FIRST PROPOSED BY Elon Musk, the theoretical transportation system we call hyperloop would propel people- or cargo-filled pods over long distances through steel tubes. Magnetic levitation and big vacuum pumps would do away with pesky friction and air resistance, letting those bus-sized vehicles zip along at speeds approaching Mach 1. It wouldn’t just be fast, the boosters say: Hyperloop could be cheaper and better for the environment than the planes, trains, and cars in which humanity putzes about today.

And like so many promised panaceas, it’s actually quite simple—on the surface. The tubes and pods should be easy enough to build, but making hyperloop a reality takes more than a few good engineers and a small fortune or two. It will require a whole lot of legal maneuvering, regulatory wrestling, and a massive amount of political will and public buy-in. Infrastructure, you should know, is hard.

The tubular tizzy started in 2012, when Tesla and SpaceX CEO Elon Musk suggested The Hyperloop as a new form of transportation, one that would be twice as fast as a plane and totally solar powered. He didn’t offer any engineering specifics at the time, but in August 2013 he produced a 57-page white paper that outlined his technical thinking for how this system would work.

At its core, hyperloop is all about removing the two things that slow down regular vehicles: friction and air resistance. To do away with the former, you make the pod hover above its track, like a magnetic levitation train. Musk originally suggested doing this with air bearings, little jets of air on the bottom of the pod. Think of air hockey, he said, but where the air comes out of the puck instead of the table. Today, most hyperloop engineers have decided instead to rely on passive magnetic levitation. Where standard maglev systems are power hungry and expensive, this system uses an array of permanent magnets on the vehicle. When those magnets move over conductive arrays in the track, they create a magnetic field that pushes the pod up, no current required. A complementary magnet system (think of two magnets pushing off one another) would give the pods a push every few miles or so—the near total lack of friction and air resistance means you don’t need a constant propulsion system.

As for air resistance, that’s where the tube comes in. (Yes, tubes also just feel like the future, but that’s not the point.) The tubes enclose the space through which the pods move, so you can use vacuums to hoover out nearly all the air—leaving so little that the physics are like being at an altitude of 200,000 feet. And so, like a cruising airplane, a hyperloop needs only a little bit of energy to maintain the pods’ speed, because there’s less stuff to push through. More speed with less power gets you to where you’re going faster, greener, and—depending on energy costs—maybe cheaper too.

After explaining all this, Musk said he was too busy to build the thing himself. He was running both Tesla and SpaceX and didn’t have time to remake yet another industry. So he encouraged anyone interested to have a go. Let there be hyperloop, he said.

And there was hyperloop. Well, a hyperloop industry, anyway. Soon after Musk’s paper hit the internet, a handful of companies sprung up, bringing together engineers and VC money to solve the problems for real. From the beginning, LA-based Virgin Hyperloop One has appeared to be the most serious contender, with serious VC backing, hundreds of employees, a full bank account, and a test track in the Nevada desert where, in December, it sent a pod racing to 240 mph.

Hyperloop Transportation Technologies takes a less built-up approach. Nearly all its engineers have day jobs at other companies (places like Boeing, NASA, and SpaceX). In their free time, they work together, mostly online and in distinct groups, to solve the engineering problems standing between humanity and hyperloop. It has plans to build networks in Central Europe, South Korea, and India. Similarly, there’s rLoop, a Reddit-based community of people who study the various engineering problems in the mission of “decentralizing high technology.”

Oh, and Elon Musk is back in the game. The hyperloop progenitor started by hosting a series of student engineering competitions, using a short length of tube he built at SpaceX’s headquarters. Then, last summer, he confirmed he wants to build a hyperloop of his own. His plans are particularly vague, but he thinks the tubular system would go great with the tunnels he wants to create using another new venture, the Boring Company.

While the various companies here are mostly pursuing the same tech (passive magnetic levitation, big vacuum pumps), it didn’t take long for the young hyperloop industry to splinter. Former SpaceX engineer Brogan BamBrogan helped launch Hyperloop One, but left in August 2016 amid a bizarre and bitter legal dispute with the company, in particular cofounder Shervin Pishevar (who took a leave of absence from the company in December 2017 after several women accused him of sexual misconduct). BamBrogan (that’s his legal name) then started his own outfit, Arrivo, except now he’s working on what he calls a hyperloop-inspired system. He got rid of the tube, deeming it too expensive. “If I want to travel really fast between two cities in a low-pressure environment inside a metal tube, I would use an airplane,” he says. It’s a valuable reminder that “hyperloop” is not an invention but a clever combination of technologies that together make something very fast and very fun.

