In 2026, space computing has emerged as a major new direction for the commercial space industry. But what problems is it actually intended to solve, and where is the industry heading? Hello Space has launched its Space Computing Interview Series, speaking with leading Chinese companies to examine the real progress being made in this emerging field.
Our guest for this installment is Zhang Shijie, President of the GalaxySpace Research Institute.
Key Takeaways
- Space computing is not simply about placing computing capacity aboard satellites. It is about building the intelligent space infrastructure of the future.
- Space computing is not intended to replace terrestrial computing. Its first priority is to address the challenges associated with space-generated data.
- Onboard intelligence and computing constellations are complementary rather than competing models, much like cloud, edge and endpoint computing on Earth.
- Demand for computing power has no upper limit and will continue to expand as missions evolve.
- Large-scale computing is closely tied to large-scale energy supply. Greater computing power creates greater energy demand, making energy the foundation of the infrastructure.
- Computing power is a productive tool. The most important question is not how much computing power is available, but what missions it can accomplish.
- Satellites will undergo a transition from quantitative to qualitative change: from transmitting data to transmitting information, from two-dimensional to three-dimensional understanding, and toward much greater intelligence.
- Competition in space computing will ultimately depend not on any single chip or satellite, but on integrated capabilities combining large-scale computing, large-scale energy, large-scale communications, software and applications.
- Space and terrestrial systems will increasingly converge. Space infrastructure will need computing, networking and intelligent capabilities comparable to those on the ground, creating a genuinely coordinated and integrated space–ground system.
- Commercial space companies must strengthen their industrial capabilities and achieve low-cost, large-scale manufacturing. Even hundreds of thousands of satellites may not represent the upper limit.
- Engineering challenges such as energy supply and thermal management can be solved. The fundamental question is how much it will cost to solve them.
Since the concept of space computing began attracting widespread attention, chipmakers have started developing and testing radiation-tolerant AI chips, satellite companies have added computing modules to their payloads, and operators have begun planning dedicated computing constellations.
Yet a more fundamental question remains insufficiently answered: Why should computing power be placed in space? What can satellites actually do once they are given a “brain”?
GalaxySpace is one of the few Chinese commercial space companies that combines mass-production capabilities for communications satellites with constellation deployment experience, while systematically developing the three foundational capabilities of computing, energy and communications.
Starting from the perspective of a satellite manufacturer, this interview explores the industrial logic behind GalaxySpace’s move into space computing.
Satellites once functioned much like cameras: they captured images, transmitted them to Earth, and left ground systems to process the data. Today, however, there is growing recognition that many things would change if satellites could make decisions autonomously in orbit.
In an emergency, for example, a satellite could immediately identify a fire, anomalous vessel activity or signs of a disaster, without waiting for raw data to be transmitted to Earth and processed. Response times could then be reduced from hours to minutes.
Making this possible requires computing, communications and energy.
Computing is the core. As the hardware foundation of artificial intelligence, it determines whether a satellite can run algorithms, perform inference and make decisions in orbit.
Communications are the nervous system. No matter how powerful the computing capability may be, much of its value will be lost if information cannot be transmitted to Earth or satellites cannot coordinate with one another.
Energy is the foundation. Computing chips and communications payloads both consume electricity. Without sufficient energy, none of these systems can operate.
Based on this logic, GalaxySpace has proposed the concept of intelligent space infrastructure. Zhang Shijie has previously said that large-scale computing is the core engine of intelligence, while large-scale energy provides the power required for continuous operation.
Within this framework, computing, communications and energy are placed on an equal footing as the three pillars of intelligent space infrastructure.
Zhang Shijie is President of the GalaxySpace Research Institute and Vice Chair of the CICC Technical Committee on Space Information and Communication Technology. He has long focused on satellite internet and mass-production technologies. Named a Beijing Scholar, Zhang has received numerous honors, including the First Prize of the Beijing Science and Technology Progress Award. He has published more than 80 papers and holds over 20 granted invention patents. Source: GalaxySpace
One practical constraint lies at the heart of this development: communications capacity between satellites and the ground remains severely limited. Put simply, the more data satellites generate, the more difficult it becomes to transmit that data.
What, then, is the most urgent problem that space computing needs to solve?
Zhang offered a direct answer during the interview: If the data cannot be transmitted to Earth, process it in orbit first and send back only the results.
This is also the most practical value of space computing today. It is not about relocating terrestrial supercomputing centers into orbit. Instead, it uses onboard computing to compensate for the widening gap between data volumes and communications capacity.
