2025 COMPUTEX Compal Unveils 'The Race is On' Strategy During Forum Session

Estimated reading time: 2 minutes

As AI models and computing density continue to grow at an exponential rate, the data center industry is entering a new era where cooling capacity has become the next competitive threshold. At the “AI in Action” forum during COMPUTEX 2025, Compal Electronics Inc., Alan Chang, Vice President of the Infrastructure Solutions Business Group, delivered a keynote titled “Data Center for the Next AI: The Race is On”, emphasizing that the challenge brought by AI has already surpassed traditional compute performance, and energy efficiency and thermal management are now the core battlefield for next-generation data centers.

According to McKinsey, global data center infrastructure investment is projected to exceed USD 7 trillion between 2025 and 2030, with more than USD 5 trillion directed toward AI-specific infrastructure. Amid this surge in demand, chip thermal design power (TDP) has climbed from 500W to beyond 1,000W, and rack-level power consumption could potentially exceed 150kW, posing unprecedented challenges for both cooling and energy allocation.

To address these demands, Compal showcased the evolutionary roadmap of its liquid cooling infrastructure. The presentation outlined the transition from traditional air cooling to single-phase liquid cooling, and projected the future development of two-phase cooling and immersion cooling. Looking ahead, Direct Liquid Cooling (DLC) is expected to advance beyond the system level toward integration at the chip and packaging level, forming a truly integrated “DLC on Silicon” thermal architecture to support the increasing thermal loads driven by large-scale AI models.

Industry research shows that cooling capacity has increased from approximately 1,000W with air cooling to over 3,000W with liquid cooling (IEA, C&W, 2024), providing the necessary thermal support for high-density AI workloads and large-scale generative model training.

In addition to technological innovation, Compal also emphasized its global manufacturing and service network as a critical foundation for AI infrastructure. The company currently operates data center platform manufacturing and support facilities in the United States, Taiwan, Vietnam, Mexico, and Poland, covering a full spectrum of applications from personal devices and edge nodes to high-performance and quantum computing, with capabilities for rapid deployment and flexible delivery.

As data sovereignty, cybersecurity, edge AI, and sustainability goals continue to shape the industry, data center architectures are rapidly shifting from centralized to decentralized models. Generative AI applications are expanding from the cloud to verticals such as smart cities, healthcare, telecommunications, and finance, placing higher demands on real-time inference and local energy optimization at the edge. Leveraging its advantages in liquid cooling, server platforms, and global deployment, Compal helps customers build next-generation AI infrastructure with scalability and flexibility.

Alan Chang concluded, “AI is redefining the rules of competition for data centers. Compal is not just a hardware provider but a strategic partner advancing both energy efficiency and compute performance. Compute is a race of speed, but cooling is a race of endurance.”

The evolution of AI technology is driving data centers into a new generation. Compal not only demonstrates forward-looking deployment in hardware design and thermal management, but also adopts a system-level approach to address the key challenges of compute density, energy efficiency, and sustainable development. Through liquid cooling architecture, smart control integration, and global supply chain synergy, Compal is working hand in hand with its customers to build a new generation of AI data centers that are built to last and ready to scale—moving toward a future of intelligent computing and sustainability.