The Economist Predicts an Era of Space Data Centers

The launch of SpaceX's Starship in March this year will make space data centers a reality, according to journalists at The Economist.

The interest in orbital data centers stems from a crisis in terrestrial infrastructure. According to Sightline Climate, between 30% and 50% of the capacity planned for this year may be delayed due to issues with permits, grid connections, and local protests.

Several U.S. states have imposed construction moratoriums, while the demand for electricity for AI continues to rise.

One alternative could be placing servers in orbit, where unlimited solar energy is available. In November 2024, Starcloud launched the Starcloud-1 satellite, the size of a refrigerator, equipped with a standard Nvidia H100 chip. It successfully trained the NanoGPT language model on Shakespeare's works and launched Google's open model Gemma.

Challenges

The main obstacle remains the cost of launch. Currently, delivering a kilogram of cargo on Falcon 9 costs $3,400, while on Falcon Heavy it costs $1,500. Two other factors are also crucial:

• power-to-weight ratio — how many watts of computing power a kilogram of satellite provides;
• cost of the satellite — in dollars per watt.

These parameters directly depend on the mass and efficiency of solar panels and cooling systems.

Another concern is how radiation will affect the reliability of AI chips in orbit. To determine the feasibility of orbital data centers, assessments of all these variables are necessary, noted The Economist.

Andrew McCalip, an engineer at the space startup Varda, created a web calculator to compare the costs of orbital and terrestrial data centers of a given capacity.

According to his calculations, building a terrestrial data center with a capacity of 1 GW and operating it for five years will cost $15.9 billion. The orbital equivalent (assuming a launch cost of $500/kg, a power-to-weight ratio of 37 W/kg, a satellite cost of $22/W, and 98% sunlight availability) is estimated at $51.1 billion. Neither sum includes the cost of the AI chips (GPUs), which will range from $15 billion to $30 billion and will be the same for both options.

Solutions

Starcloud, founded in 2024 specifically to realize the idea of orbital computing, has already gathered initial practical data. The experiment with the Starcloud-1 satellite based on the Nvidia H100 provided insights into chip performance in space and helped refine key parameters for future calculations.

The power-to-weight ratio of 37 W/kg used in the basic calculations is derived from Starlink satellites. However, AI satellites do not require heavy antennas for communication with Earth — only laser links with nearby satellites.

This means a larger portion of the mass can be dedicated directly to computing. SpaceX CEO Elon Musk believes a power-to-weight ratio of 100 W/kg is achievable, and potentially even 150 W/kg in the future. Starcloud is conservatively planning for 70 W/kg.

A similar situation exists with satellite costs. Starlink's cost is around $22/W. An AI satellite, lacking expensive communication components, should be significantly cheaper.

Starcloud's CEO, Philip Johnston, set a target of "less than $5/W" (excluding GPUs). With these achievable values (70 W/kg and $5/W), the cost of an orbital data center drops to $16.7 billion — just 5% more than a terrestrial one.

However, the launch cost remains the key factor. The estimated $500/kg is about a third of current SpaceX expenses.

If SpaceX's new fully reusable Starship rocket becomes operational, the price could fall to $100-200/kg. In this scenario, the orbital center would cost $12.1 billion, making it cheaper than the terrestrial option.

The first Starcloud satellite overheated, but the second launch this year will test a radiator with ten times the cooling efficiency. Preliminary data also indicate that chips fail less frequently than expected: with a 5% failure rate instead of 9%, costs would decrease even further.

A key event for the entire industry will be the 12th test flight of Starship in March. Its success will determine whether an era of affordable launches will arrive and if orbital data centers will become a viable alternative to terrestrial ones.

Recall that in early February, Musk merged SpaceX and xAI into a single company valued at $1.25 trillion. The firm has applied for a license to create an orbital data center consisting of 1 million satellites.