Satellites can first filter, identify and analyze their data, transmitting only the information with genuine value. In short, space computing is about enabling data to create value while it is still in space.
One conclusion stood out during this interview: the space computing race may not ultimately be decided by a breakthrough in any single computing chip.
The real differentiator may be which company can first coordinate and integrate energy, communications, thermal management, software, manufacturing and other critical components into a complete system-level engineering solution.
The road ahead remains long. Energy supply, thermal management, radiation tolerance and the reliability of inter-satellite laser communications all present engineering challenges.
As Professor Zhang candidly observed during the interview, “There is nothing that cannot be done. The question is the cost.”
The direction, however, is already clear. Once satellites truly develop a “brain,” our understanding of Earth may expand beyond two-dimensional representations toward a more comprehensive three-dimensional perspective. That may ultimately prove to be the most exciting value of space computing.
The following is an edited transcript of Hello Space’s interview with Zhang Shijie, President of the GalaxySpace Research Institute. Portions of the conversation have been organized and edited for readability.
Why Has Space Computing Suddenly Taken Off?
Hello Space: Since late last year, space computing has suddenly become one of the industry’s hottest topics. Why has it taken off at this particular moment, and what are the real forces driving it?
Zhang Shijie: From the perspective of overall technological progress, there are two factors.
First, large models have caused demand for space computing to surge. Second, commercial space technologies have gradually matured.
It is the combination of demand and supporting infrastructure that has made it possible to extend computing from the ground into space. This development is also inseparable from the current state of China’s commercial space industry.
Hello Space: Some people say China’s space computing initiatives are following Elon Musk’s plans for orbital data centers. What is your view?
Zhang Shijie: We can set aside the question of whether China is following Musk. From China’s perspective, there is a clear continuity in the effort to deploy computing capabilities in space.
As early as the 1960s, the Apollo program helped drive the development of integrated circuits. At its core, that effort was about giving spacecraft computing capabilities.
In the past, however, the limited scale of the space market meant that the cost and performance of onboard computing lagged many generations behind terrestrial computing.
It was only after the emergence of large models, when edge computing could no longer satisfy demand, that a need arose to move ultra-large-scale computing capacity into space.
That is the fundamental reason space computing has taken off over the past two years. It did not emerge from nowhere. It is the result of long-term accumulation combined with the maturation of the necessary underlying conditions.
What Problems Is Space Computing Really Meant to Solve?
Hello Space: Do you believe adding computing power to satellites is primarily intended to enable “space data to be computed in space,” or to address shortages in terrestrial computing by moving some ground-based workloads into orbit? What will the principal demand be today and over the next three to five years?
Zhang Shijie: Technology is developing so rapidly that it is difficult to make predictions about the next three to five years.
But we should return to the fundamental question: Why do we want to place computing power in space?
If the objective is to process terrestrial data, the demands placed on space systems would be too high in the short term. A basic principle of spaceflight is to make full use of the advantages of space. If computing is deployed in orbit, we must determine whether its advantages are commensurate with the investment and the missions it is expected to perform.
In my view, for the foreseeable future, the priority will still be to determine how computing can unlock the value of data generated in space.
Hello Space: As satellites become increasingly intelligent, how do we determine which tasks will inevitably need to be performed in orbit and which must remain on the ground? How will this boundary evolve?
Zhang Shijie: The number of satellites is increasing, yet there is a widespread sense that very little of what they produce is genuinely usable. I think there are several reasons for this.
The amount of data that people can access is limited. Earth appears three-dimensional when viewed from space, so how do we deliver the information we observe to people?
One approach is to transmit the observed data to the ground. Another is to perform information assessment and collection in orbit, instead of merely collecting raw data.
The greatest difficulty today is that transmitting information from space to the ground is extremely challenging. This is why, at least for the current stage, we need to process data in orbit first. Only then can we obtain more useful information from space to support activities on Earth.
Let me give a very simple example. Some terrestrial astronomical observatories—including the network of ground-based telescopes that produced the first image of a black hole—transferred data physically using hard drives. This shows that the networks we rely on are already incapable of transmitting data at such enormous volumes.
That is why I say processing information in orbit can create far greater value than simply transmitting all the data to the ground.
Consider today’s remote-sensing satellites. A satellite may collect data for one minute, spend another seven or eight minutes transmitting it to the ground, and then remain effectively idle for more than 90 minutes. Its utilization rate is extremely low.