If you really want hyperloop, however, you must build a hyperloop. There are lots of renderings and promises out there: The companies in this space have announced plans to build hyperloops in California, Colorado, on the East Coast, in India, Slovenia, Dubai, and Abu Dhabi. Hyperloop One wants a commercial line in service in 2020.

Over the next few years, then, we’ll start to see answers to the real question here. It’s not “can hyperloop work”—we know the engineering make sense. As BamBrogan puts it: “It’s within the laws of physics, but hard enough to be fun.”

Here’s the real question, as put by David Clarke, director of the Center for Transportation Research at the University of Tennessee, Knoxville: “Can it compete—from a capital standpoint and an operating standpoint and a safety standpoint?”

To really work, Clarke means, a hyperloop must offer the kind of service, pricing, and safety record that will draw paying passengers away from current modes of transportation, including airlines, trains (that applies more overseas than in the US), and the personal car. Those systems may not be perfect, but they have established user bases, are more or less profitable, and are safe enough to keep people riding and regulators happy. They know how to work with governments around the world, and they know how to build the infrastructure they need to run—how to get it certified and funded and in place.

That’s why the first hyperloop systems will likely target very specific use cases with built-in passengers and minimal political hurdles. They could connect an airport to a city center or public transit hub, or send cargo from a port to an inland distribution center, so trucks don’t have to crowd into already congested areas. Tackling a real long-distance, city-to-city route will make things much harder.

To even have a shot at competing, hyperloop must start by finding a way to finagle through the bureaucratic regulations that govern what gets built where. The people running these companies insist that it won’t be as hard as it seems and that they’re already working with eager governments to get their systems built. To make things easier, Hyperloop One held a competition in which cities pitched for the right to host the thing. No doubt, places willing to clear out obstacles like pesky regulations stood out. The winners included Canada (with a route connecting Toronto, Ottawa, and Montreal), Florida (Orlando to Miami), and India (Mumbai to Chennai), but the company hasn’t announced any actual plans to start building. And, of course, it remains to be seen whether any promises will hold when local residents protest, land rights prove hard to acquire, and construction costs mount up.

More hurdles: These companies will need to figure out how to prove that traveling by tube is safe. What happens if an asteroid rips open the tube or one of the supporting pylons collapses in an earthquake? The hyperloop engineers say the pod will just slow down in the face of sudden air resistance, but a rapid slowdown is often known as a crash. And if the pod is near a ruptured tube, what happens if it flies out? Will regulators insist the pods meet crash standards, like cars, or that everyone wear seat belts at all times? Whatever the answers, expect the first working systems to move cargo, not carbon-based lifeforms.

More questions: How much energy will it take to fling those pods up to near-supersonic speeds? Doing it with renewables would be great, but can you generate and store enough solar power to run all those pods, wherever they are, whenever people want to whoosh?

Then there’s the money. Virgin Hyperloop One CEO Rob Lloyd has said it would cost about $10 million to build one mile of two-way track, less than a third of what California is paying for its stuck-in-limbo high speed rail system. Musk’s original paper estimated a hyperloop from Los Angeles to San Francisco would cost $6 billion, and you could recoup the investment and cover operating costs with $20 ticket prices. Of course, that was five years ago and doesn’t account for changes in the engineering made by the involved companies; it also comes from a man notorious for lowballing cost (and time) estimates. The truth is, we have no idea how much it will cost to maintain a working hyperloop, which requires keeping hundreds of miles of tubes nearly free of air, and won’t until we’re closer to a working system.

If you’ve ever wondered what happened to the high speed rail system that was supposed to connect Los Angeles and San Francisco, or why people stopped building super fast magnetic levitation trains after the first few systems started service, you should know that details like local politics and maintenance costs have a knack for hampering transportation innovation. And maybe that’s not such a terrible thing. Trains, hyperloops, airports, light rail lines—these are big things that cost a lot of money and impact many people’s lives. You want to be sure of what you’re doing before you throw the switch.

So what comes next? A bit more engineering, to start. Then real life—and that’s when we’ll see whether hyperloop can really change the world, or at least get rid of some of the traffic. And if you really want to ride in a hyperloop and you’re not a shipping container, you might want to move to Dubai. If any place can sweep away political hurdles and ignore potentially outrageous energy bills, it’s the city whose motto might as well be “Sounds shiny and impractical—let’s do it!”