According to CoinMeta, a representative of Amazon's cloud business stated that customers are beginning to shift from using the company's services to train AI models to integrating these models into their own business processes, which is driving an increase in the computational demands for inference. Matt Garman, the CEO of Amazon's cloud computing division, said, "We still see some companies using large training clusters, but as these models become more popular and more powerful, more and more companies are integrating this inferential capability into their workloads." He noted that the potential of the artificial intelligence business is "very substantial," and the company will continue to increase its capital expenditures to meet the growing demand. As the world's largest provider of computing power and data service leasing, Amazon announced last week that it expects its capital expenditures to reach $220 billion in 2026, higher than the previously predicted $200 billion. The increase in expenditures reflects the rising prices of storage chips and other components required for data centers.