智通财经网·2026-01-30 00:49Core Viewpoint - Recent demand for inference and training computing power is strong, leading to price increases from both Amazon Web Services (AWS) and Google Cloud [1][2] Group 1: Demand for Computing Power - The demand for computing resources for inference and training has significantly increased, supporting the need for training computing power [1] - AWS raised prices by approximately 15% for EC2 machine learning capacity blocks on January 23, 2026, followed by Google Cloud announcing price hikes for its cloud network transmission services, with North America seeing a doubling of prices [2] Group 2: Inference Side - The rapid emergence of AI agents, such as MoltBot, is expected to support the demand for inference computing power, as these agents can perform more complex tasks and require more computing resources [3] - Anthropic has raised its revenue expectations for 2026 and 2027 to $18 billion and $55 billion, respectively, indicating a strong upward trend in inference demand [3] Group 3: Training Side - The industrial sector is continuously exploring scaling limits, which supports the demand for training computing power, with new models expected to be released in Q1 2026 [8] - Models like Grok-5 and GPT-6 are anticipated to utilize larger datasets and parameter scales, increasing the demand for training computing power [8] Group 4: Financial Reporting Catalyst - The upcoming earnings reports from major cloud service providers (CSPs) will be critical in confirming the demand for computing power and capital expenditure continuity [9] - Key earnings dates include Microsoft and Meta on January 29, 2026, followed by Google on February 5 and Amazon on February 6, with NVIDIA's report on February 26 expected to further influence market sentiment [9][10]