In 1997, Wired magazine featured a cover story titled, “The Long Boom.” Some of its predictions came true – “An uncontrollable plague – a modern-day influenza epidemic or its equivalent takes off like wildfire,” – while others did not. Critics complained that the piece was overly optimistic in its view of the benefits of technology.
Twenty-five years later, even with the predicted pandemic, we’re still in the midst of that long boom – fueled by astonishing innovation in mobile devices, battery technologies, artificial intelligence, machine learning, and much, much more.
So, what’s next? To find out, we reached out to people who are making it happen: The Arm partner ecosystem. Here’s what some of them said.
Dave Garrett, chief architect with IoT AI innovator Syntiant, was influenced in part by the Wired article mentioned above.
Garrett’s prediction is something most of the industry can agree on: “I’m going to go out on a limb and say that machine learning is going to continue to make your life better in a billion ways, and you’re not even going to know about it. It will be this vague feeling that things around you are easier, faster, better, but at the core, neural-driven interfaces are going to make it happen.”
Syntiant delivers Arm-based neural processors into the market that are extremely low power, focusing on voice AI. In that context, the company sees billions of edge devices, packed with sensors, drowning in data, and bumping against energy, data transport, and even cloud compute limits.
“In 2022, you will see all the devices around you continue to hear, see, and feel (microphones, image sensors, accel/gyro sensors) at accuracy levels that will seem magical, all driven from neural networks that are seeing patterns in the data not possible with traditional algorithms,” Garrett said.
Dr. Chris Mitchell, founder and CEO, Audio Analytic, took the audio application a step further.
Products with a greater sense of hearing will deliver exceptional human experiences by recognizing and extending what we can hearDr Chris Mitchell, CEO, Audio Analytic
“Sound is physical vibration, which means it is literally touch from a distance,” he said. “It has a major part to play in ambient computing in 2022, as new device form factors and experiences will challenge the way we think about human-centered design. Products with a greater sense of hearing will deliver exceptional human experiences by recognizing, improving, helping, and extending what we can hear, which in turn will also allow our devices to anticipate our needs or remember things that are important to us.”
There’s little argument that the human-machine interface has evolved in 2022 to a point that few would have predicted just five years ago.
A more refined AI industry roadmap
Predictions around AI and machine learning focus less on how implementations will evolve, but more on what new applications will blossom around them.
Ram Naidu, senior vice president, AI, with Cambridge Consultants, sees AI expanding quickly from the big-compute datacenters.
“Currently, most AI models are trained centrally in datacenters. However, I am starting to see technological advances that mean training can take place on the endpoint itself. This is the direction of travel for many use cases,” he said. “It will allow for customizable performance and support sustainability by allowing training to take place on low power devices.”
Roeland Nusselder, co-founder and CEO of PlumerAI, whose technology aims to push deep learning to the tiniest edge and endpoint devices, sees AI beginning to flourish at the edge.
“Microcontrollers will take on more AI workloads,” he predicted. “Computer vision will come to microcontrollers, as the technical challenges that made it impossible to deploy computer vision AI on microcontrollers have been resolved.”
Computer vision will come to microcontrollers, as the technical challenges that made it impossible to deploy computer vision AI on microcontrollers have been resolved.Roeland Nusselder, co-founder and CEO, PlumerAI
As an example, he noted that until recently, surveillance cameras were discrete devices tethered to videotape recorders inside the buildings they monitored. Then AI was added, first in the cloud, and then in the camera itself for privacy and bandwidth reasons. The next step is to turn these cameras into low-cost, low-power intelligent sensors and remove the remote storage and viewing capabilities, which in many applications is not needed.
“The camera has become an intelligent sensor, which can run on a microcontroller and can run on a battery, at much lower cost,” Nusselder said.
Nusselder also predicts that consumer electronics will become person-aware.
“A light only needs to be on when there are people around. Same for the heater, AC, the TV, and so forth,” he said. “Just like a refrigerator beeps when you forget to close the door, a stovetop can start beeping when you forget the meal you’re cooking.”
“It will become increasingly clear that AI for the real world requires top-notch data curation and verification,” he added. “The community will shift its focus from inventing new neural network model architectures to new data-centric techniques.”
The big squeeze
The chip shortage will continue to affect the consumer electronics industry throughout the yearÇağatay Büyüktopçu, R&D cyber security technology manager, Arçelik
Arçelik Global, an international household appliances manufacturer, leverages Arm low-power technology to attack global warming with AI-powered refrigerators. The company is dedicated to this work for years to come, however, in the short term, global economics is becoming a challenge, according to Çağatay Büyüktopçu, Arçelik’s R&D cyber security technology manager.
