In mid-January, Quanta Computer EVP and QCT President Mike Yang attended It’s Showtime. “It’s Showtime” is an online interview show hosted by Digitimes that invites tech leaders to talk about the hottest topics in the industry. This episode that lasted about 40 minutes, the longest interview on the platform to date, tackled questions around an AI bubble, AI system design, supply chains, and strategic planning around AI.

No bursting AI bubble. Compute power demand extends from data center to edge

To start off, Mike answered the host’s question about the rumors of an “AI bubble bursting,” by saying there is no “AI bubble.” He supported this claim by saying AI infrastructure demand still remains high as all tech giants fear of missing out (FOMO), new applications are emerging at the edge accelerating monetization and fueling the AI demand.  

AI brings a paradigm shift to system design

As to how today’s system designs differ from previous CPU-based servers, Mike said that the evolution from CPU-based to GPU- and ASIC-based systems represents an escalating curve of technological difficulties. Today’s ODMs must co-develop and co-architect with GPU vendors as early as their design stage to ensure all power, thermal, and networking requirements are met in time for production. 

Fig. 1 Quanta Computer EVP and QCT President Mike Yang talks about increasing design challenges from CPU to GPU and ASIC-based server systems

Quanta and QCT bridge supply chain partners from upstream to downstream

Following the paradigm shift from system-level design to cluster-based total solution production, Mike said the entire tech industry from semiconductors to system manufacturers must collaborate closer than ever to develop new platforms, in order to catch up with the fast turnaround time of GPU servers. Quanta and QCT as system manufacturers happen to be in the right place to integrate the entire supply chain.

Fig. 2 Mike emphasizes deeper ecosystem collaboration as an indispensable part of supply chain integration

Quanta has a comprehensive global footprint for strategic planning 

Quanta has developed a global footprint as early as the early 2000s to enable fast capacity transfers. As of now, Thailand has become the most important SMT manufacturing site for the group. In the United States, we have 30 plants in two manufacturing bases to service CSP customers near where they are located. Mexico is our latest attempt at diversifying manufacturing investments, where mass production will begin in Q.1 2026. Back home in Taiwan, Quanta invests heavily in R&D human resources. By increasing R&D staff, Quanta is expanding its design capacity to keep pace with surging customer demands.

Fig.3 Mike unveils Quanta’s investments in different regions from a strategic perspective

Quanta & QCT seize a winning edge in edge AI with comprehensive 5G solutions

As the focus of AI gradually shifts from data center training to edge inference, Mike is highly confident about Quanta and QCT’s 5G + AI product offerings and their potential to fulfill edge AI applications. With a dedicated edge server product line and end-to-end telco solutions with comprehensive URLLC and network slicing features, QCT has also worked with SI and ISV partners to validate 5G + AI use cases in diverse verticals to develop pre-optimized full-stack solutions for edge AI.

At the end of the interview, Mike gave a promising prediction of the AI industry. Despite great uncertainties regarding fast-evolving hardware and software technologies, the paradigm shift from cloud server to AI server has reached full circle. In the next 5 years, he anticipates data centers to grow larger in volume to serve as AI factories, sovereign AI to get more attention as AI power will be associated with a nation’s power, and more AI applications will emerge and mature to drive sustainable demand for both GPU servers and edge servers.  

Click to watch the full interview on YouTube (in Chinese):