On the afternoon of August 14, the computing power sector experienced a sudden surge, with computing power rental stocks rebounding sharply. Data港 quickly hit its daily limit, 网宿科技 soared 20%, 奥飞数据 rose over 7%, and 云赛智联, 二六三, and 首都在线 were among the top gainers. The pure-play computing power ETF, Cloud Computing ETF (159099), saw its intraday gains spike over 2% before closing up 1.95%.
Entering August, the computing power rental concept has continued to strengthen. As of the close on August 14, its cumulative gain reached 14.59%, significantly outperforming broader tech indices like the ChiNext Index and the STAR 50. Strong earnings and share price rallies among overseas computing power rental giants have had a clear mapping effect on the domestic market. Additionally, the certainty of a computing power shortage, combined with enhancements in large models and downstream applications, is expected to further boost demand for computing power, improving the sector's future earnings visibility.
As of the latest data, the underlying index of Cloud Computing ETF (159099), the Cloud Computing 50 Index, has a computing power rental weight of 58.95%, leading peers like the Cloud Computing Index (42.43%) and the Shanghai-Hong Kong-Shenzhen Cloud Computing Index (46.11%).
Key Market Signal: DeepSeek's Price Hike
On the news front, DeepSeek's price hike plan has been implemented, adopting a peak-valley pricing model where valley prices are half of peak prices. Peak hours are set for 9:00-12:00 and 14:00-18:00 Beijing time, with the rest considered off-peak. According to media calculations, using the newly released DeepSeek V4 Pro as an example, the input price per million tokens is 12 times higher (cache hit) and 3 times higher (cache miss) than before, while the output price is 4.5 times higher. Even during off-peak hours, these three pricing tiers are 6 times, 1.5 times, and 2.25 times the original prices, respectively. This move signals a shift in the large model industry from a "price war" to a focus on sustainable profitability, while also confirming the tight supply of computing power amid the AI wave.
Notably, the price hike for large models shifts profits upstream in the industry chain, making computing power rental the most directly benefiting segment. Additionally, Shanghai recently issued the "Action Plan for Implementing Measures to Further Promote Private Investment Development," explicitly proposing to provide computing power subsidies in accordance with laws and regulations and supporting private enterprises in renting intelligent computing resources for large model R&D, training, and application deployment.
Data from the China Academy of Information and Communications Technology (CAICT) shows that the domestic computing power rental market size is expected to exceed 260 billion yuan in 2026. Over the medium to long term, the market is projected to approach the trillion-yuan mark by 2030, with the intelligent computing sub-segment maintaining a compound annual growth rate of over 43%.
Institutional Perspectives
Northeast Securities points out that the AI industry has entered a phase of large-scale commercialization, with model iteration and diverse scenarios jointly driving up demand for computing power. Leading internet companies are continuously locking in long-term intelligent computing contracts, corroborating the current tight supply of computing power. Small and medium-sized AI players, constrained by capital and operational barriers, struggle to build their own clusters, which continues to broaden the base for computing power rental demand. As token call volumes grow exponentially, the industry's business model is shifting from traditional per-card or per-hour rental to more refined token-based pricing. Coupled with tightening supply of high-end overseas GPUs and the rollout of domestic computing power support policies, the sector has a clear trend of both volume and price increases.
As a core infrastructure for the AI era, cloud computing offers strong thematic investment value against the backdrop of expanding AI computing power, accelerating data center construction, and rising cloud infrastructure prosperity. Cloud Computing ETF (159099) passively tracks the CSI Cloud Computing 50 Index, with a computing power rental weight of nearly 60%. Its top ten constituent stocks cover leading companies in core areas, including optical modules (中际旭创, 新易盛), servers/AI computing power (中科曙光, 浪潮信息), data centers (润泽科技), and software platforms (金山办公, 恒生电子), helping investors fully capture opportunities in AI computing power expansion and cloud infrastructure prosperity.
Data source: Shanghai and Shenzhen Stock Exchanges, Wind. Note: The computing power rental weight refers to the combined weight of Cloud Computing ETF (159099)'s underlying index components that are also part of the IDC (computing power rental) index. As of August 14, this figure was 58.95%. Institutional view source: Northeast Securities, August 6, 2026, "Computer Industry: Computing Power Rental: Industry Structure Reconstructing, Value Re-evaluated."
ETF fund related fees: When investors subscribe or redeem fund shares, the subscription and redemption agency may charge a commission of up to 0.5% of the standard rate. On-exchange trading fees are subject to the actual fees charged by the securities company, and no sales service fee is charged.
Risk Disclosure: Cloud Computing ETF (159099) passively tracks the CSI Cloud Computing 50 Index, with a base date of December 31, 2014, and a release date of June 12, 2020. The Cloud Computing 50 Index's annual returns over the past five complete years were: -9.71% in 2021, -23.97% in 2022, 11.81% in 2023, 27.53% in 2024, and 93.52% in 2025. Its annualized volatility over the past five complete years was: 20.91% in 2021, 27.12% in 2022, 33.95% in 2023, 42.39% in 2024, and 40.00% in 2025. The composition of the index constituents is adjusted in a timely manner according to the index compilation rules. The back-tested historical performance does not predict future performance. This fund is issued and managed by 华宝基金. The selling agency does not assume responsibility for the investment, payment, or risk management of the product. Investors should carefully read the fund's legal documents, including the "Fund Contract," "Prospectus," and "Fund Product Information Summary," to understand the fund's risk-return characteristics and choose a product that matches their risk tolerance. The fund manager assesses the fund's risk rating as R3-Medium Risk, suitable for balanced (C3) and above investors. When subscribing or purchasing through a selling agency, the specific risk rating result shall be based on the rating of the fund sales institution. Selling institutions (including the fund manager's direct sales institutions and other selling institutions) conduct risk evaluations of this fund in accordance with relevant laws and regulations. Investors should promptly pay attention to the suitability opinions issued by the fund manager. The suitability opinions of various selling institutions may not be consistent, and the risk rating results of fund products issued by fund sales institutions shall not be lower than the risk rating results made by the fund manager. The fund's risk-return characteristics and risk rating in the fund contract may differ due to different factors considered. Investors should understand the fund's risk and return situation and carefully select fund products based on their own investment objectives, time horizon, investment experience, and risk tolerance, assuming the risks themselves. The registration of this fund by the China Securities Regulatory Commission does not imply a substantive judgment or guarantee of the fund's investment value, market prospects, or returns. The performance of other funds managed by the fund manager does not constitute a guarantee of the fund's performance. Past performance does not predict future returns. Fund investment involves risks, and caution is advised!