Accelerations of investment and full-scale adoption

Generative AI, which began to emerge with considerable momentum in 2023, has led to an acceleration of investment in the computing and networking areas of cloud data centers. However, supply chain benefits have been limited to a small number of players. As implementation of generative AI for edge devices progresses, the base of companies that will benefit should expand, with benefits coming in areas such as improved product mixes (increased computational processing, enhanced storage, upgraded peripheral devices) and a speeding up of replacement cycles. We expect 2024 to be the year in which promotion and adoption by various companies begins, with full-scale adoption of edge AI coming in 2025–2027. For instance, major companies have begun launching a series of edge AI-related products such as mobile devices that can do simultaneous interpretation even when offline.

Upgrades in battery and sensor

Edge AI may help drive an upgrade in battery along with higher computation requirements on the edge side. According to UBS's twice-yearly smartphone purchase preference survey, battery performance is actually the single most important performance feature to consumers. Considering the increased computing demands associated with AI implementation and the possibility of an attendant increase in power consumption, it is natural to expect the importance of battery life to become even more important in the future.

Discussions with experts have also implied that the importance of microelectromechanical systems (MEMS) microphones for capturing audio data will increase when simultaneous interpretation functions are implemented on terminals or in the course of improving the quality of video conferencing. It will be important to increase the number of MEMS microphones but also to improve the accuracy of microphones to achieve real-time speech recognition for simultaneous interpretation. Sensors that recognise the direction of vocalisations is required for accurate voice recognition. Thus, the importance of sensors when AI is implemented in terminals is likely to increase.

In addition, the adoption of MEMS sensors is expected to expand in fields other than consumer devices, for instance in automotive applications. The Global Navigation Satellite System is used to detect the position of vehicles in real time, but radio waves may be disrupted by buildings, tunnels and other infrastructure. To achieve advanced autonomous driving technology, it will become necessary for a vehicle itself to accurately detect its own position. The use of MEMS motion sensors is therefore expected to expand in the future to process information collected by vehicles in real time.

What are the advantages of implementing AI at the edge?

  • Low latency (because of high-speed processing on the edge side)
  • Reduction of loads on communication bandwidth and lower costs
  • Enhanced privacy and safety of personal data
  • Improved productivity
  • High-level personalisation/customisation

Potential applications include:

  • Content and multimedia generation on the device
  • Simultaneous interpretation and transcription
  • Meeting note-taking
  • Improving the quality of video conferencing
  • AI-powered games
  • Increased productivity with Co-pilot

We expect the penetration rate of generative AI-enabled personal computers will be 60% by 2027, with the penetration rate of AI-ready smartphones at 46% in that year.


Explore other articles you may find interesting