Author | Ren Tianqin

Editor | Chen Xiaoran

As of the market close on July 10, the A-share space computing sector surged by 5.45% against the market trend, with five core industrial chain stocks hitting their daily limits or experiencing significant price jumps. Net inflows of principal funds exceeded 4 billion yuan, marking a clear direction for the cross-border integration of AI and the aerospace industry in the capital market.

With ground-based computing power facing bottlenecks in energy and physical space, deploying computing nodes in low-Earth orbit has emerged as a potential solution. However, the rapid growth of this sector is accompanied by pressing challenges, including technological costs, the gap between China and foreign counterparts, and the need for commercialization.

Capital Chases New Narratives

Market fluctuations in the secondary market serve as a leading indicator of industrial trends.

On that day, the Shanghai Composite Index fell by 1.0%, and the Shenzhen Component Index dropped by 2.29%, with most sectors experiencing corrections. However, the space computing sector bucked the trend, rising by 5.45% overall and attracting a total of 4.17 billion yuan in principal funds. Multiple industrial chain listed companies saw their share prices hit the daily limit.

This market rally did not occur spontaneously. Traditionally, aerospace satellites have been limited to basic functions such as remote sensing photography and signal relaying. The massive amounts of raw data collected must be transmitted back to ground data centers for processing, a model known in the industry as "space sensing, ground computing."

However, the explosive growth in demand for AI large model training and multi-dimensional remote sensing monitoring has exposed the shortcomings of this traditional architecture. Massive data transmission consumes significant bandwidth resources, and the round-trip of data introduces high latency, making it difficult to adapt to scenarios requiring real-time responses, such as emergency disaster relief and ocean monitoring.

Meanwhile, the expansion of ground computing clusters has reached a practical limit.

By 2025, the total global electricity consumption of data centers is projected to reach 485 terawatt-hours, equivalent to Japan's annual electricity consumption, and is expected to double to 950 terawatt-hours by 2030.

In Europe and the United States, delivery lead times for data center transformers have extended more than tenfold, with power infrastructure, land supply, and cooling costs escalating, trapping ground computing power expansion in a bottleneck.

The supply-demand contradictions are forcing the industry to seek new solutions. The concept of integrating servers and computing chips onto satellite platforms to create in-orbit native computing capabilities has emerged.

Satellites equipped with computing payloads can perform on-site data collection and processing, transmitting only filtered and valid information to the ground, significantly reducing transmission loads and energy consumption.

Overseas tech giants have taken the lead in technological exploration. In June 2026, SpaceX disclosed the design plan for its AI1 computing power satellite, featuring a high-power solar power supply array and computing power comparable to NVIDIA's high-end ground AI cabinets. It plans to rapidly deploy a space-based computing constellation on a large scale by the end of 2027, with long-term plans to build a massive Starmind computing network composed of one million satellites.

The domestic industry is also keeping pace with deployment, having completed the network launch of the first batch of 12 space computing experimental satellites in 2025. In 2026, the Beijing Space Computing Innovation Center was established, building a complete research and development system from chip design and payload development to large model in-orbit deployment. Leading commercial aerospace companies have even proposed long-term plans for constellations comprising thousands of satellites.

The simultaneous deployment of domestic and foreign industries has jointly spurred this concept explosion in the capital market, with space computing power officially transitioning from science fiction to industrial research and development reality.

However, for space computing power to move beyond the conceptual realm and achieve large-scale deployment, it requires coordinated support from the entire aerospace industrial chain, from upstream to downstream.

From satellite manufacturing, in-orbit energy supply, and space-ground communication transmission to mass production and assembly of components and the formation of closed-loop collaboration in ground-supporting infrastructure, the multiple listed companies leading this rally are precisely distributed across various key nodes of the industrial chain, collectively piecing together the complete puzzle of industrial deployment.

Satellite manufacturing is the physical carrier of space computing power. Only by achieving large-scale, standardized mass production of small satellites can the high-density networking requirements of low-Earth orbit computing constellations be met.

In traditional aerospace, satellites are mostly custom-developed and produced, with high manufacturing costs and long delivery cycles, making them unsuitable for networking plans involving tens of thousands of satellites. The industry is accelerating its transition to an assembly line mass production model. Leading Chinese manufacturers are planning intelligent factories with annual production capacities of hundreds of satellites, significantly shortening the manufacturing cycle for individual satellites and laying a production foundation for the launch of massive computing satellites.

