“The embedded business is a business that succeeds when the customer’s device business succeeds. AMD's Embedded Business Unit is integrating AI engines to enable edge AI and support its success across various industries.

Hee-man Lee, CEO of AMD Korea, delivered a keynote speech at the 'AMD Embedded Computing Tech Day' event held on the 16th, revealing examples of AI utilization in edge devices and AMD's strategic direction.

AMD operates across four business segments: data center, PC, gaming, and embedded.

Among them, the embedded business division has been growing rapidly in recent years by providing high-performance computing to industrial devices through adaptive computing technology based on SOC and FPGA.

The combination of CPUs and GPUs, along with the integration of AI engines, enables the development of custom SOCs and plays a key role in providing customized solutions.

Chiplet technology, in particular, is at the heart of AMD's next-generation strategy. Chiplets are a structure that modularizes various processing areas and allows customers to combine their IP and technologies, and are expected to significantly increase the flexibility and expandability of future SOC designs.

According to CEO Lee Hee-man, AMD is collaborating with companies across various industries to realize edge AI technology.

A representative example is the Japanese automobile manufacturer Subaru.

Subaru is implementing an AI-based real-time decision-making system by equipping its autonomous vehicles with AMD's high-performance processors, aiming for "accident-free driving."

Another partner, StradVision, is a company developing vision recognition technology for autonomous vehicles. It is building a system that utilizes AMD's GPUs and AI engines to quickly analyze and assess the vehicle's surroundings.

In the medical field, manufacturers of portable ultrasound devices are incorporating AI capabilities based on AMD's SOCs. This helps even non-experts make accurate diagnoses and increases accessibility to medical care.

In the field of industrial robotics, various devices, including railway track inspection robots, logistics center automation systems, and fisheries sorting robots, are being developed based on AMD technology.

Especially in the defense and healthcare industries, where personal information protection and real-time processing capabilities are crucial, the application of edge AI is essential.

CEO Lee Hee-man cited humanoid robots as the most complex form of edge AI device.

The 26 axes of rotation required to implement hand movements, the integrated processing of various sensor information, and the wireless actuator control are technically very challenging tasks.

To achieve this, three functions – data preprocessing, artificial intelligence judgment, and motor control – must work organically, and AMD’s goal is to implement this in a single chip.

Lightweight and battery-efficient edge AI devices are essential.

Regarding this, CEO Lee Hee-man said, “AMD is developing technology to increase the performance of AI-dedicated chips while reducing power consumption,” and “In particular, the AI H series chips aim to achieve both lightweight and high performance by integrating three functions.”

"Based on the philosophy that our customers' success is our own, AMD prioritizes trust with our partners," said CEO Lee Hee-man. "The embedded business is an area that cannot be pursued alone. Technology development and support tailored to customer needs are key. We will diligently support our customers to achieve success and differentiate our products."