Apolink wants satellites to stop going dark

Here is a detail about the space industry that surprises most people outside it: the operators flying satellites usually cannot talk to them.

A satellite in low Earth orbit circles the planet roughly every 90 minutes, and it can only communicate when it happens to pass over a ground station. The rest of the orbit is silence. Telemetry piles up in onboard storage. New commands wait in a queue on the ground. If something goes wrong between passes, the operator finds out after the fact, once the satellite comes back around and downloads the bad news.

Apolink is building a relay network in orbit to close those gaps. Instead of waiting for a ground station to appear below, a satellite hands its data to one of Apolink's relay satellites overhead, which routes it back to Earth. The company, founded in 2024 and backed by Y Combinator in its Fall 2024 batch, raised an oversubscribed $4.3 million seed round at a $45 million post-money valuation to build it. And unlike most seed-stage space companies, it already has hardware overhead: its first demonstration satellite launched on a SpaceX rideshare in July 2026 and is talking to the ground today.

Most of an orbit is silence

The dead-zone problem is as old as spaceflight, and the traditional fix is to buy more ground stations. Networks like AWS Ground Station and KSAT sell passes over antennas spread around the world, and adding passes shrinks the gaps.

It never eliminates them. Antennas sit on land, oceans cover most of the planet, and every added ground contract adds cost and scheduling complexity. Operators learn to live inside the constraint. They buffer telemetry, delay tasking, and fly blind during anomalies, which is exactly the resignation Apolink's own homepage pushes against.

For some missions the gaps are an inconvenience. For others they defeat the point of the mission. A wildfire-detection satellite that spots a new burn over Australia but cannot report it for another 40 minutes has lost most of the value of spotting it early. The same is true for missile-warning payloads, disaster response, and any constellation whose customers expect data in something close to real time.

The classic answer to this was NASA's TDRS network, geostationary relay satellites that gave the Space Shuttle and the ISS continuous contact. TDRS is aging out of service, NASA has been steering missions toward commercial alternatives, and nothing aimed at ordinary commercial smallsats has filled the slot. That is the opening Apolink is aiming at.

A relay layer that works with satellites as they are

Apolink's plan is a constellation of 32 satellites carrying both radio and laser links, sitting above the traffic in low Earth orbit and relaying data between customer satellites and the ground. The company targets roughly 99 percent uptime, with latency of 10 to 15 seconds at first and 2 to 3 seconds once the full network is deployed. For operators used to waiting the better part of an orbit, that is a different product category, closer to a dial tone than a scheduled appointment.

The design decision that most defines the company is backward compatibility. Starlink and Kuiper both build optical inter-satellite links, but those links serve their own constellations, and connecting to a proprietary network generally means flying that network's terminal. Apolink's hybrid RF and optical architecture is deliberately hardware independent: a customer satellite connects over radio links it already has, with no dedicated user terminal to buy, integrate, and qualify.

That matters because the satellites that need relay most are already in orbit or already designed. A hardware requirement writes off the existing fleet and adds a year of integration work for everyone else. A network that meets satellites where they are can sell to the whole market, including spacecraft launched before Apolink existed.

Continuous contact also changes what operators can do, not just how fast they hear back. Tasking becomes something you do when the customer calls rather than when the schedule allows. An anomaly gets caught while it is unfolding, when there may still be time to intervene, instead of reconstructed from logs an hour later. Apolink's site leans on the use cases where that immediacy is the product: wildfire detection, disaster response, defense intelligence, and day-to-day fleet management.

The founder built his first satellite in high school

Apolink is the work of Onkar Singh Batra, who founded the company in 2024 and was 19 when the seed round closed. He is not new to satellites, only to running a company. Growing up in India's Jammu region, he built his first website at seven. In 12th grade he developed InQube, India's first open-source satellite, and he went on to guest-lecture at IIT Jammu before starting Apolink in the Bay Area.

The seed round drew 468 Capital, Unshackled Ventures, Rebel Fund, and Maiora Ventures alongside Y Combinator, plus angels who know the terrain: Epsilon3 CEO Laura Crabtree, Pebble co-founder Benjamin Bryant, and Jump Crypto's Kanav Kariya. For a founder barely out of his teens pitching orbital infrastructure, an oversubscribed round is its own kind of diligence.

The team is small, seven people as of the raise, which is worth pausing on. A seven-person company put a satellite through design, licensing, launch, and commissioning in about two years.

The first satellite is already talking

In July 2026, Apolink's LinkONE demonstration satellite, a 3U cubesat, went up on SpaceX's July 7 rideshare mission. Contact was established shortly after deployment, and the mission is now working through its actual test: receiving S-band signals from partner satellites in orbit, storing the data, and forwarding it to approved ground stations. The satellite is designed to close low-power links with spacecraft up to about 150 kilometers away during line-of-sight passes.

The mission also cleared a regulatory path. Batra has said the company holds a first-of-its-kind experimental license from the FCC for S-band inter-satellite link operations. Licenses sound like paperwork next to rockets, but in satellite communications they are often the harder half of the problem. Spectrum for satellite-to-satellite relay at commercial scale is largely uncharted regulatory territory, and Apolink now has a precedent with its name on it.

A demo satellite is not a network. It is, however, the difference between a deck that says "backward-compatible relay is possible" and a spacecraft in orbit doing it under an FCC license. At seed stage, that is a rare place to stand.

$140 million in interest before the network exists

Apolink reports more than $140 million in letters of intent from companies including Astro Digital, Hubble Network, and Star Catcher Industries. LOIs are not revenue, and seasoned space investors discount them accordingly. But the names say something about who feels the pain. Astro Digital builds and operates satellites for other companies. Hubble Network is putting a Bluetooth-compatible receiver network in orbit. Star Catcher wants to beam power between spacecraft. All three are operators with real constellations in the works, signaling what they would pay for connectivity that does not depend on ground station geometry.

The demand backdrop helps explain why. Tens of thousands of satellites are expected in low Earth orbit over the next decade, and each one inherits the same dead-zone problem. Ground station networks scale by pouring concrete on the right coastlines. A relay layer scales with the traffic above it. If LEO becomes the industrial zone the industry expects, always-on connectivity starts to look less like a premium feature and more like plumbing.

The roadmap from here is concrete. A second demonstration mission with two satellites is planned for June 2027, commercial service is slated to begin in 2028, and the full 32-satellite constellation is expected to be on orbit in 2029.

There is real distance between a working cubesat and a 32-satellite network with optical crosslinks, and nobody at Apolink would claim otherwise. But the company has been early on every promise so far: licensed before launch, launched on schedule, in contact on the first pass. Four years from a teenager's open-source cubesat to an FCC-licensed relay demo is a pace the incumbents should probably take personally.