Is Dreame's space plan of launching 2 million computing power satellites reliable?
In the consumer electronics industry, cross - border development is no longer a novelty. From mobile phone manufacturers venturing into the automotive industry to internet companies manufacturing chips, the technological boundaries are constantly being broadened.
However, it still seems rather abrupt when a company that started with floor - cleaning robots suddenly announces a plan to build a computing power network in space.
On March 11th, at the AWE exhibition, "Xinji Chuanyue", an ecological enterprise of Dreame Technology, unveiled a rather science - fiction - like plan.
The company stated that it will launch the "Yaotai" series of computing power satellites and build a distributed computing network in low - Earth orbit to provide computing power support for large AI models.
Fu Haiyang, the person in charge, introduced on - site that the space environment is conducive to heat dissipation and energy acquisition. Deploying some computing nodes in orbit may relieve the energy consumption pressure on ground data centers.
On March 16th, at the Jiuquan Satellite Launch Center, the Kuaizhou - 11 Y7 carrier rocket was launched. The first "Yaotai" computing power base station under Dreame Xinji Chuanyue was successfully sent into space on the rocket. It will conduct systematic tests in a sun - synchronous orbit about 561 kilometers from the Earth.
What really shook the industry is the long - term scale of this plan: 2 million satellites.
This figure has almost no comparable counterpart in the aerospace field.
Since the first artificial satellite was launched in 1957, the total number of spacecraft launched globally is approximately over 20,000. Currently, the largest commercial satellite network is Starlink built by SpaceX, with a long - term planned scale of about 42,000 satellites.
If calculated according to the upper limit of the plan, the 2 million satellites proposed by Dreame are nearly 50 times the scale of Starlink and close to 100 times the total number of launches in the history of human spaceflight.
In an industry like the aerospace industry that highly depends on engineering reality, such a scale naturally raises a question: Is this a feasible path or an extreme technological narrative?
I. Three Realistic Thresholds in the Aerospace Industry
The limitations of aerospace engineering come from physical laws and international rules. Any large - scale constellation project must first cross three basic thresholds: orbital resources, space safety, and economic feasibility.
Orbits and Spectrums
Space is not a completely open space. The communication frequency bands and orbital positions in low - Earth orbit (LEO) need to be coordinated and allocated through the International Telecommunication Union.
The satellite communication resources follow the rule of "first - come, first - served". In recent years, several large - scale constellation projects have occupied a considerable proportion of orbital and frequency band resources. For example:
- SpaceX's Starlink constellation
- OneWeb communication network
- China's planned GW constellation
- The G60 Starlink promoted by Shanghai
The scales of these projects range from thousands to tens of thousands of satellites, and they have submitted relevant applications to the ITU.
If a new entrant hopes to deploy hundreds of thousands or even millions of satellites, it must first obtain the corresponding orbital and spectrum resources. Without international coordination and application procedures, it is difficult for any constellation plan to enter the engineering stage.
Orbital Capacity
Even if the orbital resources can be solved, the space environment itself has a capacity limit.
One of the long - term concerns in the aerospace community is the Kessler Syndrome. This theory points out that when the density of orbital objects is too high, the debris generated by satellite collisions may trigger a chain of impacts, eventually forming a debris cloud that is difficult to clear.
Currently, mainstream research generally believes that the safe capacity of low - Earth orbit is roughly in the order of hundreds of thousands of satellites. If the number continues to increase, anti - collision control and space debris management will become extremely complex.
In this context, a satellite constellation with a scale of millions of satellites poses requirements on the orbital traffic management system far beyond the current level.
Cost and Launch Capacity
Finally, there is the most direct engineering problem - cost.
The cost of a satellite with computing power processing capabilities, laser communication links, and a stable power supply system is much higher than that of an ordinary communication microsatellite. Even if the cost of a single satellite is compressed to the level of one million yuan through highly industrialized production, deploying millions of satellites still means a long - term investment in the trillions of yuan.
Launch capacity is also a limitation. Currently, the most promising heavy - lift vehicle is the SpaceX Starship. In theory, this system can deploy a large number of satellites at one time, but even so, it will still take an extremely long period to send millions of satellites into orbit.
