The Boring Company & the Future Underground
The Boring Company & the Future Underground


Ask Steve Davis when he plans to retire as the CEO of the Boring Company, and he’ll tell you: “the day I can commute from D.C. to Vegas in half an hour.”
D.C. and Las Vegas are about 2,100 miles apart, so that would mean averaging more than 4,000 miles an hour. It sounds absurd, but Davis isn’t the first to imagine it. In 1972, RAND researcher Robert M. Salter proposed an underground vacuum-tube system that could carry passengers from Los Angeles to New York in 21 minutes, at speeds as high as 14,000 miles an hour, and Elon Musk cited the idea in his 2013 Hyperloop white paper.

The vehicle isn’t the hard part, though. “The hard part about Hyperloop is tunneling,” Davis says. The faster you go, the straighter the path has to be, which means digging through or under obstacles rather than around them, for thousands of miles. And tunnels have historically cost far too much for that to make sense. So before it can build a hyperloop, the Boring Company has to make tunnels cheap.
Fortunately for them, Musk has spent decades making expensive technology cheap. When Charlie Rose asked him in 2009 why Tesla started with an expensive sports car, Musk said, “Any new technology is expensive when it starts out,” because the first thing engineers are trying to do is make it work. “And then when you’ve made it work, you optimize, and you optimize, and you optimize.” The first Roadster sold for $98,000 in 2008, and today a Model 3 starts at $36,990. At SpaceX, the Falcon 9 cut the cost of getting a kilogram to orbit from about $18,500 historically to roughly $2,700 in 2010, and Musk says that once Starship is fully and rapidly reusable, it could bring the cost “well below” $100 per kilogram.
Now Musk and Davis are running a version of that playbook on tunnels, and they’re starting with cities that routinely pay hundreds of millions or billions of dollars for new subway and light rail lines. The company’s first product, the Vegas Loop, offers an early glimpse of how that might work. It’s a transit system where Teslas take passengers through underground tunnels straight to their destinations, and since opening beneath the Las Vegas Convention Center in 2021, it has carried more than four million passengers. Boring now has projects underway or in the works in Nashville, San Antonio, Orlando, and Dubai, and eventually Davis wants to connect cities like these with hyperloops.
Since its founding in 2017, the Boring Company has brought the cost of tunneling below $10 million a mile, with a further halving in sight. That’s already less than what it costs to add an urban highway lane. “If we’re below $5 million,” Davis asks, “why would you ever build a road again?”

A Rocket Scientist, an Economist, and a Yogurt Shop Owner Walk Into a Bar
At the other end of Davis’s dream commute is Washington. It’s an unlikely object of affection for a rocket scientist, but he fell in love with the city after SpaceX sent him there to open its D.C. office in 2008. Before that, he spent years in Southern California, where he designed guidance systems for the company’s first rocket.
His new job in D.C., working with NASA and other federal agencies, evidently left him with spare capacity, which he spent in unusual ways. He earned a Ph.D. in economics at George Mason University to complement his master’s degrees in particle physics and aerospace engineering; his 2010 dissertation was on the debasement of American currency. Unable to find the frozen yogurt he’d gotten used to in California, he taught himself how to make it and opened a shop called Mr. Yogato. (He later sold it for $1 after holding a contest to choose his successor.) Then he opened Thomas Foolery, which the Washington City Paper called “the wackiest bar in Washington.” It gave discounts to anyone who dressed as Carlton from The Fresh Prince of Bel-Air or shouted their order during “angry hour.”
“Honestly, it’s no more than some people play golf,” he said when asked how he found the time. “I like opening fun new places.”

