Space-based Data Centers: Bringing AI Policy Into the Equation
Space-based data centers have attracted growing interest in recent years, prompting significant attention to the technical and policy challenges associated with their development. What’s less discussed is another policy question at the intersection of space infrastructure and artificial intelligence: as both technologies advance, how should AI policy be applied to data centers operating in space?
Shifting AI data centers into orbit can mitigate some of the concerns of centers on Earth, such as energy and water use. However, emerging questions over the management and transfer of data can become quite complex. Many countries and trade blocs have sought to regulate the AI industry by implementing rules on how data is stored, who can access the data, and what information can be fed into models. These regulations are ostensibly designed to address concerns over data privacy, consumer rights, and cybersecurity, but often risk impeding the development and deployment of AI models.
These rules risk fragmenting the global marketplace for AI as companies may be forced to silo their operations into specific jurisdictions, and are prevented from easily engaging in cross-border service delivery. Restrictions on the export of AI chips further fragment the market as companies struggle to fulfill dueling mandates from different countries, although some progress has been made to open avenues for the trade of AI chips between the U.S. and the EU.
The transfer of AI operations from terrestrial data centers into space would likely further complicate these matters. The current rules on AI are based under the assumption the data centers are on a stationary location within the jurisdiction of a governing body – but in the future data centers could be hubs on celestial bodies or on satellites orbiting the Earth at thousands of miles per hour, and may even operate without any human physically accessing the hardware. This raises several concerns. How will these data centers be maintained and protected, both physically and digitally? Today, bad actors can intercept data transmissions that could otherwise be sent through more secure fiber optic networks, and data may be accessed across borders as these satellites travel all over the world.
Then there is the question of jurisdiction – if a data center is not located within the physical boundaries of a governing state, how might that change how laws and regulations are enforced? For example, the EU restricts the export of personal data based on the privacy protections available in destination jurisdictions: will it recognize the country licensing a satellite as being the relevant jurisdiction, for the purpose of its data export regime?
The Outer Space Treaty requires all non-governmental space operations to be authorized and continuously supervised by a member State, which is usually interpreted as the member state that an operator is registered in. While the U.S. has a licensing regime that covers launch, spectrum use, and remote sensing, there are no specific regulations on other “novel” space activities, such as AI data centers. Several frameworks to address this gap have been introduced, such as the mission authorization framework proposed by the Office of Space Commerce in response to a recent Executive Order on the issue. With the lack of space-specific regulations or laws, the general assumption in the U.S. has been that any rules that apply to a terrestrial activity would also apply to the equivalent space activity. However, many of those rules are not practically applicable or enforceable to space technology. Other countries may also take a different legal approach to authorization and supervision to expand or contract their regulatory jurisdiction on space activities, which could create regulatory uncertainty for companies that operate in international markets. Then there’s the risk that companies may seek to fly under a flag of convenience in order to operate under a more favorable regulatory environment, as seen in global shipping.
Further complicating the issue is the fallout from the Supreme Court’s ruling on the Loper Bright case, which limits the ability of federal agencies to craft regulations without explicit Congressional direction. This ruling has already had major implications for the commercial space industry, as evidenced by the recent bipartisan letter from the House Science Committee leaders to the FCC regarding their regulatory authority on space safety and traffic. In the absence of clear rules on space activities, companies may be forced to go through unnecessary regulatory hurdles to reach compliance, and regulators may lack the expertise to provide adequate oversight to space activities.
Congress is likely to introduce commercial space legislation this year that should address some of the regulatory shortcomings. As they do so, they should consider language that:
- Prevents the use of flags of convenience for space activities.
- Explicitly states that space operators are expected to abide by the same rules and regulations that terrestrial operators follow where applicable.
- Provides mechanisms for regulators to create exceptions to rules and regulations that are not practicably enforceable for space operations.
- Delegates authority to a single regulatory body for novel space activities and provides them with enough flexibility to adopt and adapt rules as the industry evolves to reduce the need for repeated legislative action.
Additionally, the Administration should work with other major trade partners to harmonize regulatory frameworks on AI and space technology to reduce trade barriers. Orbital data centers have the potential to provide services to a wide range of customers across the globe, and the costs of manufacturing, launch, and operations often necessitate international partnerships.
By providing regulatory clarity and reducing international trade barriers, we can bring the power of AI to the final frontier.
