GigaDevice reports first-half attributable net profit of RMB 6.857 billion, up 1091.5% year-on-year.

According to Bank of America's August Global Fund Manager Survey, 32% of investors identify the AI bubble as the top tail risk event. Tail risks refer to extreme, rare events that could significantly impact financial markets. The survey shows 27% of respondents view a disorderly rise in bond yields as the second-largest tail risk. Meanwhile, investors' confidence in U.S. equities outperforming other regional stock markets has further strengthened. The survey indicates 27% of participating fund managers are over allocating to U.S. stocks relative to the global benchmark weight, up 3 percentage points from July and hitting the highest level since December 2024. Year-to-date, U.S. equities as a whole have outperformed most overseas markets: the S&P 500 index has gained 13% cumulatively, while the Euro Stoxx 600 index has risen 10%. However, Japanese and South Korean stock markets have still performed better than U.S. equities over the same period.

According to BIT (Bit.com) market data, major US stock index futures are trading lower ahead of the US market open: the Nasdaq fell 0.32%, the S&P 500 dropped 0.52%, and the Dow Jones declined 0.51%. Storage and optical communication sectors saw broad losses: SanDisk (SNDK) down 4.19%, Seagate Technology (STX) down 5.11%, Western Digital (WDC) down 5.23%, Micron Technology (MU) down 4.13%, SK Hynix (SKHY) down 4.39%. Corning (GLW) down 4.55%, Coherent (COHR) down 5.76%, Marvell Technology (MRVL) down 5.30%, Lumentum Holdings (LITE) down 5.98%, Nokia (NOK) down 3.43%. On the news front, Anthropic’s annual revenue fell short of market expectations, sparking questions about AI’s growth trajectory. Bloomberg previously reported that Anthropic’s annualized revenue run rate (ARR) reached approximately $65 billion as of the end of July. Against a backdrop where some third-party data and AI industry optimistic forecasts had pointed to over $80 billion, the $65 billion figure signals a slowing growth pace. The controversy centers on the definition of ARR, or annual recurring revenue, which essentially annualizes current revenue velocity and is not equivalent to audited full-year revenue. Sacra data shows Anthropic’s annualized revenue stood at roughly $47 billion in May, rising to $65 billion in July, but it also noted that revenue from cloud channels including AWS, Google, and Microsoft may be recognized on a gross basis, which would inflate the revenue figure and further draw market focus to gross margins and revenue quality.

The benchmark emerging market stock index ended its four-day rally, while the emerging market currency index also retreated from its all-time high. A global bond sell-off and rising oil prices weighed on market risk sentiment. The MSCI Emerging Markets Index fell 0.7%, mainly dragged down by declines in Samsung Electronics and TSMC shares. The emerging market currency index dropped 0.1%, as the U.S. dollar strengthened for the first time in four days. Chris Turner, global head of markets at ING, said a massive sell-off in the U.S. long-term bond market is typically bad news for emerging market currencies and overall market risk. (Jin10)

Nvidia is leveraging its strong cash flow and financing capabilities to extend the AI competition from the chip sector to the capital markets. Over the past week, Nvidia first struck deals with Wall Street institutions including Goldman Sachs, Blackstone, BlackRock, and Apollo to push for up to $500 billion in GPU financing; it then announced it would provide up to $105 billion in support for OpenAI’s Ohio data center project. Nvidia CEO Jensen Huang said many cutting-edge AI labs are growing faster than their balance sheets and long-term credit capacity, requiring external capital to support AI infrastructure development. Nvidia also plans to provide up to 25% backstop support for related loans, and will continue to position itself in the AI industrial chain via equity investments. As Google’s TPU and AMD catch up rapidly, Nvidia’s chip technology advantage is facing challenges. Analysts believe Nvidia is expanding its ecosystem control via a "chip + capital" model, further consolidating its market leadership while driving AI infrastructure investment.

According to a Bloomberg report, sources familiar with the matter disclosed that LimX Dynamics, a robotics manufacturer backed by Alibaba, has secretly submitted an application for an IPO in Hong Kong, with plans to raise up to $300 million. The wave of listings for Chinese robotics companies is expanding once again. LimX Dynamics is collaborating with CITIC Securities to advance the transaction. Headquartered in Shenzhen, China, LimX Dynamics is an embodied intelligence robotics firm founded in 2022, focusing on the R&D, manufacturing and commercialization of general-purpose robots. Its core products include full-size humanoid robots LimX Oli and LimX Luna, as well as modular robots of the TRON series, and it has independently developed the humanoid brain system LimX COSA. Centered on three core technologies—hardware design and manufacturing, integration of cognitive and motor intelligence, and embodied intelligence operating system—the company is committed to advancing the deployment of physical-world AGI, with its products already applied in fields such as scientific research, commercial services and industrial inspection. The firm has secured strategic investments from Alibaba, JD.com and other investors, and recently completed multiple rounds of financing.

Citrini analyst Jukan noted that MLCC (multi-layer ceramic capacitor) delivery cycles are experiencing severe divergence based on product specifications. Lead times for general-purpose products remain stable at 14 to 18 weeks (roughly 3.5 to 4.5 months), while high-capacitance and high-voltage products take about 4 to 5 months. However, high-end products for AI servers have entered a state of "severe supply tightness". Server-grade MLCCs from Murata and Samsung Electro-Mechanics have delivery times exceeding 20 weeks (over 5 months). Samsung Electro-Mechanics’ average lead time for some high-capacitance products is around 40 weeks, while Murata’s high-capacitance products of 1 microfarad and above saw their July delivery times rise to 30 weeks from 24 weeks in June. Some channel quotations even reach 36 weeks, nearly 10 months. This extreme divergence is expected to persist until the second half of 2026. The core driver of this gap is robust demand from AI servers and data centers: AI servers require 5 to 13 times more high-capacitance, miniaturized, high-reliability MLCCs than traditional servers. New capacity expansion has been delayed from Q4 2026 to 2027, so the supply-demand imbalance is unlikely to ease in the short term. Major manufacturers have simultaneously raised prices and expanded long-term supply agreements. Delivery times exceeding 5 months, even approaching a year, not only signal that manufacturers cannot rapidly boost supply in the short term, but also mark the start of enhanced pricing power and substantial improvements in profitability driven by high-value product portfolios. The AI bottleneck is spreading from GPUs and memory to the smallest components on circuit boards.