Asia’s AI strategy confronts reality as Iran war drives up energy prices and disrupts supply chains

(SeaPRwire) –   The surge in global AI has strengthened economies throughout Asia, benefiting Korean chip manufacturers, data center firms in Southeast Asia, Chinese AI ventures, and Japanese component producers.

Even the most severe Middle East conflict in decades has failed to slow this momentum. Recently, Microsoft pledged $5.5 billion for cloud and AI infrastructure in Singapore, alongside another $1 billion for Thailand in the coming years.

However, the war in Iran might eventually compel Asia to rethink its AI strategy, driven by rising energy costs and a lack of essential materials required for constructing AI infrastructure.

“The scaling laws fueling the AI boom are essentially peacetime concepts,” explains Wei Lu, a professor at Nanyang Technological University (NTU) in Singapore. “They were identified during a time of plentiful energy and growing chip supplies, relying on the implicit assumption that energy elasticity is limitless.” This has resulted in what Lu calls a “brute force aesthetic,” where increasingly powerful models are built despite the rising energy cost per unit of computation.

This approach is sustainable during prosperous times but becomes problematic when resources are scarce. “The ongoing conflict is repricing that wager,” Lu states.

Asia’s AI boom

Asia has emerged as the epicenter of the worldwide AI expansion. Nomura estimates the region accounted for almost two-thirds of global AI trade growth during the first half of 2025.

Various areas have specialized in distinct segments of the AI industry. East Asian nations such as South Korea and Taiwan have profited significantly from semiconductor manufacturing, supporting the AI capital expenditure surge in markets like the U.S. Meanwhile, Southeast Asia has seen investment concentrated on assembly, precision manufacturing, and data storage.

However, experts caution that AI operations in the region could become significantly more expensive as oil, LNG, and helium prices soar following the Iran war.

“The primary consequence for Asia’s AI boom will be increased expenses in developing AI infrastructure,” says Bo An, a computer science professor at NTU. “Chip manufacturers might encounter steeper costs for energy, raw materials, shipping, and insurance, while data center operators could see rising power and cooling expenses.”

He also forecasts that the elevated costs and supply interruptions in Asia will inevitably affect technology companies globally, considering the region’s pivotal position in the international chip supply chain.

For instance, TSMC is the primary provider of advanced chips to industry leaders like Nvidia and Apple. However, TSMC’s home base in Taiwan depends heavily on imported energy, potentially creating a dilemma for the island’s government should the Iran crisis persist. Oxford Economics projects that Taiwan’s industrial production could drop 0.7% below baseline levels if shortages last for six months.

“We are already witnessing panic buying and logistical gridlock,” says NTU’s Lu, observing that the global supply chain has effectively become “a series of single points of failure.”

Efficiency-first design

In the near term, the AI sector is robust enough to withstand concerns regarding the Iran conflict. South Korean chip exports reached a historic peak of $32.8 billion in March, surging over 150% year-on-year, per government data issued on April 1.

“We do not anticipate that the energy shock will significantly derail South Korea’s AI-driven growth path this year, especially since the current semiconductor cycle seems more robust than previously expected,” Bank of America analysts wrote in an April 2 research report.

There could even be long-term benefits for Asia. Iran’s attacks on data centers in the Middle East have underscored that server racks are now potential military targets.

Following substantial investment in the Middle East, “AI companies are beginning to turn their attention to Southeast Asia and India,” according to Sandeep Sethi, who manages the APAC data center division for real estate firm JLL.

However, in East Asia, data center operators may confront the long-term issue of restricted power availability. This is particularly true in locations like Japan, where connecting a new data center to the grid can take up to a decade.

Lu contends that AI enterprises must begin adopting an “efficiency-first” design philosophy, minimizing the energy and raw materials required to cultivate artificial intelligence.

“The most valuable type of intelligence is the one that understands how to achieve more with less.”