Australian Associated Ports Union: BHP Billiton employees and elected representatives held five hours of wage negotiations on Tuesday. BHP Billiton labor talks failed to reach agreement. Further talks with BHP are scheduled to take place on July 21. BHP Billiton workers plan to continue their strike on July 16.

Based on the views of many analysts, an inconspicuous piece of news a few days ago may become a watershed in the AI ​​hardware landscape. Meta plans to mass-produce a self-developed chip called "Iris" in September. An internal memo shows that the chip was tested in only six weeks and no major problems were found. This speed is almost miraculous in the chip industry. Meta’s goal is straightforward—to reduce dependence on Nvidia GPUs. So the logic behind this is worth remembering. When a company's core cost items are monopolized by a single supplier, self-research is always just a matter of time, not whether to do it. Meta is not the first, It won't be the last either. And Meta did something else it had never done before - charged for the first time it released an AI model. Its inference model Muse Spark 1.1 scored 51 in the artificial intelligence analysis index, an increase of 8 points in three months, ahead of Gemini 3.5 Flash's 50 points. The pricing is $1.25 for input and $4.25 for output per million tokens, which is about a quarter of the price of Claude Opus 4.8 and GPT 5.5. Taken together, These two moves may represent a strategic shift for Meta rather than a show of skill. Meta has made the Llama series free for many years, but now it has begun to charge for one of the most cost-effective orchestration models on the market. At the same time, it uses self-developed chips to vertically integrate and significantly reduce inference costs. Its goal is not to top the review list, but to take the initiative in pricing.

The yield on German two-year government bonds rose to 2.7854%, a new high since July 2024, rising 6.5 basis points on the day.

Government of India: The Wholesale Price Index (WPI) rose by 9.87% year-on-year in June (Reuters survey expected 9.15%).

Government of India: The fuel price index in the wholesale price index increased by 27.41% year-on-year in June.

There is an obvious statistical lag effect in CPI in June. If the nominal CPI falls from 4.2% in May to around 3.8%, it mainly reflects the phased cooling of energy prices in June, but does not mean that overall inflationary pressure has subsided. The Cleveland Fed model predicts that CPI may fall back to around 3.9% in June, but core CPI is still close to 2.9%, indicating that price pressures are still relatively resilient. What the market is really concerned about is changes in the inflation structure. A fall in energy prices may lead to an improvement in nominal inflation, but housing, service prices and wage costs are still supporting core inflation. A survey by the New York Fed showed that One-year and three-year inflation expectations continue to rise, which means that even if June CPI data improves, the Fed still needs to observe whether this cooling is sustainable. Therefore, the key to the impact of this CPI lies in how the market interprets this fall. If core inflation remains high, or energy prices strengthen again in July, trading logic may shift back to pricing secondary inflation risks. In the context of the widening policy differences within the Federal Reserve and the weakening of forward guidance, the CPI is not only an inflation data, but also an important signal affecting the path of interest rates and asset pricing.