This is The Stepback, a weekly newsletter breaking down one essential story from the tech world. For more on how to work anywhere, follow Thomas Ricker. The Stepback arrives in our subscribers’ inboxes at 8AM ET. Opt in for The Stepback here.
I’ve worked from home for 20 years. In that time, I’ve seen technologies supporting remote work advance considerably. But nothing has been as transformative as the arrival of Starlink, SpaceX’s internet service that lets me “work from home” anywhere I choose, be it from the open road, forest, or desolate beach.
SpaceX began launching Starlink satellites in 2019 to provide data in the large gaps that still exist between 4G, 5G, and traditional broadband networks. It now has a monopoly on high-bandwidth, low-latency internet that can be quickly and easily installed anywhere, be it land, sea, or air, or in response to a natural disaster. Other consumer satellite internet services exist with more scheduled to come, but only Amazon’s Leo (previously known as Project Kuiper) looks like it’ll be a true competitor, having just lit up the service last week for enterprise customers.
Starlink internet, like Leo, relies upon a large constellation of satellites operating in low Earth orbit (LEO) — not geostationary orbit like the Viasat and Hughesnet services, which are unusable trash by comparison. Starlink user terminals (aka dishes) lock on to the best available satellite traveling at 17,000mph about 350 miles above the Earth. Ground stations direct traffic between the satellites and the internet.
By May 2021, there were enough Starlink satellites (about 1,500) in operation for The Verge to test Elon Musk’s space internet service from the northern US. Nilay’s experience wasn’t great, calling it “very much a beta product that is unreliable, inconsistent, and foiled by even the merest suggestion of trees.”
A year later, I came away with a much different experience when testing what was then called “Starlink RV” from a camper van in Western Europe, finding it to be fast and reliable compared to the 4G/5G data networks in the remote locations I like to travel. Unlike Nilay, I was testing a more capable dish on a larger constellation of about 2,400 satellites, and I could just move my home whenever trees obstructed the sky.
I was so impressed that I immediately subscribed to Starlink and eventually bought a van to start chasing my vanlife fantasies. At the time, Elon Musk was prone to “pedo guy” outbursts and covid-19 skepticism but had yet to fully show himself.
Today, SpaceX has deployed over 10,000 Starlink satellites, of which about 9,000 are operational. That’s enough density to allow user terminals to stay connected even when partially obstructed by buildings and trees. According to SpaceX, user terminals in the US now have “10s of satellites in view,” allowing them to instantly switch to satellites traveling on an unobstructed path. These proactive switches happen many times per minute to keep the connection stable and are imperceptible to users.
Performance varies by the dish, service level paid for, time of day, and location of the user. We’re talking 20ms to 50ms latency (good enough to play first-person shooters with friends), 100Mbps to 400Mbps downloads (comparable to fixed broadband in homes), and 10Mbps to 40Mbps upload speeds (good enough for two people to video conference on the same connection).
The increasingly robust internet service allows me to work about one-third of the year from a surf shack on the North Sea and another third from my van while traveling around Europe. It’s given me the type of work-life balance I’ve long dreamed of.
Recently, however, my monthly service price went up for the first time, and Starlink took away the much-loved pause feature. Seasonal users must now cancel their subscription and sign up anew, with no guarantee that a slot will be available. It’s a scheme that’s ripe for the implementation of costly activation fees in the future.
The service clearly needs competition to keep SpaceX from exploiting a user base with nowhere else to go, be they part-time digital nomads like me, rural Americans, or users in truly remote settlements. And frankly, I, like many Starlink subscribers, would like to ditch my dependency on the incredibly polarizing Mr. Musk.
Amazon recently launched its Leo space internet service as an “enterprise preview” to a few unnamed corporate customers. A broader rollout is promised for 2026. But it can’t offer anything close to Starlink until its constellation reaches sufficient density.
Unlike SpaceX, which handles all Starlink launches, Amazon has partnered with a number of launch partners for Leo, including SpaceX. So far, it only has 153 satellites in low Earth orbit since deployments began in April. In that same period, SpaceX has deployed over 1,500 new Starlink satellites, with dozens of new spacecraft joining its constellation every week.
