SCALABILITY

The computational resources used to train notable AI models have grown at a rate of about 4–5× per year from 2010 to 2024. To support this growth, terrestrial data centers used for AI workloads will require purpose-built nuclear facilities as they approach gigawatt scales with ROI timelines stretching beyond 10 years.

StarBlades can be deployed and merged into a parent constellation, scaling overall compute capacity without power constraints.

As data centers grow in scale and frequency, the costs attributed to retrofitting latent infrastructure also scale. For example, U.S. ratepayers shoulder in excess of $10 billion annually for grid upgrades to support data centers, despite limited public benefits. This is expected to increase through 2030 with approximately $720 billion in grid expenditure in the U.S. alone.

A StarBlade constellation requires limited terrestrial infrastructure scaling without financial bottlenecks.

The time required to build new terrestrial data centers is also increasing. Moratoriums in countries such as Amsterdam, Dublin, and Singapore routinely block new data centers due to grid strain. It routinely takes four years or more to have high-capacity power lines extended for terrestrial data centers.

StarBlade constellations can be scaled without rate limits faced by terrestrial incumbents.