Chinese A-shares experienced narrow-range fluctuations today, with the three major indices posting mixed results. The Shanghai Composite Index closed up 0.11%, while the Shenzhen Component Index fell 0.09% and the ChiNext Index declined 0.27%. The combined turnover for the two markets reached 2.76 trillion yuan, an increase of 150.3 billion yuan from the previous trading day. The semiconductor sector stood out as the top performer, with particular strength seen in chip design.
The semiconductor sector surged, led by chip design companies. Boosted by better-than-expected new orders reported by a major manufacturer yesterday, domestic AI companies are demonstrating a strong turnaround with high profit growth. Today, the STAR Market Chip Design ETF (589260) and the STAR Market Chip ETF (589100) both rose over 6%. The STAR Market Artificial Intelligence ETF (589110) and the Integrated Circuit ETF (159546) gained over 5%, while the Chip ETF (512760) advanced more than 4%.
On April 24, 2026, the long-awaited DeepSeek V4 model was officially launched. The DeepSeek-V4-Pro demonstrates several notable strengths. Firstly, its Agent capabilities have significantly improved, achieving top-tier performance among current open-source models in Agentic Coding evaluations and excelling in other Agent-related benchmarks. Secondly, regarding world knowledge, DeepSeek-V4-Pro substantially outperforms other open-source models, trailing only the leading closed-source model, Gemini-Pro-3.1. It also exhibits world-class reasoning performance, surpassing all publicly evaluated open-source models in mathematics, STEM, and competitive coding tests, achieving results comparable to top-tier closed-source models.
A relevant question arises: why has the market ultimately favored the semiconductor sector? The answer is straightforward. Amid previous US restrictions, domestic computing power supply became extremely scarce. Constrained by computing resources, the development pace of domestic AI also lagged behind international counterparts. However, China possesses distinct advantages: abundant and inexpensive electricity resources. As long as domestic GPUs can be adapted to support domestic large language models at a usable level, China can compete on scale. If technology lags, competing through volume becomes a viable strategy – but this depends on meeting basic performance requirements. Consequently, the adaptation of DeepSeek V4 to domestic GPUs holds particular significance for the semiconductor sector.
This logic is not merely speculative. As the first-quarter earnings season nears its conclusion, numerous companies have recently released their reports. Financial data from certain industry leaders has indeed exceeded market expectations. For instance, VeriSilicon disclosed that as of April 20, its new order value for the year had reached 4.516 billion yuan. By April 29, this figure had further increased to 8.240 billion yuan, demonstrating robust demand. Additionally, Cambricon released its financial report, showing a net profit attributable to shareholders of 1.013 billion yuan, a year-on-year increase of 185.04%.
Looking ahead, given the significant gap between supply and demand for domestic computing power, the adaptation of DeepSeek V4 to domestic chips marks a new milestone for domestic AI chips. A market potential measuring hundreds of billions of yuan may gradually materialize in the financial reports of related companies. Interested investors may continue to monitor ETFs such as the STAR Market Chip Design ETF (589260), the STAR Market Chip ETF (589100), and the Chip ETF (512760).
In North America, positive news emerged frequently. The Communication ETF opened lower but recovered somewhat during the session, ultimately closing down 0.53% for the day.
Yesterday, the four major North American cloud providers (Google, Microsoft, Amazon, Meta) held their earnings calls, with Google significantly surpassing market expectations. Google revised its 2026 capital expenditure upwards by $10 billion to a median of $185 billion, and its order backlog doubled quarter-over-quarter to $460 billion. Meanwhile, reports for Microsoft, AWS, and Meta were mixed but revealed no major risks. The key takeaways are: (1) capital expenditure remains on an upward trajectory; (2) the order backlog provides support for future capital expenditure; (3) growth in earnings demonstrates that the hardware and software cycle for AI is becoming more mature.
In the latest Federal Reserve meeting, the target range for the federal funds rate was maintained at 3.5%-3.75%. Subsequent policy adjustments are expected to be cautious. Fed Governor Michelle Bowman dissented, as she has consistently, favoring a 25 basis point rate cut at this meeting. Furthermore, three regional Fed presidents, while supporting holding rates steady, objected to the inclusion of language suggesting a loosening bias in the policy statement. External reports indicate that four officials expressed dissent on the statement, the first such occurrence since October 1992, signaling growing internal divisions at the Fed. The Fed's statement noted that inflation remains elevated, partly driven by recent increases in global energy prices, and that developments in the Middle East have heightened uncertainty surrounding the economic outlook. In his final speech as Chair, Jerome Powell stated he would remain on the Fed's Board of Governors after his term ends and would not become a "shadow chair." Although the path for Fed rate cuts involves some uncertainty, the decision to hold rates steady had no significant impact on capital markets, as it was largely anticipated.
Regarding the Taiwan stock market, AI-related stocks have accelerated comprehensively since March. Taiwan Semiconductor Manufacturing Company (TSMC) reported March revenue of NT$415.2 billion, a year-on-year increase of 45%. For the full first quarter of 2026, revenue reached NT$803.7 billion, a monthly increase of 34.90% and a yearly increase of 45.57% (approximately 34.1% monthly and 51.5% yearly increase in US dollar terms), marking the highest level for the period in the company's history.
Overall, the AI wave continues to advance vigorously, and related companies are likely to see sustained performance. In summary, demand certainty remains resilient, macroeconomic disturbances appear limited, and supply bottlenecks are gradually easing. Interested investors may continue to monitor the Communication ETF (515880).
Risk Disclosure: Investors should fully understand the differences between regular fund investment plans and savings methods like lump-sum deposits. Regular fund investment is a simple way to promote long-term investment and average cost, but it does not eliminate the inherent risks of fund investing, guarantee profits, or serve as an equivalent substitute for savings. Equity ETFs/LOFs/structured funds are securities investment fund products characterized by higher expected risk and higher expected returns; their expected risk and return levels are higher than those of hybrid funds, bond funds, and money market funds. Funds investing in STAR Market and ChiNext stocks face specific risks related to investment targets, market systems, and trading rules; investors should take note. Short-term price changes for sectors/funds are presented only as supplementary material for analysis and are for reference only, not constituting a guarantee of fund performance. Mention of short-term individual stock performance is for reference only and does not constitute stock recommendations or guarantees of fund performance. The views above are for reference only and do not constitute investment advice or promises. When purchasing relevant fund products, please adhere to regulations on investor suitability, complete a risk assessment beforehand, and purchase fund products matching your own risk tolerance. Funds carry risks; investment requires caution.