Book: Google's TPU Development, Microsoft's SMR Investment—Why? Looking Ahead at the Next Decade's 'AI Power War'

The AI era competition is often metaphorically described as a war, with talent wars, model wars, and capital wars being common themes. This is because we have already deeply engaged in the technological frontlines surrounding AI. As technology rapidly evolves, the competitive advantages gained often transition into the next industrial hegemony. Talent, models, and capital are individual elements crucial for maintaining superiority. In 'AI Power War,' author Lee Jae-hoon argues that the speed of implementing AI technology in reality is determined by power and infrastructure. The difficulty in keeping up with power supply growth amidst technological evolution forms the background. As technology evolves, energy demand increases, yet expanding power transmission networks or power plants remains challenging. When building AI data centers or semiconductor factories, integrated and industrialized designs considering energy infrastructure are increasingly necessary. A weakness in any single element can cause bottlenecks in the technological competition landscape. The book diagnoses Korea's reality as a complex bottleneck shaped by interconnected factors: power generation, transmission, pricing, power source transition, and social consensus. It argues that the rapid scalability of power system expansion must now be linked to national industrial competitiveness, requiring a structure that can swiftly connect, stably operate, and fairly distribute power. The actions of tech giants prioritizing AI growth, such as Google, Microsoft, OpenAI, NVIDIA, and Tesla, support this argument. The book highlights how power is no longer just a purchase item but a core infrastructure strategy for AI industries, using cases like Google's long-term power purchase agreements (PPAs) securing renewable energy capacity globally, and Microsoft's SMR investments positioning nuclear energy as a long-term power portfolio.