“One of the major topics on the agenda for 2022 will again be chip shortage,” he said. “We expect it to continue to affect the consumer electronics industry throughout the year.”
“Internet of things (IoT) cyber security and interoperability will also be under discussion,” he added. “We hope to see rapid adoption of EN 303 645 – ETSI’s cybersecurity standard designed for consumer IoT devices – across the industry and in different geographies around the world.”
Büyüktopçu cautioned that if this doesn’t happen, there’s a risk that new, competing standards will arise, fragmenting the market.
From a design productivity standpoint, he noted: “We also predict that new vulnerabilities will be found in the open-source software libraries used for IoT device connectivity. This could affect the security of both edge devices and their cloud services. Software updates for IoT devices will become even more important.”
Regardless of supply and demand, however, there is an inexorable push to distribute compute resources more efficiently from the cloud to the edge and endpoints, according to Roger Thornton, director of applications, at Raspberry Pi.
With computing platforms like Raspberry Pi powered by Arm technologies, we will see IoT compute further pushed from the cloud and onto the powerful devices at the edgeRoger Thornton, Raspberry Pi
“With increased focus on the environment and sustainability we will see companies evaluate the cost of computing and pay more attention to where the computing is done,” he said. “With computing platforms like Raspberry Pi powered by Arm technologies, we will see IoT compute further pushed from the cloud and onto the powerful devices at the edge and endpoint that have a greater level of autonomy to sense and process data without sending it all to the cloud.”
Scaling design productivity
Frank Schirrmeister, senior group director, Solutions and Ecosystem, with Cadence Design Systems, said “scale” is the operative word for 2022. This is driven by varying requirements in the different vertical industries, such as hyperscale computing, consumer, mobile, 5G communications, aerospace/defense, automotive, industrial, and healthcare. And those requirements are, in turn, driven by the explosion in data.
“We are getting used to measuring data transmission in exabytes and its storage in zettabytes,” he said.
Virtual and augmented realities (the nascent metaverse) and rapidly increasing autonomy in automotive and aerospace/defense applications are creating pressures that are sparking new innovation and new approaches to design.
“The network infrastructures carrying all the data and the computing environments creating insights and storing hard-to-image amounts of data in seconds are becoming more and more workload-specific, composable, and software-programmable,” he said. “There are various opportunities to touch and make sense of data at network, far, middle, and near edges, enabling different types of scalable computing with latencies and throughput dictated by the vertical industries they are serving.”
We are getting used to measuring data transmission in exabytes and its storage in zettabytesFrank Schirrmeister, Cadence Design Systems
Of course, no one company can do it all, Schirrmeister noted.
For example, the growing ecosystem of the Open Compute Project, now in its 10th year, brings together 5,000 engineers from 280-plus companies in 29 working groups with Arm and Cadence among its partners.
“Together, we understand the requirements stemming from the desire of consumers to be always on and hyperconnected,” Schirrmeister said. “We, as consumers, can expect a lot from the era of hyperconnectivity and hyperscale computing, and as an ecosystem, the industries of IP, EDA, technical software, semiconductor, and system design are closely intertwined and co-dependent – a vibrant village that is making it all happen.”
Scott White, CEO of PragmatIC, sounded a similar tone around design flexibility.
“Recent events have shown us that being agile is key to success in fast-changing times,” he said. “At PragmatIC Semiconductor, we see demand increasingly driven by the desire for greater agility in hardware design, just as we already expect in software. For example, our FlexIC Foundry order book has more than tripled this year, driven by innovators wanting to rapidly iterate designs to create and optimize new products.”
Toward a prosperous future
Futurist Matthew Griffin, with the 311 Institute, has given us plenty of food for thought in 2021, whether it’s been his view of healthcare, retail, AI, or the metaverse. Looking ahead, he sees a lot of positive social change occurring on the back of technological innovation.
In 2022 we’ll see more of us working from anywhere as standard, companies openly discussing and investing in the health and mental wellbeing of their employees, the continued acceleration of digital transformation, and the embrace of immersive reality technologies and toolsMatt Griffin, Futurist, 311 Institute
“When we went into 2021, few people would have predicted it would end as it began, with the world still reeling, rather than recovering, from the global pandemic,” he said. “However, as we look toward a brighter 2022, my prediction is that we will see more of us working from anywhere as standard, companies openly discussing and investing in the health and mental wellbeing of their employees, the continued acceleration of digital transformation, and the embrace of immersive reality technologies and tools. And we will finally realize the first truly conversational AIs that let us talk and interact with the devices all around us.”
As we stride together into a new year of innovation, we’ll see whether the legendary American baseball player Yogi Berra had a point when he said, “It’s tough to make predictions, especially about the future.”