The dollar witnessed sharp fluctuations on Tuesday, as it fell after recording its highest level in about two months, following the decline in oil prices following an Iranian proposal to reopen the Strait of Hormuz within 7 days if the United States eases its military pressure.
The US currency fell 0.12 percent against the yen to 157.15, while the euro stabilized without significant change at $1.146. The British pound fell 0.09 percent to $1.3356, according to Reuters.
These fluctuations highlight the interconnected factors driving turmoil in foreign exchange markets, at a time when the outlook for global interest rates has been further complicated by the prolonged conflict in the Middle East.
The turmoil in the Strait of Hormuz has put pressure on global oil supplies and increased inflationary pressures, prompting investors to quickly reprice assets at any sign of changing expectations.
A senior Iranian official told Reuters, following a previous report by Kyodo News, that Iran could reopen the strait within 7 days if the United States eases military pressure and lifts the blockade imposed on Iranian ports.
Brent crude futures fell below $99. Any sustained decline in oil prices would ease inflationary pressures and reduce the need to raise US interest rates, which could reduce the dollar's yield advantage over other currencies.
The dollar index, which measures the performance of the US currency against 6 major currencies, recorded a decline of 0.02 percent to 100.4, after touching earlier in the session its highest level since late July.
Factors putting pressure on the yen
Traders were also evaluating whether the Bank of Japan would raise interest rates at a fast enough pace to close the gap with other global central banks.
Although the Bank of Japan raised interest rates last week, the Japanese currency remained under pressure, as investors view the opposition expressed by two policymakers, who called for a more cautious pace, as an indication that implementing additional increases may be difficult.
Naomi Fink, chief global strategist at Amova Asset Management, said: “The central bank’s acceleration of its pace depends on how events develop, and the issue of determining the neutral and final levels of interest rates is still a matter of debate.”
Tighter statements from other global central banks have raised questions among traders about the yen's yield advantage, although the possibility of Tokyo's intervention has limited the dominance of bearish expectations.
Last week, European Central Bank President Christine Lagarde pushed back against investors' bets on sharp increases in interest rates, suggesting that a measured response from the central bank would be enough to contain inflation.
“European Central Bank officials have so far maintained a noticeably hawkish tone, leaving the option of an October rate hike firmly on the table,” wrote Francesco Pesoli, foreign exchange strategist at ING Bank.
He added: “However, investors seem more inclined to adopt the opposite scenario, which indicates more downward pressure on the euro-dollar pair in the near term.”
Markets are pricing in a roughly 30 percent chance that the Bank of Japan will raise its benchmark short-term interest rate to 1.5 percent in October, and a roughly 50 percent chance that the Federal Reserve will raise the federal funds rate range by 25 basis points to about 4 percent.
Alibaba Group intensified its bet on artificial intelligence by announcing the development of a new model that may reach four times the size of its current main model, in addition to revealing a new artificial intelligence chip that it described as the most powerful in China, in a move that pushed the company’s stock in Hong Kong to rise by more than 5 percent.
The announcements come at a time when Chinese technology companies are racing to develop local alternatives to NVIDIA chips, amid tightening US restrictions on exporting advanced processors to China, which makes the ability to build an integrated local artificial intelligence system a strategic issue for Chinese companies.
During the annual Apsara cloud computing conference in Hangzhou, Alibaba revealed plans extending from basic models and chips to data centers, an indication of its desire to control the largest number of infrastructure layers needed to train and operate artificial intelligence systems.
CEO Eddie Wu said that the group’s “Coin” team plans to train a new model with a size ranging between 5 and 10 trillion pieces of information. With the aim of dealing with more complex tasks that take longer, as part of the company’s endeavor to develop superior artificial intelligence that exceeds human capabilities.
The “Coin 3.8 Max” model, which is the company’s current main model, contains about 2.4 trillion pieces of information. Alibaba explained that it is currently working on training the next generation, “Coin 4,” while the sizes of the “Coin 4.5” and “Coin 5” models are expected to rise in the future to a range between 5 and 10 trillion pieces of information.
Wu noted progress in the ability of models to improve their performance on their own by discovering weaknesses, conducting experiments, and producing training data with limited human involvement.
In parallel with the development of models, the group unveiled the new “Chenwoo V900” chip for artificial intelligence, developed by the T-Head Semiconductor Unit.
Wu said that the performance of the new processor is three times that of the previous generation, the M890, which the company launched last May.
Up to 500,000 of these chips can be connected into huge computing clusters to train and run the largest artificial intelligence models. Large-scale commercial production of the chip is scheduled to begin during the first quarter of 2027, with the company expecting to achieve “significant growth” in annual shipments of artificial intelligence processors.
This step reflects Alibaba's attempt to reduce its dependence on foreign technology in one of the most important components of artificial intelligence infrastructure, especially with the restrictions imposed by Washington on Chinese companies' access to some advanced American chips.
Markets welcomed the announcements, as Alibaba's Hong Kong-listed shares jumped 5.1 percent, reaching their highest levels in a month.
Spending is not limited to models and chips. Wu has set a goal to raise the total capacity of Alibaba Cloud's global data centers to more than 20 gigawatts by 2032, an indication of the enormous amount of infrastructure that the group expects to need in the coming years.
He said customer demand for AI is “exceptionally strong” and is driving cloud computing revenue growth to accelerate, but supply chain constraints still limit the speed of expansion.
He added that the expected demand in the medium and long term greatly exceeds the company’s current ability to provide computing power, noting that “Alibaba Cloud” will begin, during the current quarter, operating supercomputing nodes dedicated to artificial intelligence on a commercial scale.
This expansion outlines a strategic plan that goes beyond the competition to produce a more advanced artificial intelligence model. At the same time, Alibaba is betting on the model, chip, data center, and cloud computing, giving it greater ability to control development and expansion costs.
For China, this trend carries additional importance as US restrictions become an incentive to accelerate the construction of a local technology ecosystem. With the rise in demand for computing power, the success of companies such as Alibaba in producing chips capable of running large models will become an important factor in determining the speed of progress of the Chinese artificial intelligence industry in the coming years.
