To The Moon

Key Points

Last week, the market narrative continued to concentrate on AI and earnings. In U.S. equities, the Philadelphia Semiconductor Index rose 11% for the week and posted a rare 18-day winning streak. Intel’s earnings and guidance significantly exceeded expectations, further driving a re-rating of CPUs and the broader semiconductor sector. We believe that, as agent workflows and agentic AI applications accelerate, there remains a sizable gap between market expectations and the medium- to long-term prosperity of CPUs, memory, and related AI infrastructure value chains. On the macro front, the marginal market disruption from U.S.-Iran tensions has clearly diminished. However, given that CTA positioning is already elevated and month-end pension rebalancing may bring temporary selling pressure, volatility in U.S. equities could increase over the next two to three weeks.

This week, major technology giants will report earnings in a concentrated window. The market’s focus will not only be on whether results can beat expectations, but also on whether revenue growth and AI commercialization prospects can continue to materialize behind the backdrop of intensive CAPEX. In Greater China, DeepSeek last week released a new model adapted for Huawei chips and optimized for AI Agent scenarios. This further reinforced expectations that the closed loop of “domestic models + domestic computing power” is accelerating, making medium- to long-term opportunities across the domestic computing power, server, PCB, memory, networking and communications, and cloud infrastructure value chains increasingly worth attention.

—— Tiger Brokers (HK) Asset Management CIO David Chen

U.S. and Greater China Market

■ Last week, U.S. equities continued their strong momentum. The Philadelphia Semiconductor Index, which includes major AI upstream hardware names, rose 11% for the week and posted a rare 18-day winning streak. Intel’s results beat expectations across the board, and the company significantly raised its second-quarter revenue guidance. Its share price jumped 23.65% last Friday, driving gains in CPUs and even the broader semiconductor sector. We believe CPUs will go through three stages in the AI era:

1. On the model training side, as the industry moves from pre-training to post-training, the share of CPUs relative to GPUs increases. This was the main source of incremental demand in the past.

2. In agent workflows, processes such as planning, orchestration, and verification rely on CPU-based computation, further increasing the CPU share. This is the point the market has currently recognized and started to price in.

3. As agentic AI applications grow explosively, computational demand derived beyond the agent workflow itself — such as services behind APIs and the operation of software generated by AI coding — will create CPU demand far beyond that of human software application scenarios before 2022. This is the area receiving relatively little attention from the market and sell-side institutions, and where the expectation gap is the largest. We believe the market has still not fully priced this in. Since the end of last year, we have already increased our allocation to the CPU and memory value chains, and we will continue to hold these positions through this wave.

■ On the macro front, the marginal impact of U.S.-Iran tensions on U.S. equities has clearly declined. Although the two sides did not hold a second round of talks last week, both the duration and magnitude of the market pullback were much more contained than before, indicating that investors no longer view geopolitical risk as the dominant variable for U.S. equities, but rather as a temporary disruption. According to media reports, Iran has submitted a new three-stage proposal to the U.S. via Pakistan: first, to promote a ceasefire; second, to discuss the reopening of the Strait of Hormuz; and finally, to address the follow-up handling of nuclear issues. However, both the U.S. and Israel are prepared for a resumption of hostilities, leaving the outlook still uncertain. In addition, according to institutional data, CTA funds have largely completed their buying, leaving limited room for further inflows. Meanwhile, month-end pension rebalancing this week is expected to bring USD 25 billion of selling pressure. As a result, U.S. equities face a meaningful risk of correction over the next two to three weeks.

U.S. and Greater China Market

■ This week will mark a true earnings super week, with all five major technology giants — Alphabet, Microsoft, Meta, Amazon, and Apple — set to report results. The market will also conduct a concentrated test of the current technology rally. The first-quarter earnings season has not started poorly overall. According to FactSet, as of April 24, 28% of S&P 500 companies had reported results, with more than 80% beating EPS expectations. In addition, the S&P 500’s first-quarter net margin is expected to reach 13.4%, while the information technology sector’s net margin is expected to be around 29.2%, both at very strong levels. The most critical issue for U.S. equities this week is not whether the technology giants can beat market expectations, but whether the heavy CAPEX cycle continues to be matched by sufficiently strong revenue realization and a clearer outlook for AI commercialization. We remain optimistic about the long-term earnings prospects of the AI theme as AI Agents continue to accelerate.

■ In Greater China, the market narrative is also concentrating on AI and advanced manufacturing. Last week, DeepSeek released a new model adapted for Huawei chips and explicitly stated that the model is especially suitable for AI Agent scenarios. This directly reinforced market expectations that the closed loop of “domestic computing power + domestic models” is accelerating. Meanwhile, Victory Giant Technology surged 50% on its first day of trading in Hong Kong, becoming Hong Kong’s largest IPO in nearly seven months and once again confirming investors’ strong interest in China’s AI hardware value chain. We believe that, in the past, the market focused more on the logic of domestic substitution. What deserves greater attention now is that, as model capabilities continue to improve and Agent applications gradually land, the entire industrial chain — including domestic computing power, servers, PCBs, memory, network communications, and cloud infrastructure — is likely to enter a longer and deeper cycle of prosperity than previously expected.

Disclaimer

1. The information contained in this document is for reference only and does not constitute any financial advice or a transaction offer, solicitation, suggestion, recommendation or any guarantee for any financial product, strategy or service. You should make your own investment decisions and bear the risk of investment responsibility independently.

2. The content of this document is based on reliable data sources that the staff believed to be reliable at the time of production. The Tiger Investment Research team may adjust without prior notice. The Tiger Investment Research team does not guarantee the accuracy, reliability or completeness of the content of this document, and does not assume any responsibility for any transactions arising from the content of this article and its derivative consequences.

3. This document is confidential and non-public and can only be accessed by professionals with corresponding risk-taking capabilities and preferences. Without the prior consent of Tiger, no one may copy or distribute it in any form.

Disclaimer: Investing carries risk. This is not financial advice. The above content should not be regarded as an offer, recommendation, or solicitation on acquiring or disposing of any financial products, any associated discussions, comments, or posts by author or other users should not be considered as such either. It is solely for general information purpose only, which does not consider your own investment objectives, financial situations or needs. TTM assumes no responsibility or warranty for the accuracy and completeness of the information, investors should do their own research and may seek professional advice before investing.