This article is by Lee Jae-lim and read by an artificial voice.

Semiconductor production capacity will remain a more decisive driver of profitability than technical prowess amid the ongoing AI chip supercycle, DigiTimes founder and CEO Colley Hwang said on Monday.
The technology media outlet's founder said that chipmakers that can secure capacity and supply chain control will be the key winners in the era of artificial intelligence, arguing that the current supercycle represents a structural shift rather than a temporary surge.
"Take Nvidia, for instance," Hwang said at a seminar on AI and semiconductors hosted by People Power Party Rep. Song Seok-jun, held at the National Assembly Library in western Seoul. "Revenue was $7.19 billion in the first quarter of fiscal 2024, which grew to $46.7 billion in the second quarter of fiscal 2026. Within just two years, its scale expanded roughly six to seven times. In this kind of environment, the ability to deliver becomes more important than technology alone. Technology maturity matters, but timing and capacity have become decisive."
The global semiconductor market is valued at about $628 billion in 2024, according to Markets and Markets, and is projected to reach $707 billion in 2025. Memory accounts for about 27 percent of the market and non-memory for 73 percent. On the demand side, computing devices make up 39 percent and telecommunications 30 percent, followed by consumer electronics, industrial applications, defense and automotive.
Hwang warned that countries and companies lacking the resources to scale manufacturing - whether in chips, power systems, data centers or supporting energy infrastructure - will fall further behind as capital requirements grow.
"So, for companies and for countries, you must become No. 1 or No. 2 in your field," Hwang said. "Anything below that will be difficult to sustain. Taiwan and Korea are in unique positions because we are among the very few with the manufacturing depth and supply chain capability to serve the global market."
Hwang further noted that the next bottleneck will not be chip output alone, but power capacity, as data center expansion struggles to keep pace with AI-related demand.
Responding to these concerns, Datacrunch Global CEO Sung Soo Eric Kim pointed to rapid growth expected in small modular reactors (SMRs), referencing Bill Gates' August visit to Seoul as supporting evidence for broader global discussions on energy-linked AI infrastructure development. It was Gates' first visit to Korea in three years, during which he met with President Lee Jae Myung, government officials and business leaders to discuss collaboration on nuclear energy, biotechnology and AI.

"More than a decade ago, Bill Gates told the Korea Electric Power Corporation board that in the future, tech companies would play a major role in nuclear power development," Kim noted. "That time has now come. SMRs are becoming closely tied to AI and semiconductor growth because large-scale AI data centers require tremendous amounts of stable, carbon-free power."
He said Korea's strategy should recognize that demand for large-scale AI data centers is centered in the United States, the United Arab Emirates and Saudi Arabia, while Korea, Taiwan and Japan hold the supply-side strengths. The key challenge, he added, is shaping an effective trilateral cooperation model to meet this accelerating global demand.

Ver todos los episodios