The computing power sector received a significant boost on August 11 when the Shanghai Municipal Commission of Economy and Informatization released the 15th Five-Year Plan for the city's software and information services industry. The plan targets a total industry scale of 4 trillion yuan by 2030, with added value exceeding 1.1 trillion yuan, positioning Shanghai as a growth engine and a global hub for AI-empowered applications.

The plan emphasizes accelerating the digital transformation of the intelligent computing cloud industry. It calls for enriching diverse intelligent computing cloud services, guiding top-tier cloud providers to develop lightweight, high-throughput AI inference foundations with "expert-level intelligence" to meet large-scale industrial needs. The initiative also includes building comprehensive intelligent computing public platforms that integrate computing scheduling, model incubation, and application adaptation. Cloud service providers are encouraged to transition into "computing power + development tools + model service providers," strengthening integrated software engineering and cloud-network-security capabilities. To enhance the ecosystem, the plan aims to create benchmark application scenarios for vertical models in sectors like finance, manufacturing, education, healthcare, tourism, and urban governance. Additionally, it promotes market-oriented reforms for computing power, exploring new investment models like computing power as equity and novel "computing resources + application scenarios" recruitment approaches.

The plan also introduces a "Hundred-Thousand-Ten Thousand" intelligent computing cluster initiative, establishing a tiered supply system. This includes deploying 100,000-card-scale ultra-large intelligent computing clusters in areas like Songjiang, Lingang, and Qingpu, alongside 1,000-card-level clusters in Baoshan, Pudong, and Jiading. Traditional data centers and telecom rooms will transition to 100-card-level edge intelligent computing centers to meet ultra-low latency demands. As AI models and applications proliferate, the demand for AI computing power is surging globally.

In another development, NVIDIA announced a strategic partnership with six Wall Street institutions—Apollo Global Management, Blackstone Group, BlackRock, Brookfield Asset Management, Goldman Sachs, and KKR—to create a standalone computing financing platform, aiming to mobilize over $500 billion in third-party capital for AI infrastructure. Domestically, China's first fully domestic 100,000-card AI supercluster recently became operational, marking a new phase in computing infrastructure. Multiple computing hubs are also undergoing expansion. In Shanghai, the Commerce Commission and other agencies released a plan to promote the export of computing power and foundational models, encouraging AI product and service outbound efforts to bolster global influence.

Looking ahead, Guosheng Securities believes North American cloud providers' capital expenditures have not peaked and are expected to rise. As multimodal models, embodied intelligence, and AI agents deeply penetrate industries, AI computing demand is shifting from training to massive inference, with no signs of slowing. Guohai Securities notes that overseas CSP and NeoCloud vendors are proving that large-scale computing capital expenditure can effectively convert into revenue and profits. Driven by agents, inference demand holds long-term growth potential, and with domestic high-end AI server prices rising sharply, supply constraints are intensifying. This makes computing power leasing a structurally attractive segment, entering a phase of earnings release.

In subsectors, CICC highlights that computing expansion and electrification are driving thermal management upgrades. The global cooling components market is projected to grow at a 59.8% CAGR from 2025 to 2027, with data center cooling alone potentially surpassing 315.1 billion yuan by 2027.

Key concept stocks include Qiniu Smart (02567), which is seeing a value reassessment thanks to the AI agent paradigm shift and rising model usage demand. Its AI-related revenue reached 437 million yuan in 2025, contributing 24.7% of total revenue, up from 22.2% in the first half. Its adjusted EBITDA turned positive for the first time since listing, marking a transition from scale growth to profitability. As of December 31, 2025, the company had over 1.8 million registered developer users, with over 160,000 using AI large model services.

Converge Technology (01729), a key part of the Luxshare system, is well-positioned in data center businesses like MPO optical communications and AI servers, with automotive operations boosted by the acquisition of Leoni's cable business, aiming to become a top global automotive cable supplier. CMB International Securities sees significant growth potential, forecasting total revenue of 20.7/27.4/34.1 billion Hong Kong dollars for fiscal years 2026/2027/2028, with net profits of 1.35/2.03/2.49 billion Hong Kong dollars, maintaining a "Strong Buy" rating.

On August 11, Zhipu (02513) announced a major upgrade to its AI programming product ZCode, introducing Goal, Subagents, Remote Control, and Idle Task features, moving Coding Agent from dialogue assistance to autonomous complex task delivery. ZCode has surpassed 1 million users, and to celebrate, GLM Coding Plan user weekly limits were reset. ZCode, combined with GLM-5.2, achieved a 2.39% higher task pass rate than with Claude Code, particularly excelling in cross-file, multi-step tasks requiring final verification. ZCode also boosted effective GLM Coding Plan capacity by about 30% through cache optimization, with overall usage nearly doubling when combined with a 1.5x activity until August 31. This shows Zhipu is accelerating its "foundation model + Coding Harness + subscription service" AI programming ecosystem.

Tianshu Zhixin (09903), a leading domestic GPGPU manufacturer with the Tiange and Zhikai product lines, has received an initial coverage note from Bank of America Securities, setting a target price of HK$1,013. The bank cites three key drivers: benefiting from domestic substitution trends in China's fast-growing AI accelerator chip market; gaining market share through new product launches and expanded enterprise customers; and a diversified supply chain supporting sustained revenue growth.