Taiwan’s competitive advantage in the AI landscape

Citing examples, Tung pointed out that training advanced models like OpenAI’s GPT-4 costs approximately US$78 million, while Google’s Gemini Ultra requires about US$190 million. The escalating need for electricity and computing power, he noted, has led industry leaders like OpenAI’s Sam Altman to invest in nuclear energy solutions, emphasizing the connection between energy supply and AI innovation. Tung advocates for maintaining nuclear power in Taiwan to support this growing demand.

Taiwan is well-positioned to leverage its strengths in hardware manufacturing, according to Pegatron Co. Chairman Tung Tzu-hsien. Speaking at the 2024 Taiwan AI Academy Annual Conference in Taipei, Tung emphasized that while established tech enterprises and countries may consolidate their dominance, Taiwan’s capabilities provide a unique competitive edge.

Tung compared the rise of AI to the internet boom over the past three decades, noting that AI will increasingly integrate into everyday life. However, he cautioned that unlike the internet era, which allowed for significant market disruptions, AI may reinforce the power of those already at the top. The primary barrier to entry, he argued, lies in the substantial financial and infrastructural resources required for AI development.

Citing examples, Tung pointed out that training advanced models like OpenAI’s GPT-4 costs approximately US$78 million, while Google’s Gemini Ultra requires about US$190 million. The escalating need for electricity and computing power, he noted, has led industry leaders like OpenAI’s Sam Altman to invest in nuclear energy solutions, emphasizing the connection between energy supply and AI innovation. Tung advocates for maintaining nuclear power in Taiwan to support this growing demand.

He also highlighted the strategic retreat of tech giants like Apple, which recently shifted its focus from developing in-house AI infrastructure to partnering with OpenAI for integrating AI capabilities into its products. This move reflects a broader trend where companies recognize that having a strong market presence can be more beneficial than developing technology from scratch.

Taiwan’s semiconductor industry stands to gain significantly from the AI boom, with the overall output value expected to reach NT$5.3 trillion (about US$167.2 billion) this year, surpassing earlier forecasts of NT$4.8 trillion. This growth is largely attributed to increased demand for AI-enabled devices, which will enhance everything from cloud computing infrastructure to personal electronics.

Despite Taiwan lagging in electric vehicle (EV) adoption, with penetration rates below 10%, Tung believes that AI’s integration into various sectors, including EVs, presents an opportunity for Taiwan to catch up. He noted that the foundational technologies for modern AI advancements were made possible by Taiwan Semiconductor Manufacturing Co. (TSMC), founded by industry pioneer Morris Chang.

As Taiwan navigates the AI landscape, its strong semiconductor and hardware capabilities may prove critical in maintaining its competitive edge, reinforcing Tung’s assertion that Taiwan remains a key player in the global AI arena.