EON raises $10.75M to link AI data centers with orbital lasers

EON is a reminder that AGI‑era infrastructure is not just about GPUs and data centers; it is also about how you move bits between them. Training and serving frontier models increasingly involve shuttling enormous volumes of data between geographically distant sites. Undersea fiber is fast but fragile, hard to repair and concentrated in a small number of chokepoints. A modest constellation of high‑throughput laser satellites that can give hyperscalers dedicated, controllable links between regions is strategically attractive.

The seed round is small by AI standards, but the investor roster and target customers are telling. General Catalyst and Andreessen Horowitz are betting that AI and cloud providers will pay for lower‑latency, redundant routes on underserved or politically sensitive paths, such as Africa to South America or Europe to Australia. For AI labs, such links could reduce training wall times, smooth replication of massive datasets and improve reliability of cross‑region inference.

In the race to AGI, better interconnects shorten the distance between distributed compute islands and make global training runs more practical. They also concentrate more of the stack in a handful of highly specialized vendors. If EON or rivals succeed, backbone connectivity itself could become another strategic bottleneck and another lever for states that want to influence where and how AGI‑class models are trained.