At the conference, Thomas Kurian, CEO of Google Cloud, highlighted that the average payback period for Google Cloud’s AI servers is under two years. Meanwhile, its self-developed chips achieve payback in just half the time compared to traditional GPU solutions. The business volume of Google Cloud’s self-developed AI accelerators now exceeds twice that of the world’s second-largest hyperscale data center operator. By maintaining full control over its supply chain, Google Cloud has slashed the prices of products like Gemini Flash 3.8 to half or less of the industry average, achieving cost-effectiveness advantages of 2.7 times in AI training, 80% in AI inference, and 30% in CPU operations, respectively. Additionally, the operational costs of Gemini Enterprise—Google Cloud’s enterprise-level AI platform—for machine learning models are just 40% of those of its competitors, while delivering superior accuracy. Currently, over 90% of Fortune 100 companies have chosen Google Cloud as their AI infrastructure solution, with the majority of its cloud infrastructure contracts structured as long-term, five-year agreements.