But his eclectic interests aren’t why Elon Musk trusted Davis to run the Boring Company. “He gets more done than 11 people working together,” a SpaceX engineer told writer Ashlee Vance. Readers of our SpaceX profile may remember Davis as the fourteenth employee who joined the company right out of Stanford and built a rocket actuator that a supplier had quoted at $120,000 for just $3,900.
Over time, Davis developed a reputation at SpaceX for being “hardcore,” and became one of Musk’s most trusted engineers. When Musk bought Twitter in 2022, Davis was by his side, reportedly even sleeping in a makeshift bedroom at the company’s San Francisco headquarters with his partner and their newborn.
Musk usually gets the attention for the Boring Company, but “Steve Davis and his team,” Musk wrote last month, “deserve almost all the credit.”
Beat the Snail!
When people tell the story of the Boring Company, they usually begin with a tweet Musk sent while sitting in Los Angeles traffic on December 17, 2016: “Traffic is driving me nuts. Am going to build a tunnel boring machine and just start digging…”
But Musk had been thinking about tunnels long before that. “Tunnels, in my view, are the only solution to urban congestion because we have a 2-dimensional road network and buildings in 3 dimensions,” he explains. When everyone piles into and out of those buildings at the same time, traffic is almost inevitable. The fix is to make transportation three-dimensional too, and there are only two directions to choose from: up or down.
The long-proposed above-ground solution is flying cars, but that would mean millions of them flying overhead every day. “Inevitably somebody is not going to service their car properly, and it’s going to drop a hubcap and guillotine somebody,” Musk quips. Flying cars would also be at the mercy of the weather, and anything big enough to carry a person creates a lot of noise and wind force. You can’t fly even the quietest helicopter through a neighborhood without bothering people, he points out.

Tunnels, on the other hand, scale really well. “In theory you could have hundreds of levels of tunnels and arbitrarily solve any level of traffic,” Musk says. You don’t have to worry about bad weather. No one can see, hear, or feel them. They don’t divide neighborhoods the way highways do. And you can easily add more without the years of disruption that accompanies adding more lanes to a highway.
For years, whenever Musk gave a talk and someone asked what opportunities he saw in the world, he’d tell them to build tunnels. “I think they were hoping I’d say some sort of iPhone app that they could make,” he jokes. “I was basically talking people’s ears off about tunnels for several years,” he says, and then finally decided, “Well, let’s find out what it takes to build a tunnel.”
So on Friday, January 27, 2017, Musk asked the SpaceX team how long it would take to clear the employee parking lot across the street and start digging.
Two weeks, they said.
“Why two weeks?” Musk asked. “Let’s get started today and see what’s the biggest hole we can dig between now and Sunday afternoon, running 24 hours a day.”
Within three hours, the cars in the parking lot were moved and there was a hole in the ground.

Musk wanted the pit where he could see it from his office to make sure it was actually being built, and he put Davis in charge. The plan was to dig 50 feet down, deep enough to clear the gas and sewer lines, then turn horizontal and secure the remaining permits they would need before the tunnel reached the property line.
A week later, Musk and Davis brought a reporter from Bloomberg Businessweek along as they shopped for a used boring machine at a municipal project in Washington. New machines typically cost $15 million or more, but after a decade of subway construction in China had created a glut, lightly used ones could be bought at a 90% discount. A crane lowered the three of them into a 100-foot-deep shaft so they could get a closer look at the 400-foot-long, 1,200-ton boring machine. “Holy shit,” Musk said, wading into the mud in knee-high muck boots pulled over the suit he’d worn to meet the president that day. “This is cool.”
Musk was struck by the complexity of the operation. Besides the cutting head that actually drilled the earth, conveyor belts carried mud to the surface while trains hauled in concrete rings to line the tunnel. It was like a factory on wheels.
“We’re trying to dramatically increase the tunneling speed,” Musk told the site’s project manager. “We want to know what it would take to get to a mile a week?”
The project manager was taken aback. That was nearly 18 times the crew’s pace. But Musk wasn’t discouraged. “I think we can simplify this a lot,” he told the reporter as the crane lifted them out of the pit.
Back in Los Angeles, Davis kept looking for a machine of their own. “I didn’t know anything about construction,” he remembers, “so I just emailed all of the big construction companies,” asking whether anyone had a 12-foot-diameter machine they could send. “I think I offered like a million bucks.” Super Excavators, then a 67-year-old contractor in Menomonee Falls, Wisconsin, had a 14-foot machine built by the Canadian firm Lovat. It arrived at SpaceX’s headquarters in April 2017, four months after Musk’s tweet, along with a crew to help get it running.
They named it Godot, after the Samuel Beckett play in which two men wait for someone who never arrives. Then the team started digging, constantly asking themselves: Do we think we can make this better? “The answer was unequivocally yes,” Davis says. “If the answer had been no, we wouldn’t have started the Boring Company.”