How can that utilization rate be improved? One option is to increase satellite-to-ground transmission capacity, but communications technology advances relatively slowly. Another is to process the data sources in space and transmit information rather than raw data. That is far more efficient.
This is why the immediate priority is not to process terrestrial data in space, but to extract useful information from space-generated data more effectively.
Hello Space: If computing and communications capabilities in space become much stronger, how will ground support systems change? For example, will fewer telemetry, tracking and command systems or ground stations be required?
Zhang Shijie: There are two sides to this.
On the one hand, as satellites gain stronger processing and automation capabilities, an individual satellite will depend less on ground support systems. It will become more intelligent and autonomous.
On the other hand, the number of satellites in orbit will inevitably continue to increase. Their total number will not decline, so overall demand for ground systems will not decrease. It is like the saying: when the main river rises, its tributaries fill as well.
Hello Space: If more computing capacity and large models are placed aboard satellites, how will the satellites themselves change?
Zhang Shijie: Satellites will look very different from those we see today. They will undergo a progression from quantitative change to qualitative transformation.
Future satellites will operate more intelligently. They will provide more information instead of merely delivering raw data.
More importantly, large groups of these satellites will transform the information society as a whole. The fundamental objective is to exploit the unique advantages of space and expand humanity’s two-dimensional understanding of the world from Earth’s surface into a three-dimensional understanding.
Under those circumstances, I believe there will be profound changes in large models, the acquisition of social information and social governance.
Hello Space: If this transformation is inevitable, what preparations must the commercial space industry make, and what challenges will it face?
Zhang Shijie: From a commercial perspective, companies must possess industrial-scale production capabilities, because industrialization is the foundation of commercialization.
On the technology side, we must continue innovating to increase computing, energy and communications capabilities.
At the same time, satellite manufacturing must become genuinely industrialized, with an emphasis on low costs and large-scale production. This could support the deployment of hundreds of thousands of satellites in the future—and even that may not be the upper limit.
What Is the Relationship Between Onboard Intelligence and Computing Centers?
Hello Space: If the near-term purpose of space computing is to process space-generated data in orbit, and every satellite can be equipped with computing power, do we still need dedicated computing constellations?
Zhang Shijie: The trends in space and on the ground are actually quite similar.
There is already substantial computing power at the edge on Earth, so why do we still need supercomputing centers? Because the development of large models has created computing requirements that cannot be met by dispersed edge computing alone. Larger, independent pools of computing power—or more powerful standalone systems—are needed to meet those requirements.
Under these circumstances, cloud, edge and endpoint computing capabilities must develop in a balanced way. It is not a matter of choosing one over another.
Even if an individual satellite has sufficient computing power, centralized computing capacity will still be necessary. The two are not mutually exclusive; they complement each other and together form a complete system.
Why Will Space Computing Ultimately Require “Large-Scale Computing + Large-Scale Energy + Large-Scale Communications”?
Hello Space: You have repeatedly mentioned “large-scale computing” and “large-scale energy.” What exactly do these concepts mean, and are they equally important?
Zhang Shijie: Large-scale computing and large-scale energy are interconnected.
From an energy perspective, energy supply is itself a service capability. All infrastructure in modern society depends on energy—or, more specifically, on electricity.
In my view, the demand for large-scale energy arises from the demand for large-scale computing.
Hello Space: GalaxySpace rarely talks about computing chips. What is the company primarily developing in the field of large-scale computing?
Zhang Shijie: In the short term, we do not plan to develop computing chips. We will rely on the industrial supply chain to address these foundational issues.
Our main focus is determining how terrestrial industrial-scale computing capabilities can be adapted for satellites and extended into space. That is what we have consistently been working on.
Hello Space: GalaxySpace is a major developer of solar arrays. In which direction will flexible solar arrays evolve?
Zhang Shijie: Their form and type are simply different technological approaches. The most important objective is to provide more energy—or to provide more energy at a lower cost. That is the fundamental criterion.
Whether the arrays are flexible, what kind of solar cells they use, or which other technologies are adopted, the purpose is always to maximize energy supply at the lowest possible cost.
“Cost” here should be understood broadly. It includes direct manufacturing costs as well as launch costs.
Hello Space: Beyond computing and energy requirements and the demands placed on communications networks, are there any other essential or foundational elements of space computing?
Zhang Shijie: If computing power is treated as a productive tool, the most fundamental question is still: “What exactly am I trying to accomplish?”
