Nature Abhors a Vacuum Tube

Why We Should Have Known Hyperloop Would Fail

Attention conservation notice: 16-minute read

There was a time, not that long ago, when we thought the future of transportation was to combine a rocket ship and a Thermos.

This turns out to have been a bad idea, but it's worth spending a bit of time exploring why it didn't seem that way at first, so that we can avoid mistakes like this in future.

What makes a rocket ship great is that, at least some of the time, it is travelling through a vacuum, and the lack of air resistance means that it can move at high speeds with limited power spent on propulsion. What makes a Thermos great is that, at least some of the time, it is sealed off from the outer environment by a vacuum, protecting and preserving its contents.

Hyperloop aimed to combine these functions. Passengers would be sealed into small pods, with a capacity of about 28 people at most. These pods would operate inside steel tubes which would contain mostly vacuum, with air pressure perhaps one one-thousandth of that at sea level. The pods would accelerate to nearly the speed of sound, so more than 1,000 kilometres per hour. The near-vacuum in which they operated would mean that once that speed was reached, minimal thrust would be necessary to keep it going. And that, in turn, meant that energy costs would be low and passengers would be comfortable, not pinned to their seats by the force of constant thrust.

(Acceleration and deceleration at the beginning and end of the ride would be costly, in both energy and passengers' comfort, but you can’t have everything.)

Speeds like that would be impressive. Had it worked, the Hyperloop could have taken passengers from Toronto to New York in about 45 minutes, or New York to Los Angeles in about four hours. That's competitive with commercial aircraft speeds, and especially so if—like high-speed rail—we imagine the Hyperloop would have travelled directly between city centres. But where it would really have shined would be middle distances.¹

Seattle to Vancouver in less than 15 minutes. New York to Washington, D.C. in 20 minutes; same time from Edmonton to Calgary. Toronto to Montreal in 35 minutes; and, indeed, San Francisco to Los Angeles as well. (This one's particularly relevant. More on that in a moment.)

But it was not to be: Virgin Hyperloop One, the most prominent Hyperloop company, ceased operations at the end of 2023. The technology is no longer being developed in any serious fashion by anyone. But for a decade or so, it dominated conversation about the future of intercity travel.

Why Hyperloop seemed so great

The obvious reference case to compare the Hyperloop to is high-speed rail (HSR). Both cases involve moving a vehicle in a dedicated right-of-way at high speeds. In fact, the earliest proponent of Hyperloop, Elon Musk, endorsed it specifically as better than HSR. Why?

Most obviously, Hyperloop would have been faster. California HSR is expected, at top speed, to travel no faster than 350 km/h, or less than a third as fast as Hyperloop.

For Hyperloop to exceed HSR speeds would have been enough to make the case. In transport, the two factors that consumers care about most are speed and cost, with other factors, like comfort or reliability, coming well behind. (If this wasn't true, commercial air travel would look very different than it does.) A 200% improvement in speed for Hyperloop over HSR would have been decisive. But Hyperloop, supposedly, would also have won the cost race, with a San-Francisco-to-Los-Angeles ticket to cost $20 USD in 2013 dollars.

Why so cheap? Supposedly, this was because of Hyperloop's cost profile. Like HSR, it would have a large capital cost to build: stations, rights-of-way, and trackage, in this case elevated or buried vacuum tubes, plus rolling stock. Unlike HSR, this rolling stock would be simpler and cheaper to build, because the real gains in speed would come from the infrastructure: the vacuum tubes would feature magnetic impellors, much like maglev trains today, to propel the passenger capsules forward. 

“Hyperloop Transporation [sic] Technologies - The Great Lakes Project3”, CC BY 2.0

That simplicity and inexpensiveness would in turn would mean that many more could be purchased and operated. Headways (that is, the time between departures) would be short, meaning that Hyperloop could serve many trips a day between destinations, or even many trips each hour. That frequency would satisfy all the latent demand for trips that Hyperloop could satisfy, and would indeed induce more, increasing revenue. Meanwhile, maintenance costs would be low. As long as the tubes maintained their vacuum, the system could continue to operate.

The rest was icing on the cake. No engines meant no noise pollution. All electric power meant green, sustainable, emissions-free operation. And all this while keeping the benefits of trains: downtown-to-downtown connection, fast boarding and disembarking, limited security requirements, and more. What's not to like?

As it turns out, plenty.

Fragile and dangerous

I'd like to begin this section with the words "in retrospect," but that would mislead. All of the problems with Hyperloop were evident from the start.