Like the aerospace industry before SpaceX, tunneling was dominated by a few big incumbents with a deeply cautious culture, little standardization, and almost no incentive to cut costs. The giant machines that dig subways and highway tunnels were custom-built for each project by a handful of manufacturers, mostly in Germany and China. And when a tunnel went over budget, the public usually paid the difference. Davis and team believed that by rethinking the problem from first principles, they could improve both speed and cost by at least an order of magnitude.
The same month Godot arrived, Musk unveiled his vision on stage at TED. In a concept video, a Tesla pulled into a parking space that lowered it beneath the street, where it rode an electric “skate” through the tunnels at up to 130 miles per hour. The entrances and exits, Musk explained, had to fit “seamlessly into the fabric of the city.”

Musk also argued that self-driving cars would make tunnels more necessary, not less. Many people assume autonomy will ease congestion, he said. But once shared autonomous cars make a ride cheaper than a bus, “the amount of driving that will occur will be much greater, and traffic will get far worse.”
When TED’s Chris Anderson asked how he’d answer critics who say tunnels are too expensive, Musk agreed with them. Los Angeles had just spent about $2 billion on a 2.5-mile subway extension, “roughly a billion dollars a mile,” he pointed out.
His plan to make tunneling ten times cheaper had three parts:
Cut the tunnel diameter by a factor of two or more. Regulations require a single-lane road tunnel to be about 26 feet across to allow for crashes, emergency vehicles, and ventilation for combustion engines. But 12 feet is plenty for an electric skate, and if you halve the diameter of the tunnel, you cut the cross-sectional area by a factor of four. “Tunneling cost scales with the cross-sectional area,” Musk explains, “so that’s roughly half an order of magnitude right there.”
Continuous tunneling. Conventional machines bore for about half the time and spend the rest stopped while crews reinforce the walls. If you design a machine that does both at once, that would double the speed and halve the cost.
More power. Musk points out that these machines are “far from being at their power or thermal limits.” If you increase the power, he continued, “I think you can get at least a factor of two – maybe four or five – improvement on top of that.”
This would also bring the company closer to its goal of tunneling faster than its mascot, a pet snail named Gary. “Gary is currently capable of going fourteen times faster than a tunnel-boring machine,” Musk said. “Victory is beating the snail.”

Not a Flamethrower
In July 2017, Musk announced that the company had received “verbal government approval” to build an underground hyperloop connecting New York, Philadelphia, Baltimore, and Washington, with a 29-minute trip from D.C. to New York. That October, the Maryland Department of Transportation issued a conditional permit for the first 10.3 miles of tunnel between Baltimore and Washington, but the rest of the route ran under federal land and still needed federal approval.

To help fund the company in the meantime, Musk started selling merchandise. On October 17, 2017, he tweeted a link to “the world’s most boring hat.” By the next day, the company had sold 4,000 of them for $20 each. In December, Musk promised that if 50,000 hats sold, the Boring Company would start selling flamethrowers, a nod to the 1987 movie Spaceballs, where a flamethrower turns up in the merchandising display. (“I know it’s a little off-brand, but kids love it,” he joked.) The hats sold out less than two weeks later, bringing the total hat sales to $1 million, and it fell to Davis to make good on the flamethrower promise.
“I remember that day quite clearly because he called me and said, ‘We’re going to sell flamethrowers,’ and I’m like, cool,” Davis says. After a quick Google search though, he realized there was a small problem – quite a few countries don’t let you ship flamethrowers to their citizens.
“Great, call it Not a Flamethrower,” Musk replied. That was technically true; it was really just a roofing torch with an air-rifle cover.
“When do you want to start?” Davis asked.
“Tonight.”

