Why did @Meta an announcement about selling excess computing power cause the AI storage sector to collectively enter a high-level correction? It's because the main narrative of AI technology is facing two real-world tests:

1) The high capital expenditure of giants has entered a phase of cash flow and ROI verification, and the market has begun to price in a slowdown in capex growth, directly putting the brakes on demand for upstream storage/memory;
2) The huge investment in AI models and data centers is facing a test of corporate "ability to pay." In the past, the demand for improving productivity through token consumption tacitly justified the high pricing logic of closed-source models. But if corporate demand-side ROI falls short of expectations, the true "buyer" driving force behind AI large-model investment will be re-evaluated.