Energy supply is the core lifeline for the continuous operation of in-orbit computing power. With no atmospheric obstruction in space, sunlight intensity can reach 7 to 10 times that on the ground, making the conversion of solar energy through high-efficiency space photovoltaic cells the mainstream solution. Gallium arsenide multi-junction photovoltaic cells, with their over 30% energy conversion efficiency and strong resistance to cosmic radiation, have become essential components for the industry and a key focus for upstream material companies.

The space-ground communication link is responsible for the data interchange between the space computing power cluster and the ground control center. Inter-satellite laser and microwave composite transmission technologies can establish high-speed data channels between satellites and between space and ground, preventing in-orbit computing units from becoming isolated nodes. Relying on communication networking technologies, computing satellites dispersed in low-Earth orbit can form a distributed computing network, enabling collaborative task scheduling.

In addition, the massive constellation networking cannot be achieved without upstream component-supporting enterprises supplying propulsion systems, structural components, and other hardware. Simultaneously, a large number of ground control stations and supporting data centers require special geological construction techniques to build infrastructure. From upstream materials and mid-stream manufacturing integration to communication networking and ground infrastructure, full-chain enterprises are entering the fray simultaneously, jointly constructing the industrial framework for space computing power.

At this stage, Chinese enterprises are mostly participating in the deployment in the form of securing niche segments and providing supporting supplies, with no leading enterprise yet capable of covering the entire industrial chain. The industry is characterized by a refined division of labor, reserving business entry opportunities for listed companies in different fields.

Industry forecast reports issued by multiple institutions outline the broad growth prospects for space computing power.

According to Fortune Business Insights, the global space-based edge computing market size reached USD 168.91 billion in 2025 and is expected to climb to USD 345.04 billion by 2034. Domestic industrial planning estimates that the short-term implementation of the first phase of the computing constellation can drive tens of billions of yuan in industrial chain output value, with long-term development expected to grow into a trillion-yuan sector.

Goldman Sachs has significantly raised its global low-Earth orbit satellite installation expectations, believing that after 2029, space data centers will replace traditional satellite internet as the core driving force for low-Earth orbit constellation construction, with the global number of in-orbit satellites expected to reach 400,000 by 2031.

From an application perspective, space computing power can be applied in disaster emergency response, polar and deep-sea signal coverage, low-altitude economic control, global logistics tracking, and other scenarios, compressing the original hour-level data processing timeliness to seconds, spawning new business models for space data services. Simultaneously, deploying space-based computing power is also a strategic move to build digital space sovereignty.

Behind the promising sector prospects, unavoidable practical barriers are also constraining rapid industrial deployment, with high deployment costs being the primary challenge.

The development gap between China and the United States in this industry objectively exists. Mature reusable rockets overseas have significantly reduced launch unit prices, while similar technologies in China are still in the experimental stage, with launch costs several times higher than those overseas. Simultaneously, there is a 2- to 3-year technological generation gap in onboard radiation-resistant high-end computing chips. Overseas Starlink has completed the networking verification of tens of thousands of satellites, while domestic constellations are still in the experimental networking stage, with core technological shortcomings requiring long-term research and development.

Based on the current situation, the domestic industry has formulated a differentiated development route with step-by-step iterations. In the short term, priority is given to installing lightweight AI payloads on existing in-orbit satellites to activate idle aerospace assets and implement practical scenarios. In the medium term, the focus is on overcoming core shortcomings such as reusable rockets and high-end chips. In the long term, a collaborative architecture of "ground computing power as the mainstay and space computing power as a supplement" will be constructed, clarifying that space computing power will not fully replace ground data centers but will only focus on special scenarios that ground computing power cannot cover.

Returning to the capital market level, this round of stock rallies represents capital's advance bet on the sector's long-term value. The industry still has a long path ahead, from technological experimentation and cost optimization to commercial model profitability verification.

For market participants, there is no need to blindly chase concept hype. Instead, they should continuously track three key indicators: the implementation progress of core technologies, the extent of launch cost reductions, and the scale of real-world order implementations.

Rationally view this emerging sector that combines strategic value with implementation challenges, and calmly discern the participants with true growth potential during the industry's iterative process.