The explosion of investment in AI-powered data centres is creating serious bottlenecks in global supply chains. The warning came from Chey Tae-won, chairman of SK Group – a South Korean conglomerate that includes SK Hynix, one of the leading suppliers of key memory chips.
“I believe that these rapid changes … ultimately lead to bottlenecks around the world,” – Chey said at a business event accompanying the APEC summit in Gyeongju. The demand to build new infrastructure is so rapid that component suppliers cannot keep up. “For everything [that goes into data centres], from chips to services, I think they are creating bottlenecks,” he added.
The SK Group CEO’s comments are of particular importance. SK Hynix is a key manufacturer of advanced HBM (High Bandwidth Memory), essential for AI accelerators such as Nvidia’s GPUs. The growing demand for these specialised components already exceeds the industry’s production capacity, creating real constraints on the development of large language models and cloud services.
The problem is exacerbated by intense global competition. Chey pointed out that the race for leadership in AI has become a matter of national importance, with powers such as the US and China launching national strategies to gain an advantage. This rivalry puts additional strain on supply chains and complicates resource allocation in the global semiconductor industry. This situation puts pressure on the entire technology industry, which has to balance the rapid development of AI with real production constraints.