So Davis built a pre-order page on Squarespace, listing the devices at $500 each, and Musk tweeted the link. “Obviously, a flamethrower is a super terrible idea. Definitely don’t buy one,” he warned, “unless you like fun.” All 20,000 sold out in four days, bringing in $10 million. Then Davis had to figure out how to build them. “We literally rented out a warehouse in LA and set up a production line,” he remembers.
Merchandise was the company’s only revenue at this point, but the team was making progress on real-world transit projects too. In January 2018, the company told the Culver City council that it had finished its technology development phase and was ready to build a 6.5-mile tunnel between Los Angeles and Culver City. Instead of the private Teslas from the TED video, the pitch featured small pods that would complement the county’s existing transit, require no taxpayer money, and cost riders about as much as a $1.75 Metro bus ride. In February, Chicago Mayor Rahm Emanuel announced that the Boring Company was one of four groups bidding to build a high-speed link between downtown and O’Hare International Airport.

With bids in Washington, Los Angeles, and Chicago, it was increasingly clear the Boring Company needed to stand on its own. In early 2018, it was spun out of SpaceX, and in April it raised $112.5 million in equity, about 90% of it from Musk and the rest from early employees. SpaceX later received about 6% of the company for the land, people, and equipment it had contributed.
The next month, Musk and Davis hosted an information session in Los Angeles, with Gary crawling around a pineapple-shaped terrarium between them. Much of the evening was spent addressing common misconceptions about tunneling.
Won’t you run into pipes and power lines? No, Musk said, because the tunnels run beneath them. “Once you get below about 20 or 30 feet, there’s almost nothing.” Even L.A.’s subway wouldn’t be in the way. Boring’s tunnels would run below it.
Will the tunnels disturb people on the surface? Davis offered a rule of thumb: “If you’re two tunnel diameters below the surface, noise and vibrations are imperceptible.” With a 14-foot machine, that’s 28 feet, so the company always tunnels at least 30 feet below the surface. Walking on the sidewalk, Musk added, creates more vibration than the boring machine below.
What about earthquakes? “Earthquakes are probably the single biggest question we get,” Musk said, “but people don’t realize that earthquakes are primarily a surface-level phenomenon.” Being in a tunnel during one, he explained, is like being in a submarine during a hurricane at sea. “The safest place to be during an earthquake is a tunnel,” Davis added. Musk pointed out that Japan, one of the most seismically active countries on Earth, has one of the best subway systems in the world.
What about sinkholes? This is not a factor either, Musk says, because concrete segments lock in the shape the machine cuts, so the volume of the ground above doesn’t change.
What about induced demand? Economists have found that driving in American cities rises roughly one-for-one with new highway lanes, and drivers often endure years of construction only to find traffic hardly any different when the project is complete. Musk argues that tunnels escape this trap because you can keep adding lanes faster than demand grows. “You could have hundreds of lanes,” he says, “and there’s no real limit.”

Musk and Davis promised that the Boring Company would be a good neighbor. The only sign the company was there, Davis said, would be dirt trucks on approved routes during approved hours. “It’s not going to scale if we can’t do a 2-mile tunnel without disturbing people,” Musk added. “Then how could we do hundreds of miles without disturbing people?” Eight years later, Davis says, the sensors at the surface above the company’s tunnels have never recorded more than about two hundredths of an inch of movement.
Delete, Delete, Delete
In December 2018, Musk invited visitors to see what had become of the hole in SpaceX’s parking lot. It was now a finished, 1.14-mile test tunnel that ran under 120th Street in Hawthorne. The all-in cost with Godot was about $10 million, a figure that included the lighting, ventilation, safety systems, and track. The unveiling had the feel of a Boring Company party. There was a Monty Python-style watchtower built from bricks of compressed tunnel dirt, marshmallows roasted with Not a Flamethrowers, and a shrine to Gary. Then came a ride through the tunnel in a Tesla Model X fitted with retractable guide wheels. “You drop down the wormhole, pop out the other side,” Musk explained, and then you simply drive away.