In space computing, our objective is to support missions. On one side, we need technologies for building computing systems. On the other, we must focus on real applications and combine these elements to create missions or systems that genuinely put the computing power to use.
Why Has GalaxySpace Proposed “Intelligent Space Infrastructure”?
Hello Space: GalaxySpace has proposed the concept of “intelligent space infrastructure.” How does it differ from conventional “space infrastructure,” and what role does GalaxySpace hope to play in it?
Zhang Shijie: Intelligent space infrastructure adds a foundational layer for future intelligence to existing digital space infrastructure.
First, it must provide the digital foundations. Computing and networking are both fundamental capabilities.
It must also support intelligence. This could extend to the physical intelligence of individual systems, large models that are currently under development, and many other forms of intelligence that may eventually be implemented in space.
Space and terrestrial systems will become increasingly similar in the future. To enable coordination between space and Earth, their functions must converge. Intelligent infrastructure in space must develop capabilities comparable to those available on the ground before it can effectively acquire and utilize resources.
In addition to underlying communications and computing systems, intelligent space infrastructure will require intelligent models designed for the space environment. Once this foundation is in place, industries across the economy will be able to develop their own missions.
Hello Space: Given your expectation that space and terrestrial systems will converge, where does GalaxySpace intend to enter the market, and what are its strategic plans?
Zhang Shijie: GalaxySpace’s current priority is to build up its technological capabilities. We aim to support China’s medium- and long-term strategic plans, develop capabilities that serve national needs, work with the market to advance the industry, and continue strengthening our fundamentals.
Hello Space: GalaxySpace is positioned as a satellite manufacturer. Manufacturing is an intensely competitive business in China, and it can be difficult to generate the high profit margins needed to support a high valuation. Beyond mass manufacturing, does GalaxySpace have broader ambitions?
Zhang Shijie: GalaxySpace has largely established its manufacturing capabilities.
However, the company has never confined itself to satellite manufacturing. From the beginning, we have been developing the capabilities required to build infrastructure, including communications, computing and networking.
GalaxySpace will therefore not remain simply a company with strengths in the mass production of satellites. We will continue expanding into more areas of business.
Hello Space: GalaxySpace has consistently positioned itself around the integration of communications, navigation, remote sensing and computing. Why do you believe these capabilities will inevitably converge? Some people think integrating them on a single platform is impractical.
Zhang Shijie: Integration refers to interconnection and interaction between different capabilities or functions.
We have not imposed any restrictions on whether this must be achieved on one platform or across multiple platforms. If combining them on one platform creates greater advantages, we will do that. If separating them allows each capability to be optimized, we will take that approach instead.
Why has GalaxySpace developed its own networking capabilities? Because these functions can be integrated through network virtualization. The same process is already taking place in terrestrial systems.
Hello Space: GalaxySpace develops communications satellites, remote-sensing satellites and now computing satellites, as well as solar arrays and energy systems. It can appear that the company is involved in everything. How will GalaxySpace decide which businesses to pursue and which not to pursue? What is the underlying logic?
Zhang Shijie: The core objective is always to build intelligent space infrastructure.
We focus on whatever constrains the development of that infrastructure. Intelligent space infrastructure requires computing, data sources and communications, so these are the areas in which we will make major investments and pursue large-scale development and innovation.
Why are we also developing large-scale energy capabilities? Because both computing and communications create greater demand for energy.
These are the three core technological directions GalaxySpace has identified, and they are also capabilities the company has been developing over time. Once you understand this logic, it becomes clear what GalaxySpace is doing.
Hello Space: How do you determine when the initial infrastructure has taken shape? Is there a standard—for example, a certain number of satellites or a specified amount of computing power?
Zhang Shijie: It should not be defined by computing power alone.
The same is true of development on the ground. Terrestrial infrastructure was never considered viable only after reaching a particular amount of computing power. It is a process of continuous improvement.
Consider terrestrial mobile communications networks. A network consisting of hundreds of thousands of base stations was still a network, and today a system with millions of base stations is also a network. It continues to improve and evolve.
There is, of course, an initial stage at which the system becomes operational. For example, it may need a basic low-Earth-orbit communications network, as well as the links and channels that allow a computing center to provide basic services to users. These are the fundamental criteria for determining whether the infrastructure has been established.
What Is the Greatest Challenge Facing Space Computing?
Hello Space: Many companies are announcing plans to enter space computing, and the sector is attracting intense attention. What do you think is the biggest misconception about space computing today? Which ideas are being overhyped?
