Leveraging AI computing power to secure commercial aerospace orders across more than 40 countries, this Hangzhou-based company has stepped into the g

钱塘出海2026-11-16 09:06
SpaceWill launches the "Space String" initiative to deploy a global space computing power network.

Introduction:

Turning their sights to the global market to explore the second growth curve has become an established direction, but how to expand overseas still tests the wisdom and courage of every Chinese enterprise. Brand globalization, supply chain globalization, cross-border e-commerce... there are numerous paths, yet no one-size-fits-all solution; North America, Southeast Asia, the Middle East, Latin America... the markets are vast, each with its own set of storms and challenges. Behind every expedition lies the dedicated exploration of Chinese enterprises; on every distant coastline, more and more figures from China are emerging.

The "Overseas Trailblazers" column aims to find benchmark enterprises that have achieved outstanding results in global expansion across different sectors, dissect practical cases from a professional perspective, precipitate methodologies from consensus, and discover new possibilities amid differences.

During the visit of Haitham, Sultan of Oman, to China in early November, Oman's Lance Space signed an agreement with a Hangzhou-based space computing company for an order of 20 satellites. This is the largest satellite export order in China's commercial aerospace history to date, and also marks the first full-scale overseas deployment of a Chinese computing constellation. 

This company is SpaceWill, founded in 2021, which is dedicated to building a space computing network and providing services including space computing systems, intelligent satellites, on-orbit constellation operation, as well as artificial intelligence and big data applications. At present, its business has expanded to more than 40 countries around the world, making it China's first private commercial aerospace enterprise to achieve overseas delivery of satellites and ground systems; in 2025, the company's revenue reached 130 million yuan, with overseas markets accounting for more than half of the total revenue and turning profitable. 

According to Song Bingchen, co-founder of Eastern Starlink, SpaceWill has set its goal to build a space agent integrating chips, algorithms and models, "to become a space computing node that can think, make decisions and evolve."

Developing Chips, Algorithms and Models

Integrated Space Intelligent Agent

01

On August 5, 2026, the computing constellation jointly promoted by SpaceWill and Eastern Starlink welcomed its first batch of inference verification satellites: "Eastern Starlink 05/06" (Eastern Eye Hyperspectral AI Dual Satellites), which were launched into orbit by the JL-3 Y12 carrier rocket. The satellites are equipped with SpaceWill's self-developed "Xian" AI on-board computing platform.

In August 2023, the "SpaceWill Intelligent Emergency No.1" satellite carrying the first generation of the "Xian" platform was launched into orbit, with a computing power of 60TOPS at that time. It realized real-time on-orbit observation and processing as well as space intelligence through more than 100 algorithms, filtered out invalid data, and then transmitted the calculation results back to the ground. Taking satellite images covering 10,000 square kilometers as an example, traditional satellites need 180 days to process the data, while intelligent satellites only need a few hours of on-orbit processing to complete dynamic target extraction. 

In 2024, the computing power of the "Xian" platform was further increased to 400TOPS. However, to raise the computing power of a single satellite to above 400TOPS, it is necessary to systematically solve the engineering problems of increased energy consumption and heat dissipation caused by synchronization. 

In the vacuum environment of space, chips can only dissipate heat through radiation. Therefore, SpaceWill independently developed high thermal conductivity composite materials, which increase the thermal conductivity to several times that of aluminum alloy, and are used in conjunction with deployable radiators. Song Bingchen revealed that in the future, SpaceWill will also introduce self-developed heat pump technology to further improve the heat dissipation capacity of space servers.

The founder of SpaceWill once said frankly, "At present, we are not on the same level as SpaceX in terms of rocket technology, mass satellite production capacity, or capital scale." 

To achieve its blueprint of a thousand satellites, SpaceWill plans to make continuous breakthroughs in three technical routes: first, to develop radiation-resistant chips with domestic manufacturers, and it is expected to deploy the world's first reconfigurable chip with Qingwei Intelligence by the end of this year; second, to explore new materials such as perovskite to improve the conversion efficiency of solar panels; third, to deploy the first domestic space liquid-cooled intelligent agent next year to optimize the heat dissipation performance of satellites. 

Remote Sensing First, Then Computing

From Regional Commercialization to Global Commercialization

02

As of June 2026, the 6 satellites of the two companies have a maximum cumulative on-orbit time of more than 2600 days, and a large amount of data on the actual performance of chips in the space radiation environment has been accumulated.

In the past few years, the domestic space computing power market has not yet achieved a commercial closed loop as a whole. Therefore, SpaceWill chose to make money and accumulate customers and trust through mature remote sensing data services first, and then gradually cultivate and superimpose space computing capabilities. Due to the limited demand in the domestic remote sensing satellite market, and the fact that the team has a group of talents from ZTE with more than ten years of overseas market experience, the overseas market has become the top priority for SpaceWill to realize marketization. 

"Many countries in the Middle East, Central Asia and Southeast Asia are also eager to build their own aerospace capabilities and space sovereignty." Song Bingchen said, SpaceWill adopts the combination of "whole satellite export + data service + talent training" to enter the niche market of the third world countries, which is ignored by the traditional European and American aerospace systems, but has strong willingness to pay and relatively less intense competition.

On November 11, 2024, SpaceWill delivered the first "OmanSat" to Oman. The main payload of this intelligent remote sensing satellite is an optical camera with a panchromatic resolution of 1 meter and a multispectral resolution of 4 meters. It has the capability of continuous push-broom imaging for up to 45 minutes, and can be applied to fields such as national land detailed survey, urban planning, forest survey, and disaster monitoring. 