The electric skates from the TED video were gone. Musk said they’d proven too complex. The elevators that were supposed to lower cars into the tunnel turned out to be slower and more expensive than letting the boring machine dig its own ramp to the surface, a technique the team would call “porpoising.” The passenger pods would also be deemed unnecessary, and eventually even the guide wheels in the test tunnel demo would be removed. Critics accused the company of backtracking, but Davis and the team were really just applying the first three steps of the Algorithm they learned from Musk at SpaceX:
Question every requirement.
Delete any part or process you can.
Simplify and optimize.
They applied it to the boring machines too. “We did not know how to dig tunnels at all, actually,” Musk admitted in 2018. “Godot was a lot of waiting and figuring things out.” The big time sinks were stopping drilling to install the concrete lining segments, hauling dirt out and supplies in through a single-lane tunnel, and stringing miles of heavy high-voltage cable behind the machine.
The next machine, Line-Storm, made progress on most of these problems. (In keeping with the company’s habit of naming its machines after poems and plays, it took its name from Robert Frost’s “A Line-Storm Song.”) It was a heavily modified conventional machine, and the key change was rethinking the assumption that digging and lining the tunnel had to be separate steps. By automating the placement of the concrete segments, Line-Storm was a big step toward continuous mining, where the machine digs and builds the tunnel at the same time. Other fixes were much simpler, like adding a passing lane so concrete segments could come in while dirt went out. By Davis’s estimate, the passing lane, a faster way to dump muck, and automated segment placement would triple the machine’s speed on their own.
The team electrified everything else it could, borrowing heavily from Tesla. Industry precedent was to haul dirt out of tunnels with diesel locomotives, which required a ventilation duct. Boring built a battery-electric locomotive out of Model 3 motors and battery packs that could pull a quarter of a million pounds of muck. Musk even wanted to get rid of the power cable that trailed behind the boring machine. “Two miles of high-voltage cable is pretty expensive,” he said. “We want to make this more like a cordless drill.”

The machine’s length got the same treatment. Behind the cutting head, a boring machine trailed a long string of sections called gantries. They’re kind of like the cars of a train, and Davis describes them as “basically just storage bins.” This set off what Davis calls an “engineering death spiral,” because “every single person wants to add something, and then you have an infinitely long machine.” The first machine the team worked with had 14 of these sections. “So I was like: delete them all.” The next one had only five. “We deleted too much, which is classic.” Musk’s rule is that if you don’t end up adding back at least 10% of what you deleted, you didn’t delete enough. Boring’s machines now run with six to eight.
Teslas in Tunnels
Boring’s first customers were some of the hardest cities in America to tunnel under. In June 2018, Chicago picked the company to build an express link between downtown and O’Hare, but the deal fell through after Mayor Rahm Emanuel left office without finalizing the contract.
Elsewhere, Boring’s plans got buried in paperwork. The Washington-Baltimore project, and the New York-to-D.C. hyperloop meant to follow it, stalled for years in federal environmental review before the highway regulator declined to approve it. In Los Angeles, a lawsuit over the environmental review of a single test tunnel under Sepulveda Boulevard was enough to make the company set its plans aside. Musk had picked L.A. because he lived there, but he later called it “one of the worst places to dig tunnels, mostly because of the paperwork.” Davis says the problem isn’t the people but the system. “It was the same thing when SpaceX was launching,” he says. “The system was set up for a six-month launch permit, not a two-day one. Nobody is set up to permit Loop, because how could they be?”
Las Vegas was different. In 2019, the Las Vegas Convention and Visitors Authority needed a way to move people across its expanding convention campus, and Boring offered to do it underground for about $52.5 million, roughly a quarter of the original estimate for an above-ground people mover from the Austrian firm Doppelmayr. Boring also assumed the financial risk. Two-thirds of the money would be held back until the system worked, and if it never earned a certificate of occupancy, the authority would get its money back. But not everyone was convinced. Las Vegas Mayor Carolyn Goodman, a board member, opposed the deal, citing the company’s record of completing zero commercial projects, and cast the only vote against it.
Tunneling began on November 15, 2019. The original Loop of three stations connected by 1.7 miles of tunnel was built in about a year for roughly $47 million, without a single road closure. And it has carried passengers since 2021, with a demonstrated peak capacity of more than 4,500 passengers an hour and more than 32,000 in a day. It has since become the first piece of a much bigger network, the Vegas Loop, which now has 29 stations and more than 10 miles of tunnel. Local officials have approved plans for 68 miles of tunnel and 123 stations.