Under the current technological conditions, this scale is more of a long - term concept rather than a short - term engineering plan.
II. The Technological Paradox of Space Computing Power
Deploying computing power in space is not a new idea.
In the aerospace and computing fields, some research institutions have indeed discussed similar concepts. The basic logic is that the space environment is close to a vacuum, with better heat - dissipation conditions than on the ground, and solar power can be directly utilized.
However, when this concept enters the engineering stage, some practical problems quickly emerge.
First is communication efficiency.
AI training relies on high - density parallel computing, and a high - speed interconnection network is required between a large number of servers. Ground data centers usually use high - speed fiber - optic networks with extremely low latency.
Even if laser links are used between satellites, their bandwidth and latency are still difficult to compare with ground networks. If training data needs to be frequently transmitted between the ground and orbit, the communication cost may offset the computing power advantage.
Second is radiation environment.
There is high - energy particle radiation in orbital space, which has a significant impact on the stability of electronic devices. Aerospace - grade chips usually require special design, and their performance density and energy efficiency are often lower than those of commercial GPUs or AI accelerator cards.
Third is equipment iteration cycle.
The update speed of AI hardware is extremely fast, and servers are usually updated every three to five years. Once a satellite is in orbit, the cost of maintenance and upgrade is extremely high. This difference in cycles means that space computing power nodes may quickly lag behind ground technology.
From an engineering perspective, space computing power is more suitable for performing edge computing tasks, such as remote - sensing data pre - processing or satellite network optimization. Moving a large - scale AI training center into orbit still lacks a practical path at present.
III. What the Real Players Are Doing
If we look at the entire industry, we will find that the paths taken by major participants are significantly different from Dreame's narrative.
One of the key prerequisites for Elon Musk to promote the Starlink plan is a complete aerospace system. SpaceX not only operates a satellite network but also continuously researches and develops reusable rockets and promotes the Starship project.
By reducing launch costs, SpaceX is building a high - frequency space transportation system. Launch capacity is the foundation of its business model.
Domestic technology companies adopt more conservative strategies. For example, Huawei has focused on data - center liquid - cooling technology and energy - efficiency management in recent years, reducing the energy - consumption cost of AI computing through engineering optimization.
In the commercial aerospace industry, companies such as LandSpace and Galactic Energy still invest their main resources in improving rocket launch capacity and frequency.
Overall, the main line of the industry is still the construction of basic capabilities.
IV. Technological Narrative and Capital Imagination
From a business perspective, such grand space plans often have another meaning.
Dreame started in the fields of floor - cleaning robots and cleaning appliances. In recent years, this market has been highly competitive, product prices have been continuously declining, and enterprises generally face growth pressure.
In the capital market, the valuation systems of different industries vary significantly. Traditional home - appliance enterprises usually use the manufacturing - industry valuation model, while enterprises involved in AI, chips, or aerospace technology may receive a higher technology premium.
Judging from the content of the press conference, Dreame is trying to enter multiple technological narratives: self - developed chips, robot platforms, autonomous - driving computing, and space computing power networks. These directions happen to be the most active areas in current technology investment.
In this context, a grand space plan can not only demonstrate the enterprise's technological ambition but also change the market's perception of the company's industry.
Conclusion
The aerospace industry has a characteristic: all grand visions will eventually be tested by engineering reality.
Each link, including rockets, satellite platforms, orbital resources, and long - term capital investment, requires several years of continuous technological accumulation.
Two million computing power satellites is undoubtedly an impactful figure and has successfully attracted market attention. However, in terms of key issues such as orbital resources, space safety, and engineering cost, this plan is still in a very early conceptual stage.
In the history of the aerospace industry's development, projects that truly change the industry pattern often come from long - term technological investment rather than large - scale press conferences.
For Dreame, what is more crucial than the space blueprint is still the construction of the ground - based technology system. The real implementation of capabilities such as chips, robot platforms, and intelligent computing power is the foundation that determines the enterprise's future position.
After all, it's easy to talk big, but it's another matter to establish a commercial closed - loop.
This article is from the WeChat official account "Star Movement Without Bounds", author: Unilym, published by 36Kr with authorization.