Start with the obvious one: the need to maintain vacuum. A Hyperloop vehicle would only be able to obtain and maintain its speed if it operated continuously in near-vacuum.

Vacuums, notoriously, are fragile. If a Hyperloop tube suffered a small puncture, it would not be catastrophic, as the system's air pumps would be able to remove small amounts of air. But if it suffered a large puncture, the situation would become much worse, very quickly. On Earth, at least, a significant breach in the integrity of a vacuum tube will destroy the vacuum, and often the tube itself, as the pressure differential between the air outside and the absence within causes air inflow greater than any pumps can address. Decreases in the vacuum of the tube would slow all the pods, degrading service while also requiring higher energy costs, both to propel the vehicle and operate the pumping system.

But it could get worse. A truly significant breach could cause explosive damage at the inbreak. Decades of science fiction have made this a familiar concept, although on television the explosive decompression is usually from a starship to the outside; in this case, the explosion would go the other way, of air into the tube. On Star Trek the solution to this problem is air locks: if one part of the Enterprise is punctured, bulkheads seal that part of the ship off, preserving the atmosphere elsewhere aboard. The Hyperloop would have to have such features, certainly. Each station would need one, with the passenger vehicle sealed in a locked section of tube, the air removed from that section, and then the section opened to the vacuum as a whole. But it would be impossible to seal off broken sections of Hyperloop tube.

“Hyperloop at Launch Festival 2016”, Kevin Krejci, CC BY 2.0

Why? Well, imagine there is a road incident on a highway, like a collision or a vehicle flipping. The scene of the incident must be protected from oncoming traffic. What would happen if that protection came in the form of brick walls suddenly rising up out of the pavement… without warning… in front of cars travelling 120 km/h per hour?

The flaws with this approach are obvious. And yet this approach resembles, sort of, what would have to happen in a broken Hyperloop. In this case the ‘brick wall’ would have had to have been barriers within the tube that suddenly close, to seal off the broken area. The absence of such safety measures would be even worse. Instead of pods striking compartment seals, they would instead collide with an incoming shock wave of air, possibly moving faster than the speed of sound. Passengers would be at extreme risk, irrespective of whatever safety measures might be put in place to mitigate that risk. And no pods would be able to finish, much less begin, their journeys.

So the system could become inoperable, and in fact dangerous to all users, in the event of even a single break in the line. No method of transport, except perhaps a Ferris wheel, has such little redundancy. If a train fails, other trains are inconvenienced but not endangered. If a jet fails, other jets don't even notice. But if Hyperloop fails, all users are impeded, and even endangered, by even a single dysfunction.

And that dysfunction is easy to imagine, given how fragile the system is. Leave aside entirely how bad actors might interfere. Stipulate high-quality manufacturing such that no component ever fails. Human error would be enough, such as one bad driver miscalculating on the highway. In California, at least, it was proposed that the Hyperloop right-of-way might be built along highway medians. So one truck, colliding at speed with a truss holding up the tube, could cause enough force to induce a break. 

And then, of course, there are earthquakes.

Hyperloop's solution to this problem was to rest the tube on elevated pylons. By not fixing the tube to the surface of the Earth, nor burying the tube within it, Hyperloop designers could reduce the risk of catastrophic damage. A car collision or earthquake would rock or displace the tube without breaking it. Imagine a hammock slung between two tree branches; in an earthquake, the trees would sway, and the hammock would too, but the hammock would be unlikely to tear. Conversely, if (for some reason) a hammock was nailed to the ground, without any scope for flexion, a sufficiently-violent earthquake would tear that hammock.

That explanation seems glib, as does the proposed solution. Earthquakes not only shake trees, they can fell them too, and any hammocks attached to those trees are torn as a result. How safe would vacuum tubes mounted to pylons be? How much force would be sufficient to pose a risk? How often do earthquakes of such force occur? The Hyperloop's lack of progress means that the matter was never put to the test, but it seems likely that regulators would have found a great deal to concern them with Hyperloop operation.

So too would have management consultants.

How to fail in business without really trying

Transport economics depends on striking a balance between convenience and scale.

The most convenient trip is point-to-point: the system takes you from where you are to where you want to go. A taxi or ridehail trip, for instance, maximizes convenience, because I go from wherever I happen to be to my destination, without intervening stops. Personal automobile or bike trips are often like this, although parking can add difficulty at either end.

These trips, however, don't scale; they serve only me. Most private auto trips are taken alone, with only the driver in the car; and all bike trips are taken alone.² The trip only serves one person.

