It can be hard for industry observers to look beyond the uncertain geopolitical pressures which affect the global economy. The geopolitical power balance looks set to shift in unpredictable ways.

But underneath the febrile surface of political affairs, there are long-term technological trends at play, which will drive growth in the global economy for the rest of this decade and beyond. The two most important are the wider adoption of artificial intelligence (AI), and the drive towards electrification.

The electrification trend is in part a response to the climate crisis, and the need to dramatically reduce the use of fossil fuels. But across many applications—from electric vehicles (EVs) to heat pumps to industrial processes—users find that the electric alternative to fossil fuel power is more reliable, performs better, and costs less to run and maintain, as well as being more environmentally sustainable, especially when electricity is generated from non-fossil sources such as solar or nuclear.

Whereas electrification is a question of broadly implementing technological solutions which are already well developed and understood, in AI we are at the early stages of learning what the technology can do, and how to implement it. But what we already know about the potential of AI is incredibly exciting.

Public attention has been fixated on generative AI, the large language models (LLMs), and self-driving cars— ‘Big AI’, where inferencing requires the enormous compute resources of a data center.

But consumers will see at least as big an impact on their everyday lives from the implementation of AI at the edge. It could be by enabling new, more intuitive user interfaces, which recognize natural language commands and gestures; creating wearable devices that can process audio intelligently so that you can converse more comfortably or listen to music in more comfort; or wireless sensors in smart cities, which can intelligently make sense of many streams of real-time data about the local environment.

Power/performance Ratio Determines Winners and Losers for Edge AI Devices

There are widespread concerns about the enormous energy requirements of Big AI’s data centers. It’s not inconceivable that constrained energy supplies or government regulation could put the brakes on growth of Big AI’s centralized infrastructure.

But at the edge, the pressure from the AI industry is to use less energy, not more: many edge applications are battery-powered, so embedded device OEMs are trying to figure out how to do more AI, faster, and with less power.

This means that the semiconductor companies that can provide the biggest and fastest improvements in the edge AI power/performance ratio will be the winners in this market.

This efficient AI capability is going to be required in 2025 and beyond because the amount of data generated by edge devices is growing exponentially. The total volume of data created is forecast to be 150 zettabytes (ZB) in 2025, and to climb to over 500 ZB in 2030.

The more that data is produced, the more essential the role devices will play at the edge in streamlining, curating, and contextualizing the data as applied to their AI and machine learning (ML) engines to produce meaningful and immediate insights while operating in relatively constrained environments that are typically powered by batteries. Without thoughtful management of data and power, the sheer amount of data to be processed at the edge would not make it possible to meet power consumption and inference latency targets.

A New Perspective on the Function of Personal Technology

We expect to see a surge in demand for AI-optimized embedded systems in 2025 to support new implementations of AI at the edge. Further ahead, it is possible to foresee a wider reshaping of the world of personal technology. Demand for compute capability could be set to move from CPU- and GPU-powered hubs to an environment in which device users’ interaction with the digital domain is via devices running a real-time operating system (RTOS).

This more-distributed compute environment will allow for the user’s experience of AI-driven functions to feel seamless and fast. The use of devices will feel more natural, embedding technology in the fabric of the user’s life.

This will represent a different usage model for personal technology: until now, the basic operating mode of personal devices such as the smartphone or PC has been to interrupt the user and repeatedly draw their attention with notifications and alerts. When AI-optimized embedded devices take center stage, technology will be a natural part of the user’s life, rather than something that intrudes on their life.

This has the potential to affect people’s lives profoundly: as more tasks are delegated to smart devices, which are embedded in the environment and can communicate with each other—whether in health monitoring, energy optimization, security, or other applications—users’ dependence on smartphone apps and PC-based internet services will lessen. A new diffused, decentralized intelligence will challenge the monopoly that centralized devices and services currently hold.

And by interacting with technology in a more seamless and less intrusive way, new AI-driven embedded devices promise to improve life for consumers and businesses alike.

By: DocMemory
Copyright © 2023 CST, Inc. All Rights Reserved