Satellite Lasers Take Aim at Undersea Fiber's Monopoly on Global Data Traffic

The Infrastructure Tension Nobody Talks About

Every time a hyperscaler provisions capacity in Mumbai or São Paulo, someone has to figure out how the bits get there. The answer, for decades, has been undersea fiber: thousands of kilometers of glass cable laid on the ocean floor, expensive to install, slow to repair, and increasingly inadequate for the volume of data AI workloads demand. Endeavor Optical Networks thinks the solution isn't another cable, but a constellation of satellites equipped with optical communications terminals capable of beaming 2.4 terabits per second between ground stations on different continents.

The company emerged from stealth in early August with $10.75 million in seed funding led by General Catalyst and Andreessen Horowitz. CEO Charlie Horowitz and CTO Tyler Presser are betting that advances in laser technology and satellite power systems have finally made space-to-ground optical links viable at the throughput levels data center operators require.

At DailyTechWire, we've tracked the slow maturation of optical satellite communications for years. What makes EON's plan notable isn't the technology itself, which NASA and a handful of private operators have already demonstrated in limited form, but the target performance. Most optical links demonstrated to date have achieved around 2.5 gigabits per second. EON is aiming for nearly 1,000 times that figure in its initial deployment.

Why Fiber Dominates and Where It Fails

Modern submarine cables can carry 200 terabits per second or more across ocean basins, an order of magnitude beyond what any satellite system has achieved. That capacity, combined with latency measured in tens of milliseconds, makes fiber the default choice for intercontinental data transit. Radio frequency satellite systems can't compete on bandwidth; even the most capable broadband constellations max out in the low gigabit range per beam.

But fiber has structural weaknesses that become more visible as data center footprints expand into secondary markets. Cable routes are geographically constrained, repair cycles can stretch into weeks, and certain corridors remain underserved because the economics don't justify new deployments. A link between France and Australia, for instance, might traverse multiple cable systems with different operators and pricing structures. Routes connecting Africa and South America face similar challenges, with limited direct infrastructure.

EON is targeting exactly those gaps. The company plans to sell dedicated capacity on specific routes, giving hyperscalers and AI labs full control over their transit without the capital expense of laying new cable or the operational risk of sharing capacity on congested systems.

The Atmospheric Problem

Laser communications through space face a fundamental challenge: Earth's atmosphere. Water vapor, aerosols, and cloud cover all distort and attenuate optical signals. This phenomenon, known as atmospheric turbulence, is the same effect that makes stars twinkle. For a communications link carrying terabits of data, any signal degradation translates directly into packet loss and retransmission overhead.

Horowitz says EON has developed techniques to mitigate this problem, though the company isn't disclosing specifics. The strategy involves careful site selection for ground stations, leveraging real-time weather data to route traffic through the clearest atmospheric windows, and deploying redundant terminals in each region to ensure link availability.

The initial constellation will consist of about 20 satellites, each providing a dedicated link between two continents. That architecture differs from mesh networks like Starlink, which route traffic dynamically across hundreds or thousands of satellites. EON's approach prioritizes link quality and predictability over coverage area, a trade-off that makes sense when your customers are data centers with fixed locations rather than mobile users.

Building the Terminal

Most of the technical risk in EON's plan centers on the optical communications terminal itself. The company intends to purchase satellite buses, the platforms that provide power, propulsion, and attitude control, from vendors like Apex Space. That allows EON to focus engineering resources on the payload: the lasers, gimbals, and signal processing systems that will maintain a stable link while the satellite moves at orbital velocity.

Horowitz previously worked at Apex, first as chief of staff to CEO Ian Cinnamon and later as director of special projects. That background gives EON access to proven bus designs and a supply chain for the commodity components of the spacecraft. The challenge is in the exquisite systems, particularly the gimbals that must point the laser with sub-microradian precision while compensating for satellite motion and atmospheric distortion.

The company's technical team includes Tyler Presser, who holds a PhD in astronautical engineering and has worked on mission design at NASA, along with Michael David Francois, a former Google executive who spent years managing global network infrastructure. Wesley Baxter, an optics engineer who worked on Amazon's Project Kuiper satellite network, rounds out the core team.

The Demo and the Market

EON plans to use its seed funding to establish an optics laboratory, expand its engineering headcount, and conduct ground testing before launching a demonstration satellite in late 2027. That spacecraft is expected to achieve at least 800 gigabits per second of optical downlink throughput, and potentially a full terabit, which would represent the highest performance yet demonstrated for a space-to-ground optical link.

Jeannette zu Fürstenburg, president and managing partner at General Catalyst, framed the investment around two themes: artificial intelligence infrastructure and supply chain resilience. She told reporters that demand isn't the constraint. The real question is execution, whether the team can deliver the promised performance on the timeline they've outlined.

That skepticism is well-founded. Blue Origin has announced plans for TeraWave, a 5,048-satellite constellation targeting up to 6 terabits per second for large-scale users. Blue's plan is more ambitious but also more capital-intensive and further from deployment. EON's smaller fleet should reach orbit faster, but both companies face the same physics: maintaining link quality through a turbulent atmosphere while moving data at speeds data center operators will trust.

Industry analysts point out that satellite connectivity has historically been a last-resort option, chosen only when terrestrial alternatives don't exist. Data centers, in particular, have stringent requirements for uptime, latency, and predictable performance. Convincing them to route production traffic over a satellite link, even one with terabit throughput, will require sustained proof of reliability.

Where This Fits in the Stack

EON's focus on existing, terrestrial data center customers distinguishes it from the more speculative idea of building compute infrastructure in space itself. Horowitz says the company has a rule: no physics problems. The market they're addressing exists today, with identifiable customers moving data over routes that are expensive, slow to provision, or lacking redundancy.

That pragmatism aligns with the broader trend we're seeing in infrastructure investment. As AI training runs scale and inference workloads distribute globally, the cost and performance of inter-region data transit becomes a first-order concern. Hyperscalers are already investing in private fiber routes and peering arrangements to optimize their network topology. A satellite link that offers comparable throughput to fiber, with faster deployment and route flexibility, fits neatly into that calculus, assuming the reliability and cost structure work.

The question isn't whether optical satellite links are technically feasible. NASA and several private operators have already proven the concept. The question is whether they can scale to the performance and reliability levels that data center operators demand, and whether the economics justify the capital and operational expense. EON's bet is that the answer is yes, at least for the subset of routes where fiber is weak. If they're right, the next generation of intercontinental data infrastructure might bypass the ocean floor entirely.