SpaceX has FCC approval for as many as 12,000 Starlink satellites with a stretch goal of 42,000. Amazon has FCC approval for 3,236 Leo satellites, which will take a decade to deploy at this rate. Leo deployments could accelerate in 2026 once Blue Origin’s reusable New Glenn rocket — from Jeff Bezos’ other company — is done teething.
The Leo service is launching with the promise of speeds up to 1Gbps, something Starlink can’t match until SpaceX gets its larger Starship rocket operational. And while Amazon hasn’t announced any pricing yet, I’m sure Bezos has the funds to undercut Starlink’s monthly subscriptions until more Leo satellites come online, especially if the consumer offering gets bundled into Prime memberships.
Starlink’s service offerings have grown increasingly complex, with prices varying significantly by location, dish and router selection, and data needs. The US is one of the most expensive markets due to high demand, especially in rural areas that lack competition. Conversely, those same satellites that pass over the US arrive over my tiny country of the Netherlands with lots of excess capacity. SpaceX has to keep prices low to attract customers that already have access to cheap and fast broadband and mobile data.
In the US, you’ll currently pay $349 for the Standard Residential kit and then $120 each month thereafter. In NL, that hardware costs about the same, but the monthly service fee costs less than half at €50 (about $58). Starlink hardware costs drop to $0 in major markets if you’re willing to commit to a 12-month contract.
I paid €299 (about $315) for my Starlink Mini dish and €89 (about $103) each month for unlimited data anywhere in Europe. That same service in the US costs $599 for the dish and then $165 each month. Starlink says it has over 8 million active customers globally. With any luck, Leo will create some much-needed competition to help bring prices down, and provide an off-ramp for the Musk-averse.
In addition to Amazon Leo, there are several private and government-backed entities trying to compete with Starlink.
The Eutelsat OneWeb constellation of about 650 satellites has been operating in low Earth orbit for years, but it doesn’t sell directly to individual consumers. There’s lots of talk about the debt-laden service one day competing with Starlink, but it only ever amounts to talk. Likewise, China’s Spacesail Constellation (aka Qianfan/G60) has launched just 108 of the 648 satellites it had targeted by the end of 2025. Operational issues and a lack of reusable launch vehicles have hindered deployments that are supposed to yield a constellation of up 15,000 satellites by 2030.
Europe’s IRIS² constellation is also in the works, with plans to launch 290 satellites into LEO (with an additional 18 in medium Earth orbit) by 2030. But the sovereign space internet service is expected to be restricted to EU citizens, businesses, and government entities.
Naturally, the prospect of all those spacecraft flying around in similar orbits has raised concern. Astronomers have complained about light reflecting off existing constellations, while others are concerned with the risks presented by an increasingly crowded sky. These concerns are being addressed but are far from resolved.
As we roll into 2026, Amazon’s Leo space internet service from that other billionaire with his own baggage is looking like the best near-term hope for consumers to break free from their Elon Musk dependency. But it’s going to take a few more years for the Leo constellation to grow into a viable Starlink competitor. And with no immediate exit strategy, I’ve subjected myself to what I’m calling an “ethical offset tax” by donating to the Center for Countering Digital Hate (CCDH). Musk hates the CCDH, and that helps me sleep better at night no matter where I lay my head.
- Starlink satellites have a lifespan of roughly five years before they’re intentionally de-orbited to burn up in the atmosphere. Of the over 10,000 that have launched, about 1,400 have been decommissioned.
- With one month to go, 2025 has already been a record year for SpaceX, averaging one Falcon 9 rocket launch every two days, the vast majority carrying a few dozen Starlink satellites. Starlink should end the year with about 2,500 new satellites in its constellation, up from the 1,700–1,900 deployed in each of the last three years.
- More powerful third-generation Starlink satellites are expected to enter service in 2026. They’ll reduce latency and provide gigabit speeds for subscribers. But first SpaceX has to get Starship into an operational launch cadence to deliver these massive satellites into LEO.
- My review of the Starlink Mini, tested in a variety of scenarios including on my moving van’s dashboard and inside a backpack.
- This battery I reviewed is a must-have accessory to untether the Starlink Mini dish.
- The Verge examines the threats posed by megaconstellations like Starlink and Leo.
- The New York Times explains why even China can’t compete with Starlink.
- Scientific American describes the steps astronomers are taking to cope with all those bright satellites.