Building Enterprise-Centric AI Applications Is Key to Turning Technology into Competitive Advantage

---

As one of the organizers of COMPUTEX and the InnoVEX startup showcase, the Taipei Computer Association (TCA) emphasized that the fast-evolving landscape of Generative AI (GenAI) and Multimodal AI is pushing businesses to accelerate their digital transformation journeys. According to McKinsey & Company’s Superagency in the Workplace report, AI is expected to unlock $4.4 trillion USD in long-term value, transforming not only the roles of executives and employees but entire business models—much like how steam engines revolutionized industry in the 19th century.

To help business and ICT leaders formulate strategic AI adoption plans suited to their operations, TCA hosted the AI NEXT FORUM – Afternoon Session: “Enterprise AI Centers — The New Goldmine” on May 6 at Hanlai Hotel Taipei. The session featured experts including Chris Hung, Director General of the Market Intelligence & Consulting Institute (MIC); Miller Chang, President of Embedded Sector, Advantech; Samson Hu, Co-CEO of ASUS; Steve Lin, Chairman of Auras Technology; and Ares Chen, Vice President and General Manager of Delta’s Power and Systems Business Group. Topics covered breakthroughs in AI 2.0, edge AI in robotics, AI infrastructure, data center power and cooling, and more.

To give attendees hands-on insights into enterprise AI deployment, the forum showcased solutions from Intel, ASRockRack, AsiaRF, and Z-COM. Exhibits included the latest Intel Core Ultra and Intel Xeon processors, NVIDIA Blackwell enterprise AI servers, AI-powered edge surveillance appliances, and AI edge computing platforms. The National Science and Technology Council (NSTC)’s IC Taiwan Grand Challenge (ICTGC) also introduced its third call for submissions on-site.

Tailored AI Applications Are Essential for Turning Technology into Business Value

In his opening remarks, Tzu-Hsien Tung, Honorary Chairman of TCA, emphasized that Taiwan has played—and must continue to play—a pivotal role in the AI revolution. From the early days of personal computers to today’s AIoT and GenAI boom, Taiwan has consistently contributed to the global digital ecosystem. He noted that while Taiwan excels in hardware, it must now focus on enterprise-level AI application development to stay competitive.

Tung explained that AI development relies on three pillars: computing power, algorithms, and data. Taiwan already dominates in foundational infrastructure—semiconductors, PCBs, systems, and server manufacturing. The rise of AI can be traced back to the 2017 introduction of the Transformer algorithm, which gained global traction with OpenAI’s GPT-3.5. Now, as conditions for data, compute, and algorithm maturity converge, AI is moving from theory into vertical applications. Businesses must build their own AI centers to transform technology into sustainable advantage.

Edge AI, Human-Centered Design, and Diverse Applications Will Drive Adoption

Chris Hung of MIC delivered a keynote on “The New AI Era: Taiwan's AI Hardware & Software Landscape and Future Outlook.” He noted that large language models (LLMs) and small domain-specific models (SLMs) are evolving rapidly, with open-source versions now rivaling proprietary offerings. AI applications, he added, will become increasingly human-centric, integrating into both daily life and industrial settings through AI agents, edge devices, and humanoid robotics.

Hung cited Taiwan-based use cases such as:
– Pathology diagnostics (CytoAurora)
– Collaborative robots (Techman Robot)
– Software-defined factories (MetAI)
– Risk alert systems (Fubon Bank)
– AI-powered animation (Ziboo)

He projected that AI PC demand and new model innovation—such as DeepSeek—will further drive adoption in 2025. In turn, edge AI and localized data centers will gain traction, enabling scalable and energy-efficient solutions tailored for specific applications.

Advantech Empowers Robotics with Modular Edge AI Platforms

Miller Chang, President of Embedded Sector at Advantech, outlined the future of robotics and edge AI in his talk “Edge AI-Powered Robotics in Action.” He noted that the global Edge AI market could reach USD 143.6 billion by 2032, driven by adoption in smart factories, healthcare, cities, and robotics—including AGVs, AMRs, robotic arms, and humanoid robots.

To support this trend, Advantech introduced a modular robotics architecture, offering building blocks such as AI platforms, perception-control modules, and full system integration. The company also provides a Robotic Suite SDK to accelerate time-to-market for industrial robot solutions. Chang emphasized that Edge + AI will become the industry’s new standard.

ASUS: Local AI Data Centers Empower Enterprise Privacy and Customization

Samson Hu, Co-CEO of ASUS, spoke on “AI Data Center Layout and Development.” He explained how the evolution from GenAI to Agentic AI, Physical AI, and ultimately AGI is driving infrastructure needs both in the cloud and on-premise.

ASUS offers full-stack AI server solutions—from hardware and platforms to deployment and maintenance. Since 2011, the company has built several AI data centers in Taiwan and recently contributed to Ubilink, Taiwan’s largest AI supercomputing center, delivering 45.82 PFLOPS. ASUS also provides enterprise-grade AI servers optimized for fine-tuning LLMs and SLMs, and developed the Ascent GX10, a compact AI supercomputer, along with the Multi-LM Tuner—a tool enabling local model tuning to meet enterprise needs.

Auras Technology: Liquid Cooling Is Now Inevitable in AI Data Centers

Steve Lin, Chairman of Auras Technology, noted that as “compute is national power,” power-hungry AI servers are reshaping infrastructure needs. He emphasized that traditional air-cooling systems are no longer sufficient for high TDPs (thermal design power), and that liquid cooling has become the default trend.

For example, NVIDIA’s NVL72 consumes 132kW, making air cooling impractical. Future racks may exceed 300kW, necessitating advanced liquid cooling systems to ensure performance and energy efficiency.

Delta: Boosting AI Data Center Efficiency with High-Voltage DC and Rack-Level Capacitors

Ares Chen, VP and GM of Delta’s Power and Systems Business Group, shared insights on “AI Computing Challenges: Power and Thermal Solutions.” He noted that AI data centers are expected to consume 261 TWh globally in 2024—nearly equal to Taiwan’s annual energy consumption.

Delta’s Grid-to-Chip power architecture simplifies power conversion, from high-voltage grids to the low-voltage requirements of AI chips. The system includes high-voltage DC (800V), DC-DC modules, rack-mounted capacitors, and intelligent load balancing—improving end-to-end efficiency from 87.6% to as high as 92%.

Watch the AI NEXT FORUM Replays

>>COMPUTEX AI NEXT FORUM – Chinese Version (YouTube) https://youtu.be/BmLiHItX2Cw
>>COMPUTEX AI NEXT FORUM – English Version (YouTube) https://youtu.be/k9ZR96VSSYY

Pre-Register Now for COMPUTEX & InnoVEX 2025

To stay updated and gain free admission to COMPUTEX and InnoVEX 2025, international buyers and professionals are encouraged to pre-register online. Approved registrants will enjoy full access to the show and receive the latest news via the official website and newsletters.

Pre-registration: https://www.computexonline.com.

■About COMPUTEX TAIPEI & InnoVEX

COMPUTEX TAIPEI, initiated under the leadership of TCA Chairman Stan Shih in 1985, is co-organized by TCA and TAITRA.
InnoVEX, launched in 2016, is Taiwan’s premier platform for tech startups. It connects innovators with global partners and accelerates industry collaboration to foster a thriving innovation ecosystem.

Back