One particularly interesting thing about this launch is its payload. Alongside a satellite that will be used by the Spanish defense ministry, there are two prototype satellites for SpaceX’s planned Starlink constellation—the company’s contender in the growing push to provide cheap internet services to people on the ground.

High altitude wind shear data shows a probable 2% load exceedance. Small, but better to be paranoid. Postponing launch to tomorrow, assuming winds are better then.

— Elon Musk (@elonmusk) February 21, 2018

Here’s what you need to know about Starlink, and where it sits in the satellite-broadband scene.

Is satellite broadband new?

Not at all. The first satellite built for two-way broadband communications—e-BIRD, built by Boeing (BA) for the French-headquartered Eutelsat—went up some 15 years ago. There have been many more since. However, so far it’s a relatively niche market.

The advantage of satellite-based broadband is that it can cover entire regions without the need to build out expensive land-based internet infrastructure—you just need a satellite dish to use it, which makes it good for serving rural areas. And while it started out pretty slow, it’s gotten a lot faster over the years (though it’s still not in the same league as a decent fiber connection).

There are two big disadvantages, though (apart from interruptions due to cloud cover). The first is that it tends to be pretty expensive, and the second is that it usually suffers from high latency, because of how high the satellites are.

Latency is the time it takes for a signal to get from one place to another. While it’s not a big issue when you’re sending emails or trying to view a static webpage, high latency can be a showstopper for video, virtual reality, and real-time communications.

So what’s SpaceX doing that’s different?

SpaceX’s idea is to put its satellites into much lower orbit than usual, in order to cut the latency of the services. A typical internet satellite in geostationary orbit is more than 22,000 miles above ground. According to SpaceX’s FCC filings, the company wants to put its Starlink satellites in low Earth orbit, between 684 and 823 miles in the air.

The issue there, of course, is that bringing the satellites closer to the ground means each satellite can only cover a much smaller patch of territory than would otherwise be possible. So in order to provide competitive coverage, SpaceX will need a lot of satellites that all talk to one another.

And that’s the plan. Elon Musk’s company wants to initially deploy 800 satellites in low Earth orbit, in order to cover “initial U.S. and international coverage.” Then it wants to throw over 7,000 more into the sky at “Very Low Earth Orbit” (VLEO, in this case around 211 miles up) to fill in the blanks as needed.

That sounds… expensive?

Sure, but less so if you’re in the business of launching stuff already—this week’s planned launch is a case in point, with SpaceX’s Microsat 2a and 2b satellites tucked in there alongside Spain’s Paz satellite. And don’t forget that SpaceX’s biggest innovation is being able to reuse its rockets.

SpaceX has already become adept at recovering and reusing the first stage (the body) of its Falcon 9 rockets, and it’s hoping to become as successful at recovering the fairing (the bit at the top, which contains the payload being sent into orbit). To do this, it’s built a giant net with the delightful name of “Mr. Steven.”

Indeed, SpaceX seems to want Starlink to be its big money-spinner, ultimately financing Musk’s Martian ambitions. The company’s launches operate on thin margins—a successful Internet business would bring in a lot more cash.

And what are the challenges?

Apart from the numerous technical challenges that SpaceX faces, it’s also annoyed communications satellite operators such as OneWeb, which are worried that SpaceX’s vast planned constellation will pose a physical threat to their own satellites in low Earth orbit (which are generally used for voice communications).

SpaceX has countered that its plans include plenty of space between satellites—at least 31 miles—and that it will also be developing “ongoing innovations” to reduce the risk of bringing its satellites down, once they reach the ends of their working lives.

The company seems likely to get its way, albeit slowly. Earlier this month, Federal Communications Commission (FCC) chair Ajit Pai said he was in favor of approving SpaceX’s application to operate its constellation. However, other bodies—notably the International Telecommunication Union—also need to give the green light. It will be some years before SpaceX’s service is fully operational.

The other big challenge is competition. OneWeb, Telesat, and Space Norway have also received the FCC’s go-ahead for similar (if not quite so ambitious) satellite plans. Then there are rival plans that aren’t based on satellites, such as X’s (formerly Google X’s) Project Loon.

And let’s not forget good old mobile broadband. While some experts are skeptical about the practicalities, one big promise of the looming “5G” cellular technology is that it will be able to provide broadband services to rural areas that don’t currently get good fixed-line services.

There are plenty of competitors out there for Musk’s scheme. But that said, the others don’t have Mars on their roadmap.