According to recent Forbes statistics, developers and cloud computing companies have issued approximately $80 billion in bonds related to US data centers this year, of which about $37 billion are junk-rated, accounting for about 46% by amount. This proportion is projected to be around 24% for the whole of 2025, nearly doubling this year. Junk-rated bonds, also known as high-yield bonds, indicate a higher risk of default. These companies need substantial capital to build data centers, and borrowers with lower credit ratings typically need to pay higher interest rates to attract investors. I specifically addressed this issue in an interview with 168X. The main reason for the current decline in AI and semiconductor stocks is excessively high capital expenditures. In particular, the monetization potential of AI is not very transparent. Providing equipment for AI is profitable, as seen with companies like Nvidia or semiconductor companies, but what about the profitability of AI itself? Currently, it's insufficient to support such large capital expenditures. Furthermore, the monetization potential of AI is not very transparent, and the money earned may not keep up with the investment. Companies that provide equipment for AI are profitable, like Nvidia, whose profits are already reflected in their financial statements. However, many companies that buy equipment, rent computing power, and provide models and applications are not yet generating enough revenue from AI to keep pace with their investments, making it insufficient to support continued procurement through their own profits. Companies that build their own data centers have to pay upfront for chips, servers, and server room construction. The equipment depreciates over time, becoming part of the cost, while electricity, daily operations, and loan interest are continuously paid. If the cash flow from AI business doesn't keep up, and further expansion is still needed, companies reliant on external financing will typically have to issue bonds or take out loans. For borrowers with already low credit ratings, if investors become increasingly concerned about their repayment ability, new loans and renewals may become more expensive and harder to obtain. At that point, these companies may postpone or scale back some expansion plans, and equipment procurement may slow down. Companies that provide equipment for AI currently derive their profits from completed sales. If buyers slow down their purchases due to increased financing costs, the first to be affected will likely be market expectations for these companies' future order and profit growth. A @Gate, trading more markets