Global Vision Media Focus
The company was featured on CCTV a total of 15 times, and its reach through Party media, central media, and local official media reached a total of 50 million people.
CAS Xi'an Aerospace unveils "Quantitative Hyperspectral Remote Sensing Intelligent Computing Constellation": Competing in the "Space Computing Power" Track
On April 24, during the 2026 China Space Conference held in Chengdu, CAS Xiguang Aerospace released the "Quantitative Hyperspectral Intelligent Computing Constellation". This innovative achievement deeply integrates hyperspectral remote sensing with on-board intelligent computing, realizing the transformation from "space sensing and ground computing" to "space data and space computing".
With the accelerated integration of commercial aerospace and artificial intelligence, space computing power has become a hotly contested sector in the global technology and investment communities. The "Quantitative Hyperspectral Remote Sensing Intelligent Computing Constellation" significantly enhances the efficiency and commercial value of aerospace information services with its unique technological advantages and application capabilities.
Release of the "Quantitative Hyperspectral Remote Sensing Intelligent Computing Constellation"
In recent years, China's remote sensing satellite industry has continued to develop rapidly, making significant progress in technological breakthroughs, constellation networking, and application expansion. Data from the China Geographic Information Industry Association shows that by 2025, my country had launched over 120 remote sensing satellites, bringing the total number of civilian remote sensing satellites in orbit to over 640, maintaining its position as the world's second-largest remote sensing nation. These satellites cover optical, hyperspectral, infrared, and microwave types, enabling all-weather, all-time Earth observation.
Remote sensing satellites are artificial satellites equipped with sensors that conduct long-distance detection of the Earth's surface from space. They receive electromagnetic waves reflected or emitted by the Earth, convert them into electrical signals, and transmit them to ground stations. After processing, these signals form usable data and images, enabling comprehensive monitoring of the Earth's surface, atmosphere, ocean, and other systems. Remote sensing satellites are one of the core carriers of the aerospace information industry. Based on sensor type, remote sensing satellites can be divided into optical remote sensing satellites (visible light, infrared), synthetic aperture radar (SAR) satellites, and hyperspectral remote sensing satellites.
Unlike traditional remote sensing, which primarily provides "image information," hyperspectral remote sensing captures the spectral characteristics of ground objects across continuous bands, enabling the inversion of physical and chemical parameters of substances and achieving "quantitative" identification rather than "qualitative" observation. The "Quantitative Hyperspectral Intelligent Computing Constellation" recently released by CAS Xi'an Optics & Light Aerospace deeply integrates hyperspectral remote sensing with onboard intelligent computing, allowing the satellite to complete data processing, target identification, and information extraction while in orbit, without relying on ground-based computing power for data transmission.
This technological shift means that remote sensing applications will move from the traditional "space-based sensing and ground-based computing" model to a "space-based data processing and computing" model, which integrates sensing and computing in space. This will significantly improve timeliness and reduce data transmission pressure, providing crucial support for scenarios with extremely high timeliness requirements, such as emergency rescue, environmental monitoring, and precision agriculture.
Currently, commercial spaceflight is standing at a true industry-level inflection point. The explosion of low-Earth orbit constellations and the surge in demand for AI computing power have combined to mean that spaceflight is no longer just about "sending satellites into space," but is moving towards an era of space computing power where "data can be calculated and models can be run directly in space."
Against this backdrop, on April 24, CAS Xiguang Aerospace released the "Quantitative Hyperspectral Remote Sensing Intelligent Computing Constellation". This intelligent computing constellation is positioned with "computing power networking" as its core, abandoning the single "photographing satellite" model, and plans to build a space-based intelligent sensing system with 158 hyperspectral satellites working together, high-frequency revisit, and rapid response.
"Software and hardware synergy": Creating a new benchmark for "daytime data and calculation"
Why plan and build a "quantitative hyperspectral remote sensing intelligent computing constellation"?
Qin Jing, Chairman of Xi'an Institute of Optics and Precision Mechanics CAS Aerospace Science and Technology Group Co.,LTD, stated, "Currently, the global remote sensing satellite industry is entering a period of rapid development, with commercial remote sensing becoming the mainstay of the market. Traditional hyperspectral remote sensing faces three core pain points: limited storage, expensive transmission, and delayed timeliness. Massive amounts of data are discarded in orbit, and processing cycles can take several days, making it difficult to meet the rapid response needs of scenarios such as emergency monitoring, precision agriculture, and ecological environmental protection." Qin Jing added that, in contrast, the "Quantitative Hyperspectral Remote Sensing Intelligent Computing Constellation" promotes the transformation of remote sensing data processing from "serial processing on-board" to "parallel processing on-board" by deploying distributed computing nodes in space. This can significantly improve timeliness and reduce data transmission pressure, truly transforming space-based information from "after-the-fact" to "early-stage."
The "Quantitative Hyperspectral Remote Sensing Intelligent Computing Constellation" released by CAS Xi'an Aerospace has the following characteristics:
Firstly, in terms of "hard reconstruction," the satellite platform is equipped with a hyperspectral payload with a resolution better than 10 meters, covering the visible to near-infrared spectral bands, and has the ability to accurately analyze the composition of ground objects; it integrates a high-performance CPU+GPU+NPU heterogeneous computing architecture (single-machine computing power output of more than 248 TOPS, which can realize the deep integration of on-orbit intelligent mission planning and real-time data processing); multi-level orbital coordination covers SSO 530km/560km and 40° inclination orbits, taking into account both global surveys and high-frequency monitoring of specific areas; the communication system integrates Beidou short message service, 200Gbps inter-satellite laser link and satellite Internet to achieve integrated data acquisition and transmission, breaking through the transmission bottleneck.
