The focus of computing power demand is shifting from training to inference, and the global construction of AI infrastructure is accelerating comprehensively.

Recently, the well-established technology company Oracle shocked the global capital market with a one-day surge of 40%. The reason is that the company's performance data for the first quarter of the 2026 fiscal year showed a 359% year-on-year growth in remaining performance obligations in its cloud infrastructure business, reaching $455 billion, with nearly 60% coming from a 5-year $300 billion inference power order signed with OpenAI. The head of business at a leading domestic cloud computing company told reporters that although there are uncertainties in delivering such a large order from signing to actual delivery, it also sent a clear signal to the entire industry chain about the shortage of computing power, especially inference power, enhancing the confidence of the industry chain in investing in AI and cloud fields.