ASE Group Adopts AMD Processors, Evaluates Instinct MI300

According to reports, semiconductor assembly and test leader ASE Technology Holding has fully integrated AMD EPYC processors into its server infrastructure and adopted Ryzen CPUs across its client PCs and laptops.

This infrastructure transition has delivered significant operational and financial benefits for ASE Technology. Compared to previous systems, the company achieved a 50% improvement in system performance while reducing power consumption by 6.5%. These enhancements have collectively lowered the total cost of ownership by 30%.

Although AMD has not disclosed the exact processors previously used by ASE Technology or confirmed full migration across all systems, the notable performance gains indicate substantial adoption of AMD solutions.

Jekyll Chen, Director of IT Infrastructure at ASE Technology, explained that the company handles extensive data analysis, including AI applications and advanced smart factory technologies. Serving numerous semiconductor clients, ASE Technology faces stringent demands for high performance, low latency, and high core counts, while also adhering to its environmental, social, and governance (ESG) commitments. Stability and scalability remain the two primary objectives for their IT infrastructure.

Chen further emphasized that during the evaluation of new servers, multiple factors were carefully assessed, including performance, stability, core count, efficiency, total cost of ownership, AI processing speed, and multitasking capabilities. ASE Technology's decision to select AMD was based on these rigorous criteria, confirming that AMD processors meet their demanding computational requirements.

Notably, as the world's largest outsourced semiconductor assembly and test (OSAT) provider, ASE Technology is actively evaluating AMD's Instinct MI300 series processors to support its growing internal AI workload. AMD noted that many companies are adopting or assessing their Instinct processors for on-premises AI inference.

ASE Technology may be the first major company to publicly confirm evaluation of such large-scale accelerators, suggesting they are nearing formal adoption of the Instinct MI300 GPUs for their internal AI tasks.

Chen also highlighted ASE Technology's AI chip needs, stressing the necessity of executing complex data processing, running AI algorithms, and ensuring smooth, efficient, and flexible operation of smart factories. For client PCs, the focus is on meeting engineering design requirements and achieving high-performance digital transformation goals. ASE Technology's positive assessment of the Instinct MI300 series demonstrates its proactive deployment of cutting-edge AI infrastructure to strengthen its semiconductor industry leadership.

The report underscores ASE Technology's pivotal role in the global semiconductor sector, with packaging facilities across China, Japan, South Korea, Malaysia, Singapore, and Taiwan. The longstanding collaboration between ASE Technology and AMD dates back to 2007 with joint work on advanced 2.5D packaging technology. This enduring partnership significantly contributed to the development of high-bandwidth memory (HBM), a critical technology for modern high-performance computing and AI accelerators.