At Singapore's DBS Bank, every frontline customer service employee works alongside a dedicated AI agent — a digital partner that retrieves and integrates the full history of each incoming caller before the human representative has said a word.
DBS Shows What AI-Integrated Work Already Looks Like
The agent is not a helpdesk chatbot. It operates in real time, running silently in the background, assembling fragmented data from across the bank's systems so the human employee can focus entirely on the customer. No manual lookup. No commands typed into a search field. The AI functions, in essence, as the worker's "digital twin" — an always-present counterpart shaped by the same institutional knowledge, ready to surface whatever the moment requires.
What makes this possible is a deliberate, infrastructure-scale commitment: enormous computing capacity, deep data integration across the bank's systems, and the willingness to redesign workflows from the ground up. DBS made those investments. Taiwan has not.
AI Is Not a Tool Anymore—It Is a New Form of Intelligence
Writing in Storm Media on July 17, columnist Zhang Jingwen argues that the AI revolution deserves a more fundamental frame than the familiar shorthand of "computers learning to talk." She describes it as a civilizational encounter between two forms of intelligence: carbon-based life — humans and all biological organisms — and silicon-based entities, the integrated circuits that underpin modern computation.
What has shifted, Zhang argues, is the arrival of genuine reasoning capacity in machines. AI systems are no longer simply executing fixed instructions; they are beginning to exhibit something closer to comprehension — what technologists call "emergence." When measuring a country's true AI standing, raw computing power is only the baseline. The more revealing question, she writes, is whether that country can cultivate silicon-based intelligence of its own.
In practice, this intelligence takes the form of large language models, autonomous AI agents, and high-efficiency software systems. What they share is not physical form but functional capacity: the ability to analyze, decide, and act without step-by-step human direction. AI agents have become, Zhang argues, a new management paradigm — not just a productivity tool, but a structural force reshaping how institutions at every level are organized and run.
TSMC Supplies the World's AI Hardware; Taiwan's Own Industries Lag
The contradiction at the heart of Zhang's argument is one Taiwan can no longer comfortably defer. Through Taiwan Semiconductor Manufacturing Co. (TSMC), the island produces the advanced chips that power virtually every significant AI system in the world. American and Chinese technology companies depend on Taiwan's fabrication capacity to run the models transforming their economies.
Yet Taiwan's own financial sector, healthcare system, and traditional industries remain distant from the kind of carbon-silicon collaboration that DBS has already made routine. Critical industrial data sits locked inside aging, siloed legacy systems. Regulatory frameworks designed for an earlier technological era have been slow to adapt, leaving companies reluctant to innovate even when the technical means are at hand.
"This is Taiwan's great paradox," Zhang writes. "We used TSMC to build the world's most powerful hardware and computing chips — but Taiwan's own finance, healthcare, and industries have yet to enter the era of intelligent carbon-silicon collaboration."
Sovereign AI Means Owning the Intelligence, Not Just the Chips
The policy gap Zhang identifies leads to a sharper redefinition of what sovereign AI actually requires. The concept is most often discussed in terms of data security — keeping sensitive information beyond foreign reach. Zhang sets a more demanding standard: sovereign AI is a nation's capacity to independently cultivate, train, and control silicon-based intelligence that embodies its own culture, values, and industrial knowledge.
The consequences of failing that standard are concrete. If Taiwan's industries do not build their own AI foundation, they will have to rent intelligence from American or Chinese technology giants. And when a country's workflows, management practices, and decision-making logic must conform to AI systems designed abroad, for foreign markets and foreign ethical frameworks, Zhang argues that country has effectively surrendered digital sovereignty — regardless of who manufactures the underlying hardware.
"This is a silent war over national sovereignty and Taiwan's future," she wrote, "and there is no retreat."
Taiwan's Hardware Comfort Zone Is Running Out of Time
The urgency in Zhang's argument reflects a competitive dynamic that is accelerating beyond Taiwan's borders. Countries and companies embedding AI agents into their core operations now are accumulating productivity advantages that will compound with each passing year. Those that delay are not holding position — they are falling further behind at an increasing rate.
For Taiwan, the path forward requires more than incremental system upgrades or cautious regulatory adjustment. It demands a deliberate decision to build a domestic AI ecosystem — one capable of producing silicon-based intelligence shaped by Taiwanese institutions, Taiwanese industrial expertise, and Taiwanese values. TSMC has demonstrated what Taiwan can build when it commits fully to a technological frontier. Whether Taiwan's policymakers and industries are prepared to make the same commitment to the intelligence layer — rather than only the hardware beneath it — is the question that will define the next chapter. （Related： INFINITIX Signs MOU with Malaysia's SAINS to Advance Sovereign AI Infrastructure ｜ Latest ）
