Humanity has never created and consumed as much data as it does now. Each search, each conversation with artificial intelligence, each photo stored on a cloud platform contributes to enlarging the global information volume. The explosion of AI, especially generative systems capable of writing, drawing and analyzing like humans, is opening up new experiences, making work and life more convenient. Users can ask AI to summarize documents, plan or create content in just a few seconds.
However, behind that convenience is a giant infrastructure operating silently. These are data centers with tens of thousands of servers, consuming huge amounts of electricity and water to maintain operation. As the demand for training and operating AI models increases sharply, data centers are increasingly expanding in scale, leading to pressure on power grids, water resources and the environment.
In many parts of the world, residents are beginning to clearly feel the changes. Forests, fields or quiet residential areas are gradually giving way to high-rise concrete blocks, emitting noise from cooling systems and gas turbines. Electricity bills tend to increase, while water sources are competing between agricultural production and industrial technology.
Amidst that context, a new direction is put forward: If the Earth is overloaded, can the human data warehouse be moved out of orbit? The idea of a space data center, once considered fictional, is gradually becoming a serious proposal in the technology race between the US and China.
A big and complex problem
The data center is the heart of the digital world. Inside these seemingly discreet buildings are server systems, network equipment and storage drives that operate continuously to process information. From email, social networks to advanced AI platforms, all depend on the rows of computers placed in a space that tightly controls temperature and humidity.
Each AI model needs to go through two stages that consume a lot of energy. The training stage requires tens of thousands of graphics processors to work together, "swallowing" huge amounts of data to optimize parameters. After completing training, the model enters the inference stage, responding to user requests. Each word entered must be analyzed in relation to the entire set of learned data, causing electricity consumption to increase with the volume of interaction.
Most data centers depend on local power grids. In many countries, power still mainly comes from fossil fuels, leading to greenhouse gas emissions. Some companies seek to build their own power sources with gas turbines or backup diesel generators when the power grid does not meet demand. This increases air and noise pollution to the surrounding community.
Not only electricity, water is also a key factor. High-intensity servers generate large heat, forcing data centers to use water cooling systems. Water absorbs heat and is put on the cooling tower and part evaporates during this process. Direct water consumption is the loss through evaporation, and no one can be sure when or where it will return to its natural cycle. In drought-stricken areas, this becomes a real concern.
Environmental impacts also extend to the supply chain. AI processors use many rare metals, the mining and production process consumes resources and can cause environmental degradation. Meanwhile, the expansion of data centers requires upgrading the power grid, and this cost is sometimes transferred to people in the form of increased electricity bills.
Faced with the development speed of AI, many experts worry that energy demand is exceeding the expansion capacity of renewable energy. As companies build new centers at a rapid pace, they tend to rely on available fossil fuels instead of waiting for green infrastructure to be completed. This makes the emission spiral continue to expand.
The human data warehouse is therefore not only a symbol of technological progress, but also a complex environmental problem. When AI brings new experiences to users, the price to pay lies in huge buildings that consume electricity and water, silently operating to meet each requirement in the digital world.
Solutions from outside Earth
When ground data centers faced limitations in energy, water and space, the idea of putting them out of orbit began to be seriously discussed. Many technology companies in the US and China expressed interest in building orbital data centers, taking advantage of abundant solar energy and the ability to dissipate heat in the space environment.
At a global economic forum, technology billionaire Elon Musk said that placing energy-consuming data centers in space is a reasonable choice, because continuous sunlight and natural cooling conditions can help reduce operating costs. Companies involved in space and AI programs are considering the possibility of developing satellites integrating computing and storage functions.
China is also implementing a plan to build a supercomputer in space, considering it a new front in the technological competition with the US. Some private enterprises are participating, aiming to complete the orbital infrastructure in the next few years. This idea is seen as a leap forward, similar to how mobile phones once helped many countries skip the landline phase.
In theory, an extra-orbital data center can reduce pressure on power grids and water resources on Earth. If AI training that consumes a lot of energy is transferred to space, countries with overloaded power systems can reduce the burden. However, this solution opens up new questions about sovereignty and governance.
When citizen data is processed outside the territory, who will take control? The country where the data is created, the satellite launching country, or the company operating the orbital data center? Without a multilateral cooperation framework, new infrastructure can consolidate the dominance of large corporations instead of narrowing the digital gap.
For many developing countries, the lack of capacity to participate in space infrastructure ownership may make them the only data provider, while the decision-making power lies elsewhere. The human data warehouse then not only leaves Earth physically, but also deviates from the sphere of influence of many governments.
The idea of sending a data center into space shows the urgency of the environmental problem posed by AI. It reflects a reality that technology is progressing faster than the regulatory capabilities of existing laws and infrastructure. If managed transparently and fairly, orbital infrastructure could become an opportunity to reduce the burden on the planet. Conversely, it could create a new layer in the global digital order, where access to artificial intelligence depends on spatial access.