Zhang Shijie: The attention is actually a good thing. It demonstrates that more resources will be invested in developing this industrial chain.
At present, everyone has their own interpretation, and no one can determine which path is best. We should let a hundred flowers bloom and a hundred schools of thought contend, making full use of the capabilities of industry—particularly the capabilities of China’s terrestrial industrial system.
I also do not believe the sector is overheated. The real question is whether people are genuinely committing resources.
One of the biggest problems facing China’s space industry in the past was that it was too detached from ordinary industry and had relatively limited participation. The opening and development of commercial space over the past several years has allowed the broader terrestrial industrial system to support and participate in the sector. That is the greatest positive development.
Hello Space: Neither the industry nor the general public yet has a clear concept or definition of space computing. Could you define it in one or two sentences?
Zhang Shijie: There is no need to establish a single definition of space computing at this stage.
In my view, any approach that increases intelligence or digital capabilities can be described as space computing. The concept should be broad enough to accommodate many different forms.
Hello Space: Once space computing infrastructure has been established, who will be its largest customers, and what will its principal business models be?
Zhang Shijie: We will only know where the applications are once the infrastructure exists.
Anything involving data processing or AI applications could potentially become an application of space computing.
Hello Space: There has been relatively little discussion of data storage in space. How do you think space-generated data will be stored in the future? Will it be transmitted to Earth, or will some data no longer need to be retained after processing? What form will space-based storage take?
Zhang Shijie: Storage is indispensable.
The need for data storage in space is not new. Conventional remote-sensing satellites and other spacecraft have always carried a certain amount of storage capacity. However, today’s demand for large space-based data centers places much greater requirements on that capacity.
Storage could take many different forms. Whether a dedicated satellite should be used for storage or storage media should be distributed across multiple satellites is an engineering choice between different technical approaches. There is no need to pursue a single storage model.
Different individuals and companies will adopt different models to create commercially viable systems.
Hello Space: GalaxySpace is also developing the Mini-Spider low-Earth-orbit broadband communications test constellation. In your view, what is the greatest difference between a constellation management system and a conventional software platform for managing an individual satellite? What challenges will constellation management face?
Zhang Shijie: A constellation management system is an open, large-scale system of systems.
Managing a system of systems requires greater robustness and more management elements, while the scale itself creates additional complexity.
The second consideration is the challenge presented by space itself. Terrestrial networks are generally designed on the assumption of persistent connectivity and high reliability. Space network topologies, by contrast, are dynamic, and unpredictable interruptions can create management challenges.
The inherent complexity of such systems, combined with the unique characteristics of the space environment, will create enormous challenges for constellation operating systems.
Hello Space: From the perspective of an industry researcher, which areas of the space computing value chain deserve the greatest attention?
Zhang Shijie: First, we need to examine how computing power can be increased.
From the chip perspective, whether developers use radiation-hardened components or adapt industrial-grade components for the space environment, the ultimate objective is to deliver greater computing power at an appropriate cost.
Second is interconnection and communications. Space computing requires supporting infrastructure, including both energy and networks. How these capabilities can be provided in space deserves close attention.
Third are energy and thermal management. Thermal dissipation, like energy generation, is linked to the available surface area: a given area can provide only a corresponding amount of heat dissipation.
The underlying physical limitations cannot be eliminated, but it is still worth exploring whether better engineering solutions can be developed.
Fourth are software systems. Hardware–software coordination and the eventual development of user applications are also important areas to watch.
Hello Space: Will delivering continuous, around-the-clock services from space-based data centers be the greatest challenge?
Zhang Shijie: From one perspective, it is the greatest challenge.
Providing services continuously under all operating conditions is much more difficult than operating for a period and then pausing.
It is technically achievable from an engineering perspective. The fundamental question is how much it will cost to provide that capability.
Perspective
This interview prompted us to reconsider a fundamental question. While the industry is still debating what form space computing should take, the people actually building these systems have already broken the challenge down into executable modules.
The first step is to define the mission. Only then should the required computing, energy and communications capabilities be determined. This was the logic Zhang repeatedly emphasized throughout the interview.
In our view, instead of rushing to impose a fixed definition on space computing, the industry should allow different technological approaches to develop. Some companies are working on chips, some are building constellations, and others are developing energy systems.
Those that can create a fully operational system while keeping costs under control will have the greatest opportunity to define the industry’s next stage.
This may be the most accurate description of the space computing industry today: the direction is clear, but the path is still being explored.
