Goldman Sachs sharply revises upward its forecast for the optical module market, projecting its size could reach nearly $150 billion by 2028.

According to EmberCN’s monitoring, Garrett Jin, the agent of the so-called "BTC OG insider whale", has just closed out a BTC long position worth approximately $107 million, totaling 1,331 BTC. The position was held for around three and a half months, netting a profit of roughly $1.02 million. Currently, his account only holds a large ZEC short position: some 39,760 ZEC with a nominal value of about $47 million, and an unrealized loss of approximately $24 million. Previous data indicates that Garrett Jin’s ZEC short position has been incurring massive unrealized losses for some time. Notably, Garrett Jin added to his heavily loss-making ZEC short position earlier today, before closing out his entire profitable BTC long position. EmberCN quipped that this trading behavior is quite similar to that of ordinary investors: selling profitable positions as soon as possible, while clinging to loss-making positions and continuously adding to them.

According to an announcement from Changxin Memory’s official WeChat account, its independently developed LPDDR6 chip has entered mass production and commercial use, making its debut in the Xiaomi 18 Fold foldable flagship smartphone. This marks the world’s first commercial deployment of LPDDR6 in a flagship mobile device. Earlier, Changxin’s semi-annual report revealed that its LPDDR6 chip delivers a comprehensive performance boost over the previous LPDDR5X generation: through architectural innovation and data transmission optimization technologies, it fully unlocks data transfer potential, with a peak speed of 12800Mbps and a maximum capacity of 16GB per chip. LPDDR6 is the latest international generation of LPDDR products. In July 2025, the JEDEC (Joint Electron Device Engineering Council) first released the LPDDR6 standard, and news of Changxin’s LPDDR6 samples being sent for testing emerged in the industry in March this year. From aligning with standards, sample delivery, verification to mass production and commercialization, Changxin has successfully achieved the world’s first launch of LPDDR6. Against the backdrop of surging AI demand, LPDDR6 is regarded as a key pillar for unlocking edge AI computing power, with applications spanning smartphones, PCs, smart cockpits, AI data centers and more. It also lays a critical foundation for domestic mobile phone and other terminal manufacturers to seize AI opportunities.

According to HTX market data, Bitcoin briefly dipped below $79,000, with a 0.85% decline in the past 24 hours.

According to Bloomberg, Iranian Foreign Ministry spokesman Esmaeil Baghaei said Monday that Iran and Oman have entered the final stages of an agreement on shipping management in the Strait of Hormuz, with a finalization expected in the coming days. The two sides plan to establish a temporary security corridor and submit the relevant arrangements to the International Maritime Organization (IMO) for registration. Baghaei noted that negotiations are progressing "very smoothly" and expressed hope they will not be interfered with by third parties. If the deal is finalized, it could encourage more oil tankers and merchant ships to use this globally critical energy transport route, while reducing the risk of Iran targeting shipping. Currently, around 7 million barrels of crude oil and refined products pass through the Strait of Hormuz daily, far lower than the roughly 20 million barrels per day before the conflict erupted. However, Iran and Oman’s plan to strengthen shipping management in the strait may also signal that Iran’s de facto control over this key route is further consolidated. The U.S., meanwhile, wants the Strait of Hormuz to return to the pre-conflict state of free passage, with clear differences remaining between the two sides’ positions. The ongoing U.S.-Iran conflict has now entered its seventh month, and the blockade of the Strait of Hormuz has continued to push up prices for crude oil, fuel, and natural gas, exacerbating global inflation. At the same time, U.S. diesel prices have hit record highs, and rising energy costs have put greater political pressure on the Trump administration.

According to official data, Trade.XYZ has launched its prediction market trading page, currently offering 26 trading items across three categories: finance, sports, and others. On September 5, Trade.XYZ deployed its HIP-4 DEX on Hyperliquid; transaction records confirm the deployment was completed at 21:15 Beijing Time. Hyperliquid’s HIP-4 is a specialized outcome/prediction market protocol (event result contracts supporting binary/multi-outcome settlements, settled to 0 or 1, no leverage liquidation, etc.), differing from HIP-3 perpetual contracts.

Beating AI News Flash: OpenAI’s latest flagship model Astra has reignited market enthusiasm for AI infrastructure, with memory chips emerging as the first beneficiary sector. U.S. semiconductor stocks rallied collectively last Friday, with Micron surging over 6% and SanDisk jumping nearly 12%; Asian markets extended the gains on Monday, with SK Hynix up around 8%, Samsung Electronics rising more than 5%, and Kioxia climbing roughly 10%. Institutions believe that as the capabilities of cutting-edge AI models continue to improve, computing power demand will further rise, and bottlenecks in memory and networking segments may also intensify, making HBM, DRAM, etc. likely to re-emerge as key themes in AI infrastructure trading. Goldman Sachs and Morgan Stanley project that global AI capital expenditure will reach $1.3 trillion to $1.5 trillion by 2027, with more than half allocated to memory infrastructure. Charu Channa, Chief Investment Strategist at Saxo Bank, stated that Astra’s launch further underpins the sustained growth of AI spending, and the rising computing power demand from advanced models will benefit memory chip makers like SK Hynix and Samsung Electronics. As AI models grow more powerful, the "computing power trade" that solely focused on GPUs is expanding to the "memory trade" involving HBM, DRAM, etc., and memory chips may become a key beneficiary segment in the new round of AI infrastructure investment.