Technology stocks, semiconductors, US bond yields, and the probability of the Federal Reserve raising interest rates have all been lowered The market's concerns about AI investment and financing are driving funds to buy US Treasury bonds, and also lowering expectations of future interest rate hikes. On October 23rd during the Asian session, the yield on 10-year US Treasury bonds was around 4.96%, while on Wednesday it was still around 5.13%. The main reason, as I mentioned in yesterday's interview with 168X, is that the capital expenditure for AI is too high, and building data centers and buying chips require a lot of money. Part of this money comes from the company's own operating cash flow, but a considerable portion also relies on external financing such as issuing bonds and borrowing. The interest rate for corporate bond issuance is roughly based on the yield of US Treasury bonds with the same maturity, plus an additional risk compensation. Investors who lend money to companies usually earn more than buying US bonds, and the greater the risk of the company, the more they need to give. Currently, the yield of US Treasury bonds is still high, and companies usually have to pay higher financing costs for issuing new bonds. The amount of money spent is increasing, but the market is concerned that the revenue and cash returns of some AI projects may not keep up with the investment. The investment scale is increasing, and the market's demand for returns will also be higher. If the project returns cannot keep up, the company may slow down in order to retain cash. The bonds issued by some AI companies are already non investment grade, commonly known as "junk bonds". For example, some of CoreWeave's bonds. The yield of US Treasury bonds is already high, and investors' concerns about AI risks are increasing. As a result, some funds are turning to buying US Treasury bonds as a safe haven, pushing up bond prices and corresponding yields. At the same time, the market is also lowering its expectations for future interest rates. If future interest rates are lower than previously expected, the relative attractiveness of existing fixed interest bonds will also increase, so even if the Federal Reserve has not adjusted its policy, bond prices can still rise first. This change has already been reflected in interest rate pricing. Although the market still tends to raise interest rates in December, the implied probability of a rate hike has dropped to about 85%, which was previously fully accounted for in this rate hike. It may seem like a good thing, but in fact, this is where the biggest trouble may arise! The decline in US bond yields reflects changes in capital hedging and interest rate expectations, but does not mean that borrowing for AI companies has become easier. If the returns cannot keep up and financing becomes more expensive and difficult, companies that rely on external financing may slow down the expansion of data centers. In this way, new orders from chip, server, and related equipment manufacturers may be affected, and the market's profit expectations for these suppliers may also be lowered accordingly. Although the decline in US bond yields may provide some support for the valuation of technology stocks, this support may not necessarily offset the impact of lower profit expectations and increased risk compensation. This is the current situation that AI and semiconductors are facing. If the scope of investment slowdown continues to expand, from data center construction to upstream supply chain, investment and recruitment may decrease accordingly, which may truly drag down overall investment and employment growth. Moreover, if oil prices continue to push up inflation, it will be more difficult for the Federal Reserve to quickly alleviate the slowdown in growth through interest rate cuts. The probability of an economic downturn is likely to increase with the decline of the AI industry. One @ Gate, trade more markets