EUR/USD
1.14145
0.007%
Gold
4024.81
0.433%
Oil
82.088
-0.209%
USD/JPY
162.505
0.013%
GBP/USD
1.34358
0.042%
GBP/JPY
218.343
0.058%
Real-time News
July 21 – Capital Economics strategist Thomas Mathews stated in a report that gold remains valuable as a hedge against a stock market crash. Recently, golds performance has arguably been more of a "risk" asset than a "safe" one, given its recent price volatility, which has mirrored that of the benchmark S&P 500. Golds poor performance amidst Middle East conflicts appears to have weakened its supposed inflation-hedging role. However, the link between gold and real bond yields appears to remain intact, and any decline in real yields is expected to boost gold. If the US economy is hit and prompts the Federal Reserve to significantly cut policy rates, the recent positive correlation between gold and the stock market will not be sustained.On July 21, it was reported that the highest 7-day annualized yield of Tencent Wealth Managements "Current Account +" was 1.4500%, and the lowest was 0.7710%. The highest 7-day annualized yield of WeChat Pays "Lingqian Tong" was 1.0520%, and the lowest was 0.9220%. The highest 7-day annualized yield of Alipays "Yuebao" was 1.0670%, and the lowest was 0.9860%.The ChiNext index rose by more than 1%, with memory chips, semiconductors, and gaming sectors leading the gains.Newly listed stock N Tainuo-U opened 246.82% higher; the prospectus shows that the company is dedicated to blood product replacement therapy.On July 21st, Z.ai announced the completion of a 1GW-level domestically produced AI computing power data center, utilizing entirely domestically produced AI chips. Simultaneously, Z.ai also officially completed its acquisition of XCore Sigma, a domestic AI heterogeneous computing power software company. XCore Sigma originated from the Compiler Laboratory of the Institute of Computing Technology, Chinese Academy of Sciences, and has long been deeply involved in heterogeneous computing power software stacks and compiler optimization, considered one of the top AI infrastructure teams in China. Sources indicate that these two moves respectively fill two key gaps in computing power supply and release capabilities. The former provides the computing resources needed for large-scale model training, while the latter improves the utilization rate of heterogeneous chips, reduces inference costs, and enhances model deployment efficiency through basic software capabilities such as compilers, runtimes, and inference engines.
