Endeavor Optical Networks emerged from stealth on August 4 with $10.75 million in seed funding from General Catalyst and Andreessen Horowitz, and a plan to move hyperscaler traffic off undersea fiber and onto a mesh of laser-equipped satellites. The startup, founded in May and known as EON, is targeting 2.4 terabits per second of throughput per link between continents — roughly 1,000 times the 2.5 Gbps ceiling that prior space-to-ground laser demonstrations have hit. Co-founders Charlie Horowitz and Tyler Presser want a demo satellite in orbit by the end of 2027.
The pitch takes direct aim at a structural bottleneck in AI infrastructure. Hyperscalers moving training data and model weights between global data centers rely on undersea fiber that carries 200 terabits per second but is expensive to install and slow to repair. Radio-frequency satellite links can't match that bandwidth, which has left orbital comms as a technology of last resort. EON is betting that optical terminals, better satellite buses, and careful ground-station siting have finally closed enough of the gap to sell dedicated capacity to AI labs and cloud providers.
EON plans an initial fleet of about 20 satellites, each dedicated to a two-continent link with 24-hour coverage. The company is prioritizing routes that are either lengthy or under-served — France to Australia, Africa to South America — where fiber alternatives are scarce or expensive. Ground stations will be sited redundantly and switched based on weather data, since atmospheric distortion, especially cloud cover, is the primary technical enemy of a laser downlink.
Key facts
- 01EON raised $10.75M in seed funding from General Catalyst and Andreessen Horowitz, emerging from stealth on August 4, 2026.
- 02The startup targets 2.4 terabits per second per satellite link, roughly 1,000x the 2.5 Gbps demonstrated by prior space-to-ground laser systems.
- 03EON plans about 20 satellites providing 24-hour coverage between continents, with a demo launch targeted for the end of 2027.
- 04Blue Origin's competing TeraWave network plans 5,048 satellites and up to 6 Tbps speeds, but on a longer deployment timeline.
- 05The demo satellite is expected to deliver at least 800 gigabits and possibly a full terabit of optical downlink throughput.
The seed round will go toward an optics lab, engineering hires, and ground tests before that 2027 launch. Horowitz expects the demo satellite to hit at least 800 gigabits per second of optical downlink and possibly a full terabit — numbers that would beat any laser link demonstrated to date. To get there, EON is concentrating in-house engineering on the optical communications terminal itself, particularly the precision gimbals that steer the laser, while buying off-the-shelf satellite buses from suppliers including Apex Space.
The team's résumé is space-industry heavy. Horowitz was previously chief of staff to Apex Space CEO Ian Cinnamon and later Apex's director of special projects. Presser is a PhD astronautical engineer who worked on frontier missions at NASA. The technical bench also includes Michael David Francois, a longtime Google executive focused on global network infrastructure, and Wesley Baxter, an optics engineer who most recently worked on Amazon's low-Earth-orbit satellite network.
General Catalyst's Jeannette zu Fürstenburg, who led the round, framed the bet as a convergence of the fund's AI and resilience theses. The unspoken subtext: AI workloads are pushing terrestrial infrastructure harder than any prior application, and a laser mesh that bypasses undersea choke points is a hedge against both physical vulnerability and capacity constraints.
Competition is already visible. Blue Origin, Jeff Bezos's space company, has announced TeraWave, a planned 5,048-satellite constellation targeting up to 6 Tbps for large customers. That network is more ambitious in scale and speed, but also further from launch. EON's argument is that a smaller fleet with a narrower use case — point-to-point data-center connectivity, not consumer broadband — can be deployed faster and reach paying customers sooner.
Skeptics point out that hyperscaler standards are the hard part, not the physics demo. Caleb Henry, director of research at Quilty Space, notes that data-center customers demand quality and redundancy that satellite internet has only recently begun to approach, and that optimizing spacecraft for that market will take longer than founders typically project. EON's 2027 timeline gives it roughly 16 months to build hardware, prove the atmospheric-correction stack, and secure enough customer commitments to justify the follow-on capital that a 20-satellite constellation will require.
If EON hits its numbers, the story rewires part of the economics of AI infrastructure. Training runs and cross-region inference already push the limits of what fiber can carry cost-effectively, and cable-cutting incidents in the Baltic and Red Sea over the past two years have made hyperscalers more attentive to route diversity. A terabit-class orbital link priced as dedicated capacity is a real product for a real buyer, and it's a category where the incumbents are geography-bound fiber operators rather than deep-pocketed cloud rivals. The bet is whether a $10.75 million seed can survive long enough to reach the first paying customer — and whether Blue Origin's bigger constellation stays on the drawing board long enough for EON to book the routes that matter.
Working on something we should cover, or seeing a story we missed? Send leads, documents, or feedback to hello@aichatdaily.com. For sensitive tips, see our secure tips page for Signal and PGP options.
Spotted an error? Email hello@aichatdaily.com with the URL and the issue, or read our full corrections policy.