Secondly, regarding the "soft engine," CAS Xiguang Aerospace pioneered the on-board deployment technology of hyperspectral AI large models. Through model compression and lightweight inference, large models are efficiently embedded into the on-board platform. Simultaneously, vertical models for industries such as agriculture, forestry, water bodies, mining, and carbon dioxide are constructed, and data decoding, cloud judgment, correction, feature extraction, and quantitative inversion are completed in orbit.
Furthermore, relying on onboard intelligent processing and multi-satellite collaborative scheduling, the constellation achieves end-to-end rapid response, significantly compresses data processing cycles, reduces data transmission costs by more than 80%, and greatly improves data utilization, realizing "data acquisition, processing, and decision-making in space," thus upgrading satellites from "space cameras" to "space intelligent terminals."
Qin Jing stated that the "Quantitative Hyperspectral Intelligent Computing Constellation" has created a new era of "on-orbit computing and real-time intelligence" by building a space-based intelligent network that integrates communication, navigation, and remote computing. This will drive the remote sensing industry to upgrade from "providing raw data" to "providing direct decision-making capabilities".
It is worth mentioning that, in order to promote the "Quantitative Hyperspectral Remote Sensing Intelligent Computing Constellation" to a new stage of "constellation intelligence, data autonomy, advanced models, and abundant computing power", CAS Xiguang Aerospace also signed a strategic cooperation agreement with Zhijiang Laboratory during the 2026 China Aerospace Conference.
Currently, Xi'an Optics & Light Technology Co., Ltd. has successfully launched and is operating 11 satellites. The world's first hyperspectral remote sensing intelligent computing satellite, "Xi'an Optics-2 01", is scheduled to be launched in mid-2026.
The race for space computing power heats up
Space computing power, represented by the "Quantitative Hyperspectral Intelligent Computing Constellation," is constantly expanding its application scenarios, widely covering various key areas related to national economy, people's livelihood, and national security.
For example, in the field of agricultural production, by rapidly analyzing remote sensing data in orbit, it is possible not only to identify soil moisture but also to accurately invert nitrogen, phosphorus, and potassium content, diagnose crop nutrient deficiencies, and even provide early warnings of pests and diseases through subtle changes in leaf spectra, making farming more scientific and efficient. In the field of ecological and environmental protection, it is possible not only to "discover whether there is pollution" but also to "identify what kind of pollution it is," building an early warning system of "rapid perception and immediate response." In the field of resource exploration, relying on hyperspectral technology to accurately identify the spectral "fingerprints" of minerals, it is possible to quickly delineate prospective mineral areas and identify mineral distribution, overturning the traditional exploration model, significantly shortening the mineral exploration cycle, narrowing the target area, reducing costs, solving the problem of exploration in complex terrain, and promoting the leap of exploration towards "data-driven and intelligent mineral exploration."
As application scenarios open up new possibilities, space computing power is moving from proof-of-concept to large-scale implementation, becoming a hot track for global commercial aerospace competition. Leading domestic and foreign companies and research institutions are increasing their investment to form differentiated technology routes and industrial layouts.
Internationally, SpaceX is upgrading its Starlink constellation towards space computing power and space-based data centers, leveraging its advantages in large-scale networking and low-cost launches to seize a leading position in the global computing infrastructure market. Amazon has made a strong entry into the space computing field through Project Kuiper, planning to deploy 3,236 low-Earth orbit satellites to build a global space computing network. Starcloud, in collaboration with technology companies such as NVIDIA and Google, is advancing the on-orbit training and deployment of large models to achieve the engineering verification of space GPU computing power. Remote sensing giants such as Planet and Maxar are also strengthening their on-board edge computing capabilities to improve the efficiency of real-time delivery of remote sensing data.
China has also been actively involved in this space computing race. Zhejiang Lab is advancing the construction of its "Three-Body Computing Constellation," having already completed the networking of 12 computing satellites, thus building a leading space-based distributed computing platform in China.
Unlike the general space computing power approach, CAS Xiguang Aerospace has chosen to deeply integrate with the vertical field of hyperspectral computing, building a technological barrier with "dedicated payload + exclusive algorithm + proprietary computing power", which has stronger applicability in high-value industries such as agriculture, forestry, water bodies, mining, and carbon.
"Our goal is clear: to build China's largest, most comprehensive, user-friendly, and practical remote sensing constellation," said Qin Jing. Looking towards the 15th Five-Year Plan, CAS Xi'an Aerospace will continue to deepen its expertise in hyperspectral remote sensing and space-based intelligent computing, focusing on key areas such as on-board intelligent computing, hyperspectral AI large-scale models, and inter-satellite collaborative networking. The aim is to create a new generation of intelligent computing remote sensing systems with wider coverage, faster response, and more accurate diagnosis, continuously empowering various industries with aerospace technology.
Multiple data sources indicate that space computing power and hyperspectral remote sensing are experiencing rapid growth. According to Fortune Business Insights, the global space AI market reached $2.36 billion in 2025 and is projected to exceed $15 billion by 2034, with a compound annual growth rate (CAGR) of 22.91% from 2026 to 2034. A Research And Markets report points out that the global on-orbit data center market will reach $39 billion in 2035, with a CAGR of 67.4% from 2025 to 2035, making it the fastest-growing segment of the space economy. Domestic securities research institutions estimate that the market size of my country's satellite internet and space-based computing power is expected to exceed one trillion yuan in the long term, with hyperspectral remote sensing and intelligent computing services accounting for an increasingly larger share.
