferation, where there’s a higher probability every day that the car alongside you at a red light is a Google prototype with no driver behind the wheel—seismic smart technology shifts are taking place right before our eyes. As represented in the image below, the IoT represents an enormous opportunity for tech companies.
Think about what’s driving IoT expansion. Moore’s Law, which posited that the number of transistors on a chip doubles about every two years, along with computing power, has just about run its course. I’m not here to argue about whether it has or has not—those are my thoughts and I believe it has. But at the very least, we’ve reached the point where incremental improvements are failing to justify the enormous capital investments required in state-of-the-art manufacturing facilities.
Moore’s Law has been replaced by Metcalf’s Law, which states that the value of a telecommunications network is proportional to the square of the number of connected users of the system (n2). This new law perfectly captures the dynamic of exponential IoT growth. Viewed another way, pushing computing power out to the periphery of the network is a logical extension of our need for continued performance and functionality improvements.
There’s also a human component to IoT growth: Consumers want technology to make their lives easier and safer and to save them time and money. There are fewer and fewer of us who haven’t experienced the benefits of a connected home—smart meters, smart appliances, smart smoke detectors, for example—or cars where Advanced Driver Assistance Systems (ADAS) handle such functions as collision avoidance and blind-spot detection.
In the business world, as well, there are significant cost and efficiency advantages to collecting data wirelessly and in real time—say, information about the uptime of the manufacturing lines in your factory—and crunching that data in the cloud, as opposed to having people walking around with clipboards.
At the high end of the automotive market—but quickly moving into the mainstream—cars are using adaptive cruise control to make decisions about safe speeds and safe distances between our cars and the vehicles around them.
Today, we live in a world dominated by connected devices. If I take the example of a smart home, the “smart” is really provided by the user, not the home, through the use of mobile apps or pre-set preferences, causing slow and reactive responses. The red line in the graph below shows where the intelligence of a connected device lies, any level of intelligence, even if it’s trivial.
The next stage of the Smart Revolution is already well underway: Building products and designing software on Artificial Intelligence (AI) platforms. Companies such as Amazon Web Services (AWS) and Microsoft, with its Azure cloud computing platform, offer a broad range of services for the IoT including machine learning, cloud analytics, and large-scale node connectivity through specialized cloud hubs.
The beautiful thing about an AI-based platform is that the more data you feed it, the “smarter” it becomes—and the faster it learns. More and more nodes will send their data to the cloud for processing to perform functions like facial recognition or predictive control and maintenance.
I have a security camera with built-in facial recognition installed in my home. Even with cloud-based intelligence, as a user, I still have to react to my notifications from my app when the camera detects something at the house. This adds latency into my decision making but provides a higher level of cloud-powered intelligence to the node.
What’s next in the “Smart Revolution” is shifting computation muscle out to the edge, and transitioning from platforms where humans drive the decision making, to those capable of making decisions autonomously, learning from their environment and changing their behavior on the fly. This is what I call “Contextual intelligence,” an environment with true device independence, as captured in the image below.
There will be enormous opportunities in the semiconductor industry as the IoT and AI revolutions continue to accelerate. Just as the demand for better, faster, more feature-rich PCs and mobile phones drove the need for more powerful microprocessors, microcontrollers and denser memories, the demand for more powerful edge computing and smarter, more autonomous nodes will require more powerful sensors and MCUs, along with more robust wireless connectivity.
The ability of a semiconductor company to bundle competencies such as low power, programmability and sophisticated device security, and offer them to customers in system-level solutions, will matter just as much as our ability to introduce discrete new products in this environment. Equally as important, powerful, easy-to-use software will matter as much as hardware.
New generations of devices that are able to Sense, Connect, Learn and Respond will be required. This will change semiconductor companies as we know them today.
Not every company will succeed in transforming itself in this way, but we’ve positioned Cypress to compete and win in this new environment.
Much more on that in future posts.