Interview | Taiwan's AI Chief Explains Why Sovereign Tech Holds Key to Island's Future

In an era where artificial intelligence is reshaping global influence, Taiwan is racing to establish its own "sovereign AI" framework, viewing it as essential for safeguarding economic growth, societal resilience, and defense priorities. This push stems from lessons learned in semiconductors and personal computing, now applied to the high-stakes world of computing power and advanced models.

Wu Han-chang(吳漢章), who leads ASUS Cloud and Taiwan AI Cloud(華碩雲端暨台智雲)—one of the island's flagship AI infrastructure initiatives—shared these insights in an in-depth discussion ahead of Storm Media's upcoming forum on geopolitical shifts and technological disruption. Set for March 11, the event will bring together Wu, former Premier Sean Chen, MiTAC Chairman Chou Wei-kun, and Kneron founder Albert Liu to unpack U.S.-China rivalries, trade barriers, and AI's transformative role.

Taiwan's journey toward AI independence began well before the current hype, Wu explained. As far back as 2018, government contracts spurred local firms to develop GPU servers and supporting ecosystems, mirroring the strategies that built the island's dominance in chips and hardware. The goal was clear: foster end-to-end integration to avoid dependency on foreign suppliers.

By 2023, the rise of generative AI amplified these efforts, shifting focus to "model sovereignty." Unlike basic cloud storage, AI models shape user experiences, embedding cultural nuances, linguistic preferences, and societal values. "Outsourcing this risks losing control over information flows that could undermine national identity," Wu cautioned, equating AI's reach to that of military assets. For Taiwan, maintaining domestic oversight isn't optional—it's a bulwark against external vulnerabilities.

Beyond geopolitics, AI offers practical answers to Taiwan's pressing internal challenges, particularly its aging population and shrinking workforce. Public sectors like healthcare, education, and long-term care are straining under labor gaps that digitization alone can't bridge.

Wu pointed to real-world gains: In the U.S., the Food and Drug Administration has slashed drug review times from weeks to hours using AI, while similar tools optimize hospital scheduling, analyze records, and dispense medications. "Productivity leaps like these are our only scalable fix," he said, positioning AI as "new infrastructure" vital for stability. Without sovereign control, disruptions could ripple into security threats, underscoring why Taiwan must prioritize homegrown solutions.

Contrary to perceptions of scarcity, Taiwan's AI investments—spanning public and private sectors through 2029—rival those of leading economies. The issue, Wu argued, lies in access. High-end GPU servers carry steep prices, limiting uptake among smaller players.

In contrast, U.S. giants like OpenAI, Google, and Meta channel billions into model training, absorbing vast resources. Taiwan's manufacturing-heavy economy, with its modest R&D spending, fragments demand. Here, government intervention is pivotal: through funding, shared platforms, and incentives, it can democratize access for startups and SMEs. "Letting markets decide would entrench inequality and stall ecosystem growth," Wu warned.

Wu frames AI's evolution in two stages. The first, centered on tasks like facial recognition, matured technically but barely altered workflows. The second—powered by large language models—delivers tangible involvement in creativity, decisions, and operations.

"Users once dreamed of possibilities; now they're surprised by AI's depth," he observed. This shift accelerated in 2024-2025, fueled by open-source advancements nearing commercial quality, plummeting costs, and rising global tensions pushing localization. What began as buzzwords has become actionable policy, with "sovereign AI" driving investments worldwide.

Skeptics question AI's profitability timeline, but Wu attributes the unease to outdated metrics. Traditional finance overlooks how AI boosts efficiency, innovation, and risk mitigation—benefits that don't immediately show on balance sheets.

A natural delay exists between tech adoption and revenue gains, he noted, drawing parallels to historical indicators like GDP. "Economists will devise better tools to capture long-term impact," Wu predicted, urging patience amid corporate pressures.

Taiwan's semiconductor prowess is legendary, but Wu highlighted two overlooked assets in the AI arena. First, its "engineering economics"—the knack for turning complex science into affordable, reliable products. This spans optics, algorithms, and assembly, honed over four decades in everything from chip fabs to supercomputers.

Second, a vibrant yet undervalued network of about 1,000 B2B software firms embedded in diverse industries. These act as bridges, adapting AI for real-world applications in manufacturing and beyond—a "last-mile" advantage few nations possess.

Still, these companies grapple with generational handovers, skill updates, and tech refreshes. Wu suggested remedies: automated tools, like those from Taiwan AI Cloud, to migrate legacy code without full overhauls. Additionally, state-backed consolidation could merge niches into scaled-up entities with greater resources. Challenges persist, though: "Not everyone wants to give up being the boss," he quipped.

As AI integrates deeper into daily life, Taiwan's strategy emphasizes self-reliance, blending foresight with practical execution. For an island navigating superpowers, sovereign tech isn't just about competition—it's about survival.

Addressing market doubts about AI commercialization progress, Wu acknowledged that "when to monetize" represents the most anxiety-inducing question, but attributed this to misplaced measurement framework standards. Current financial models and investment evaluation methods poorly reflect AI's contributions to productivity and long-term value, creating enormous pressure for corporate budget decisions.

Time gaps naturally exist between new technology enhancing productivity and converting to financial benefits, he explained. Markets need new measurement tools evaluating AI investment's long-term contributions to organizational efficiency, innovation capacity, and risk management, similar to how GDP emerged as an economic indicator. "Economists will solve these problems," Wu said.

Beyond Taiwan's well-known semiconductor dominance, Wu highlighted two frequently underestimated yet crucial competitive advantages in global AI competition. First is robust engineering economics and system integration capabilities. AI represents a "total assembly" of optics, chips, software, algorithms, and multiple technologies. Converting cutting-edge science into stable, reliable, cost-effective products represents Taiwan's core strength accumulated over 40 years, from chip manufacturing and GPU server assembly to supercomputer construction and maintenance—integration capabilities gaining value in the AI era.

Second is an underestimated B2B software company ecosystem. Despite Taiwan's reputation for "strong hardware, weak software," approximately 1,000 small and medium B2B software companies penetrate various industries. Though small-scale, they represent the crucial "last mile" for AI technology deployment into physical industries and serve as indispensable intermediaries—a deeply-rooted industrial software ecosystem that most countries lack.

However, these small enterprises face structural challenges including succession planning, employee skills, and technology iteration, making transformation urgent amid the AI wave. Wu proposed two solutions: technical guidance reducing transformation barriers, such as Taiwan AI Cloud's development of tools automatically converting legacy programming languages to modern ones, helping outdated systems rapidly modernize without expensive complete rewrites.

Second is capital integration through National Development Fund-driven industry consolidation, combining thousands of niche small enterprises into several large-scale software companies with economic scale and transformation resources, enhancing overall industry competitiveness. But Wu acknowledged this approach faces challenges: "Everyone likes being the boss!"

（Related： Interview | Taiwan's Drone Industry Has a Brain Problem — And Minwang Technology Is Solving It ｜ Latest ）

​