Astranis builds satellites the size of a washing machine.

A satellite factory inside an old shipyard.

Walk into Pier 70 on the eastern edge of San Francisco today and the building still smells faintly of metal and bay water. A century ago, it built dreadnoughts. Today, it builds spacecraft that fly to 22,236 miles above Earth and stay there.

Astranis is the company doing the building. As of May 2026 it has five satellites on orbit, a contract backlog north of a billion dollars, roughly 520 employees, and a fresh $300 million Series E announced earlier this month. It is the rare hardware company in San Francisco that ships things heavier than a laptop.

And it ships them small. A traditional geostationary communications satellite is the size of a school bus and costs hundreds of millions of dollars. An Astranis MicroGEO is closer to a commercial washing machine. The economics, predictably, are different.

"We design, build, test and operate spacecraft under one roof. There aren't many places left in America where you can say that." - Paraphrased from John Gedmark interviews

Roughly four billion people still don't have decent internet.

The standard answer for the last decade has been low-Earth-orbit megaconstellations - thousands of small satellites whizzing overhead, handing connections off every few minutes. It works. It's also extraordinarily expensive to build, and the bandwidth gets shared by everyone in your beam.

The classical answer before that was the school-bus GEO satellite: park one rock at 22,236 miles, light up a continent, charge accordingly. That worked too. The catch was that each rock took five years to build and ate a billion dollars, so only the rich-country incumbents could play.

Astranis spotted the gap between the two. What if a single country, or a single telco, or a single airline could afford its own dedicated GEO satellite? Not a slice of someone else's shared bird. Their own. Built in eighteen months, not five years.

"The space industry's biggest unsolved problem isn't physics. It's the cost of the satellite." - The bet that built the company

Two engineers, one washing machine.

John Gedmark spent his early career as founding executive director of the Commercial Spaceflight Federation, which means he watched, up close, the regulatory and economic plumbing of the U.S. space industry for years before deciding to start one of its companies. Ryan McLinko came from the small-satellite world - the engineering side, where weight is a religion and every gram counts.

They started Astranis in 2015 with an idea that sounded slightly absurd at the time: take everything the smallsat people had learned about packing capability into tiny boxes, and apply it to geostationary orbit instead. Then make the radio software-defined so a single satellite design could serve dozens of customers in dozens of bands.

Andreessen Horowitz wrote the first big check - $13.5 million in 2018 - and the company built and launched its first commercial spacecraft, Arcturus, in April 2023. It went to work delivering Ka-band internet to Alaska, which has the kind of geography that makes terrestrial fiber a punchline.

A dedicated piece of high orbit, sold like real estate.

The flagship is the MicroGEO - roughly 400 kilograms, software-defined, and parked in geostationary orbit where it sits over the same patch of Earth forever. A customer buys, in effect, their own piece of sky. The bandwidth doesn't get shared with anyone else. The coverage doesn't drift.

Then there's UtilitySat, which the company calls the world's first multi-mission commercial GEO satellite. It can hop between Ka, Ku and Q/V bands, and it has onboard propulsion, which means it can be moved around the geostationary arc to plug coverage gaps wherever they appear. It's the on-orbit equivalent of a roving spare.

Vanguard is the newest product, a tactical mobile ad-hoc network co-developed with Persistent Systems, Kymeta and SatCube. And then there's Resilient GPS - small, radiation-tolerant PNT spacecraft for the U.S. Space Force, because the existing GPS constellation is, depending on whom you ask, either fine or alarmingly fragile.

"A satellite the size of a washing machine, dedicated to a single country. That didn't exist before us." - The Astranis pitch, distilled

Five customers in five very different places.

Pacific Dataport gets the first satellite, for Alaska, because Alaskan rural broadband is one of the harder problems in American telecom. Apco Networks - also known as AiTelecom - gets two MicroGEOs dedicated to Mexico, the first Ka-band capacity ever specifically purposed for that country, potentially connecting five million people to affordable internet.

Andesat takes capacity for Peruvian cellular backhaul, the unglamorous infrastructure that makes mobile phones work in places mountains have made expensive. Orbits Corp signs on for the Philippines, with a goal of reaching two million users. Anuvu, in August 2025, brings two Astranis satellites online for in-flight Wi-Fi over the United States - what the press called the first private customer-dedicated GEO network in operation globally.

The U.S. Space Force, separately, picks Astranis to prototype the next generation of GPS. That contract alone is the sort of thing that puts a startup on a different trajectory entirely.

Get the next four billion people online.

That's the number Astranis quotes, and the number the satellite industry has quoted for a generation. It hasn't moved as much as anyone would like. Fiber doesn't reach mountain villages. Cellular towers don't follow you to the Bering Sea. Low-Earth orbit constellations help, but they're expensive to build and the unit economics are still a moving target.

The Astranis bet is that the answer isn't one giant constellation owned by one giant company. It's many small satellites owned by the people who care most about a specific patch of Earth - the Peruvian telco, the Mexican operator, the Philippine government, the Alaskan ISP. Sovereignty over your own piece of sky, sold for tens of millions instead of hundreds.

"We want to be the company that connects the next four billion people." - John Gedmark

The shipyard is busier than it's been in a century.

Return to Pier 70. The pier is loud again, the way piers like that are supposed to be. Engineers in clean-room bunny suits move components around a factory floor that, a hundred years ago, was forging steel hulls for ships that would cross oceans. The hulls Astranis builds now cross something further.

The $300 million Series E, announced in early May 2026, is for more of the same - more satellites, more launches, more dedicated bandwidth for countries that have spent decades waiting in line for someone else's spare capacity. The contract backlog suggests a lot of countries were waiting in that line.

Whether the small-GEO thesis is correct in five years is a question nobody can answer yet. What's clear is that the bet has been placed, the spacecraft are flying, and a corner of San Francisco that used to build battleships is now building something quieter, smaller, and pointed at the stars instead of the horizon.