EON's orbital lasers offer geographic redundancy, not a replacement for undersea cables

Intercontinental data doesn't ride the cloud. It rides a few hundred thin glass threads on the ocean floor, and that fragility is the actual infrastructure story of the AI era. EON's pitch starts from that fact, and its $10.75M seed from General Catalyst and Andreessen Horowitz reads as a bet on the gap between cable fragility and AI-scale demand, not as a moonshot.

The 2.4 Tbps per-link goal Charlie Horowitz's team has set is the wrong number to fixate on. It is a target, not a result, and current space-to-ground laser demos top out near 2.5 Gbps. The order-of-magnitude jump is the real bet, and the company has zero satellites in orbit.

EON's actual product is geographic redundancy: about 20 satellites, each linking two continents, paired with redundant ground stations in different climate regions, with weather data used to route around clouds. The underserved routes, France to Australia, Africa to South America, are the routes the cable industry skipped because the demand curve was not there. AI training runs and hyperscale inference are making routes that were previously uneconomical viable.

The mechanism repeats whenever a critical dependency sits on a single physical layer. The mechanism earns EON its $10.75M only if the atmospheric-correction 'secret sauce' holds in cloud cover on those routes, and only if hyperscalers pay for insurance they have not historically bought.