[Peace&Chips] The New Growth Engine 'On-Device AI' Spotlighted by Samsung Semiconductor

'On-Device AI' Implemented on Devices
Faster, Smaller Low-Power Semiconductors Needed
Samsung Electronics "LLW DRAM Targeting Mass Production Next Year"

Editor's NoteSemiconductors, known as the rice of modern industry. Although it's a term we hear every day, it's often hard to explain. Peace & Chips will make the complex concepts and overall trends of the semiconductor industry easy to digest for you. Just bring your spoon.

"While generative AI is developing based on servers, on-device AI, which implements AI directly on devices for efficiency and security reasons, is also expanding." This was said by Kim Jae-jun, Vice President of Samsung Electronics Memory Business, during the conference following the Q2 earnings announcement last week. Introducing the memory business being promoted in the AI era, he highlighted on-device AI as a major AI trend recently gaining attention.

Earlier, Kyung Kye-hyun, President of Samsung Electronics DS Division, also mentioned on-device AI. Through his social media account this month, he stated, "Starting from generative AI in the cloud, discussions on on-device AI at the edge are actively progressing," adding, "It is time for us to seriously consider what more we need to do to create and capture value in the AI arena." What exactly is on-device AI that it is being mentioned like this?

Samsung Electronics Hwaseong Campus view / Photo by Samsung Electronics

On-device AI refers to AI that can operate on various devices without network connectivity. Generally, AI services we know are implemented by processing large volumes of data on high-performance servers, making network connection essential. However, on-device AI collects and computes data independently on the device without going through servers to perform necessary functions. This enables much faster service.

Also, since data does not need to be sent to servers, security issues such as personal information or sensitive data leakage can be resolved. It can also reduce energy consumption inevitably caused by processing data on servers. The expansion of on-device AI is considered essential to realize the future envisioned by AI technology advancements.

Recently, the semiconductor industry has high expectations for the increased demand for high-performance semiconductors driven by generative AI, and on-device AI is expected to bring equally significant business opportunities. Some even foresee it forming a larger market. "From drones and robot vacuum cleaners to autonomous vehicles, on-device AI will become increasingly important for all devices that must operate even when network connections are lost," they say, actively developing related technologies.

For on-device AI to be well implemented on various devices, semiconductors at the core of the system must perform their roles well. To achieve this, semiconductor chips need to be faster in response speed, consume less power, and be smaller in size than existing products. From the perspective of memory semiconductors, this is similar to the technical challenges previously required for mobile DRAM.

Samsung Electronics, in this conference call, announced plans to develop DRAM specialized for on-device AI and mass-produce it by the end of next year. This is the LLW (Low Latency Wide IO) DRAM. This product has significantly higher bandwidth than existing mobile DRAM, which offers great advantages in real-time processing of data generated on devices. Bandwidth is like the road through which data flows, and the higher the bandwidth, the faster data can be transmitted. We can look forward to new semiconductors emerging alongside on-device AI.

This article is from [Peace & Chips], published weekly by Asia Economy. Click subscribe to receive articles for free.

☞Subscribe

This content was produced with the assistance of AI translation services.