At the other end of the spectrum are trips that scale very well. Commercial air travel, subway trains, ferry boats; all of these carry immense numbers of people in one vehicle. I have to go out of my way to take such vehicles, and when I disembark, I have to find some other mode to continue my journey. That’s inconvenient, but worth it, because of their speed, reliability, or route. In fact these modes are so useful that many travellers go out of their way to take advantage. That attractiveness is crucial. The cost to build and maintain these conveyances, and their allied infrastructure, is high, and the service is only viable if they can spread their costs over many users. 

The business logic is inexorable. A mode at scale must be able to accommodate the needs of many travellers at once, because of the variety of travellers there are. And so airports and bus depots typically serve multiple destinations. If they served only one destination, they would not be able to aggregate enough passengers to make themselves viable. Even rail lines rarely operate express trains between a single origin and destination; instead, express trains travel long distances, making multiple stops along the way, albeit many fewer than a local. In this way, express trains serve (and even create) hubs wherever possible.

As a transport technology, the Hyperloop made several risky bets at once.

Hyperloop's technology would not permit such aggregation. The combination of high speeds, and the difficulty of accelerating and braking relative to maintaining speed, all suggest A to B travel only. Worse, the technology discouraged branching. Traditional trains may follow one eastbound track out of Chicago all the way to Lake Erie, and then branch to Detroit on the north shore, or to Cleveland on the south. A Chicago Hyperloop would need to pick one of these to serve, and the other would be out of luck.

Possibly it could have passengers disembark at the lake and then board a new vehicle going in their own tube; but that would add friction to travel plans, depressing use. No one likes to have to change planes, or subway vehicles, or any sort of mode on a trip, which is why transport planners speak glowingly of the “one-seat ride”. Hyperloop would in most cases be unable to offer one. Alternatively, the tube could be made to branch, but a branching vacuum tube is a significant engineering problem. I’m not sure how it could have been made to work, but doing so would certainly add extreme difficulty and cost to construction, imperilling the viability of the whole system.

For all these reasons, we have to imagine that Hyperloop would never have been a network in the true sense of that word. By design, it would not operate in parallel, but only in serial. It would not serve many trips at once, but only one trip, over and over. And no more than 28 passengers at a time! Given these constraints, as well as the immense capital cost to construct, the mooted $20 ticket price seems highly suspect. One wonders whether the operators could set any price point for a ticket that would simultaneously allow them to turn a profit while remaining competitive with air travel. 

So the Hyperloop offered a system that would cost immense amounts of money, attract heavy scrutiny from regulators for its riskiness, and could not scale up to accommodate broad demand. The answers to the question "why did this fail" seem obvious. The more difficult question is "why did anyone ever think this would work?" And that is a question. In this case, as I noted earlier, foresight was 20/20; all of these concerns with Hyperloop were obvious from the beginning. So why were they overlooked for so long?

Give 'em the old razzle-dazzle

Two words: Elon Branson.

Hyperloop was first proposed by Elon Musk in 2013. In the white paper introducing the concept, Musk is explicit that he was inspired by California's plans to build HSR, which he described there as "a bullet train that is both one of the most expensive per mile and one of the slowest in the world". He wanted something better for "the home of Silicon Valley and JPL". Hyperloop was to be that better thing

Four years later, Richard Branson's Virgin Group invested in a Hyperloop company, subsequently renamed "Virgin Hyperloop One", with Branson himself named chairman.

This very short history gives us most of what we need to know about how Hyperloop managed to dominate conversation for years, despite the weaknesses of the idea.

“Virgin Hyperloop One XP-1 pod -– side view”, CC BY 2.0

New technologies and new firms face barriers to entry. Developing new technology takes money and time... and the more time it takes, the more money is required. Fundraising is the first job of a startup, and failure to gain enough funding leads directly to the firm itself going defunct. Investopedia says 50% of startups fail in the first five years of operation, and insufficient capital is the principal reason.

Enter Elon Musk and Richard Branson. Each has been a very successful serial entrepreneur, to the extent that they are household names, so much so that I don't need to explain to you who they are. As successful entrepreneurs, they have natural skill in building attention and hype around their projects, which they've honed over decades. Importantly, that skill in attention-building doesn't require that any given project succeed. By making themselves the brand, they can keep investors interested in whatever they're doing, provided they are constantly doing exciting things. And so Hyperloop was just one more card that each man played: for Musk, Hyperloop joined OpenAI, Neuralink, and the Boring Company as his big ideas of the 2010s, keeping attention and money flowing to Tesla Motors and SpaceX. For Branson, Hyperloop joined Virgin Hotels, Virgin Media, and Virgin Galactic, keeping attention and money flowing to Virgin Mobile and other parts of the Virgin Group.