Selfie of OmanSat in space

After two months of on-orbit operation, "OmanSat" has flown more than 1200 circles around the earth, completed more than 300 measurement and control tasks and more than 100 data transmission tasks, received about 3000GB of data, and successfully realized the first space practice verification of on-orbit AI recognition of ships and aircraft, space-based distributed operating system and space-based remote sensing large model, as well as satellite-ground network interconnection.

"No private company in China has ever done satellite export before. We are not wading across the river by feeling the stones, but we can't even find the stones." From contacting customers, demand docking, scheme design, satellite development, launch coordination, on-orbit testing to the final successful launch of "OmanSat", SpaceWill spent nearly two years, running through the whole process from satellite export approval, on-orbit delivery to foreign exchange settlement. "It is equivalent to opening up a commercial road that did not exist before." 

The success of "OmanSat" has been written into middle school textbooks in Oman, and it has also become a stepping stone for SpaceWill to open up more overseas markets. Since then, its business has expanded to more than 40 countries including Indonesia, Uzbekistan, Turkey, as well as southern Africa and Central Asia. 

In the process of serving more sovereign countries, SpaceWill provides product combinations that meet local needs according to the differentiated demands of different continents and regions. For example, Middle Eastern countries have relatively sufficient funds, pay more attention to sovereign capabilities, and need whole satellite export and talent training; Southeast Asian countries focus on livelihood applications such as agriculture and fisheries, and mainly purchase data and services; Africa attaches importance to disaster early warning and resource management. 

Song Bingchen concluded that the core competitiveness behind the differentiated service of "satellite going global" lies in three points: technical strength is the admission ticket, delivery capacity is the basis of trust, and long-term operation builds trust relationships.

Compliance is the premise and lifeline of all of the above. "The particularity of this industry determines that we would rather lose customers than cross the red line." Song Bingchen said that SpaceWill carries out a complete set of internal training from post training, delivery internal control to audit specifications, all services go through formal national approval, and do not involve sensitive fields such as military affairs. 

Build the Strategic Framework of "Remote Sensing + Computing + Communication"

Create the Future "Space Store"

03

As the company's cash flow business, remote sensing data services support the company's operation and customer accumulation, and it is also the first step for SpaceWill to realize its final strategic framework of "remote sensing + computing + communication".

According to Song Bingchen, SpaceWill plans to combine remote sensing satellites and computing satellites into one in the next step to form a sensing and computing integrated satellite, use space-based data to test and verify the value of the computing platform, and gradually form the core product of the space computing platform. 

For the space computing constellation to operate, communication capability is a crucial prerequisite. In order to achieve full-link capability autonomy, SpaceWill has obtained communication licenses in overseas markets and started to build its own communication network. The "Xian" platform itself already has functions such as narrowband communication and networked distributed computing. 

"Our ultimate goal is to form a space capability with full-link independent controllability of "remote sensing + computing + communication" and three constellations complementing each other, which is the underlying guarantee for us to advance the thousand-satellite blueprint at our own pace." Song Bingchen said. 

The prerequisite for the space computing platform to achieve a commercial closed loop is that the comprehensive cost of space is lower than that of the ground, so that customers will be willing to buy computing power in the sky. This is also one of the reasons why SpaceWill chooses to expand overseas first and then enter the domestic market. 

"The ground computing power cost of these overseas countries is much higher than that in China, so the reversal of the cost difference will come faster." Song Bingchen said that the SpaceWill team judges that it will take at least 7-10 years for space computing to achieve a global commercial closed loop, "We may achieve it first in the Middle East and countries along the Belt and Road, around 2031." 

This time judgment is not much different from the cost modeling calculation of SpaceX's orbital data center by many international investment banks. A calculation by Deutsche Bank in July 2026 shows that the current deployment cost of space data centers is about 6 times that of the ground. By the SpaceX AI 2 satellite phase in 2032, the launch cost will drop to 170 US dollars per kilogram, and the cost multiple is expected to become 0.6 times. Morgan Stanley also predicts that SpaceX's orbital computing cost will drop to 6.5 US dollars per watt per year by 2031, which is lower than the 6.8 US dollars of ground data centers. 

Song Bingchen believes that there are two main variables that ultimately determine the time for the commercial closed loop of space computing. The first is the cost of rocket launch. At present, the comprehensive cost of space computing power is about 8 times that of the ground, a large part of which comes from launch costs. The second is technological iteration, including chip radiation resistance, energy efficiency, heat dissipation efficiency, and networking scale. 

Before the commercial closed loop is realized, the technical closed loop must be completed first, that is, satellites with computing power above 10P can work full-time in space, receive data at any time, maintain stable reasoning and return results at any time, "as stable as ground data centers. From the current technical point of view, this is not yet realized, and completely replacing ground data centers is also a false proposition in business." Song Bingchen said. 

He described the specific product form to realize the commercial closed loop of space computing power as a "space store": after the satellite is in orbit, new algorithms can be continuously loaded, deleted and uploaded like a mobile phone, and different customers can call the computing power of the same satellite on demand. 

This capability has been verified to a certain extent in SpaceWill's past products. In 2025, SpaceWill cooperated with Turkey to dynamically deploy algorithms on the satellite, and successively completed three different tasks of agricultural detection, urban planning and disaster assessment on the same satellite. 

* All opinions contained herein are from the interviewed institutions or individuals, for reference only. 

This article is from the WeChat official account "Zhejiang Enterprises Overseas Comprehensive Service Port".