From this point of view, Hyperloop was an exciting Big Idea. If it worked and made money, that would count as a win. But if it didn't, well, it kept attention being paid to the larger enterprises, and that would also count as a win.

Some cynical observers have, on slender evidence, suggested that the Hyperloop was so flawed an idea that it must have always been a con; an attempt to sandbag California HSR with a plausible alternative, just plausible enough to keep that project from receiving enough time and money and support to succeed, so that the personal automobile could continue to be the dominant mode of intercity transport in California. That seems unlikely to me. There is an alternative to Hyperloop and HSR, and it isn’t ‘driving a car’, it’s ‘commercial air travel’, which is doing well enough everywhere it doesn't need to sabotage its competitors.

I can't testify to what either Musk or Branson believed about Hyperloop. Were they lying? Were they operating in good faith? Perhaps. But there’s plenty of hypothesis-space between those extremes. Maybe they were happy to endorse any idea that burnished their reputations and kept attention and money flowing. But even if true, when choosing among such ideas, it’s easier to pick the ones that seem plausible. 

There’s some evidence to support the hypothesis that the proponents thought Hyperloop really could solve the problem of intercity transport. It's telling that Musk's original Hyperloop white paper considers a route from Los Angeles to San Francisco, referred to in some places in the document literally as "San Francisco / San Jose". The Hyperloop was aiming to solve a particular problem that the audience for the proposal found compelling: getting to and from Silicon Valley to California’s biggest cities. 

This is an example of what transit thinker Jarrett Walker refers to as "elite projection": the tendency of decision-makers to imagine that what the world needs is a system optimized to satisfy their own preferences, rather than the preferences of most people. Musk's problem was getting from San Jose to Los Angeles, and so he proposed a system to solve it. The fact that it could not easily branch to accommodate other trips did not factor in as important.

The road ahead

As a transport technology, the Hyperloop made several risky bets at once:

1. A dedicated right-of-way (in this case, an elevated tube)...

2. ...that was remarkably fragile, difficult to maintain, and dangerous to the whole network if it failed...

3. ...that accommodated only A-to-B travel, not travel among multiple points

The truly innovative transport technologies of recent decades have not made these bets. Trackless trams take bet 1, but not 2 or 3. High-speed rail and maglev trains take bets 1 and 3, but not 2. And automated driving, both for people and for goods, takes none of these. 

In fact, it's hard to think of any other technology ever proposed that takes all three of these bets. The best I can come up with is a space elevator... and even that, should one be built, could justify its existence with the fact that it would be unbeatable, in both cost and scale, at getting people and goods to its destination (Earth orbit). Hyperloop had too many competitors—air travel, auto travel, even train travel—to make a similar claim.

So Hyperloop is the latest manifestation of the "gadgetbahn" temptation, a tendency so common that transport specialists have given it its own name. It stems from the painful truth that genuine innovation in transport systems is hard. Moving people or goods through space at scale faces fundamental constraints in geometry. These constraints—at root, that two objects cannot share the same space at the same time—are not ones that a new technology can easily overcome. 

But it’s human nature to look away from the stubborn facts. It’s also natural to want revolutionary change, and to build something genuinely new and exciting. And if that new thing could also yield monopoly profits, large-scale manufacturing employment, and a place in history? These desires can easily prompt motivated thinking, making it easy to imagine that all the world needs is a new travel technology: a gadgetbahn. That’s the temptation: to focus only on the benefits while ignoring the problems. Hyperloop seems to me to be the latest expression of it.

But we can recognize the temptation, while also bearing in mind that material progress is always possible. There are new transport technologies waiting to be found, and we should give new proposals serious consideration, even the ones that seem quite unlikely at first glance. That is the tragedy of Hyperloop: it got so much adulation that it may have made it harder to recognize, fund, and build an actual good idea when it is proposed. 

All ideas can travel quickly through the vacuum of hype. The good ones will survive an atmosphere of scrutiny. 

Respect to Rob Tracinski, Steve Newman, Kevin Kohler, Julius Simonelli, and Sarah Constantin for feedback on earlier drafts.

Thanks for reading Changing Lanes! Please let us know how we’re doing by answering the poll below. And if you’d like to respond to this post, please leave a comment.

1. Musk's original 2013 white paper on Hyperloop understands that 1,500KM is the right range for Hyperloop, and beyond that supersonic aircraft would be better: "…a quiet supersonic plane immediately solves every long distance city pair without the need for a vast new worldwide infrastructure” (p. 2).

2. Yes, yes, tandem bikes; go away, kid, you bother me.