Vegas was the ideal proving ground because of how the Loop product works. A subway has to stop at every station, so the more stations you add, the slower every trip gets. “There’s only two ways to solve the problem,” Davis says. “You don’t stop, or if you’re willing to stop, you can’t have that many stations.” Loop is point-to-point and never stops. It works more like a private Tesla on its own highway that takes you straight to your destination. Davis estimates that if a train stopped at every hotel, getting from one end of the Strip to the other would take about 90 minutes, but on Loop, it takes about eight, no matter how many stations are in between.

Another advantage of a point-to-point system is that stations can be small and cheap. A subway station can cost $250 million to $500 million, but the Boring Company’s stations range from less than $1 million all the way up to $10 million for the really big, fancy ones. Some are as small as a few parking spaces and cheap enough that even a taco shop on the Vegas Loop is buying one. “By doing so, this taco station is going to be on our map and will have point-to-point transportation to anywhere in Vegas,” Davis says. That has been part of Musk’s pitch since 2018: “If you have many small stations that are not much larger than a parking space, then you could really seamlessly integrate a new transport network into a city.”
And as the network grows, so does a station’s return on investment. In 2022, Resorts World converted part of its underground garage into a station, so guests can get to the convention center in just over a minute. Since then, the resort has seen demand rise about 30%. Stations can also be added after a tunnel is built without slowing anyone down. “The last thing you do is pop a hole in the tunnel and that becomes your exit,” Davis explains. The only catch is that the tunnel has to run under the property first, which is why Boring asks every property owner along a route whether it wants the tunnel to clip its land or go around it. “Almost anyone who is being rational will say, of course, clip my property,” he says. Several customers have told the company they’ll open a station once 20 others are open. “Great,” Davis replies. “Once there’s twenty open, we’ll cut a hole in the tunnel.”

Davis treats the ride experience as an engineering problem too. When the convention center Loop opened, he spent two months standing in the station writing down notes. He timed riders getting in and out of different cars, and decided that the fleet would be Model Ys after observing that people got out of a Model Y two or three seconds faster than a Model 3. “Capacity is driven by how fast you can load and unload,” he says. Asking riders to leave the doors open for the next passenger raised capacity by 10 to 15%. Some stations now have no staff at all – you press a button and a car comes. Davis says it took about twenty iterations before people stopped getting confused.
The business model was another area of innovation. A new transit line makes the land around its stations more valuable, but in most American cities, taxpayers pay to build it, fares cover only part of the cost to run it, and the property owners who gain the most usually don’t pay anything extra. Boring flips that. Property owners who want a station pay for it, riders pay fares that run about half the price of a rideshare or less, and the company has been buying land along its routes, so it shares in the value its tunnels create. “We don’t ask for any taxpayer dollars,” Davis says, and he credits low tunneling costs for making that possible. That gives Boring very different incentives from those of a contractor bidding on a government project. It’s also an old idea. Developers did something similar more than a century ago, paying for streetcar lines to reach the neighborhoods they were building as people moved to the suburbs.
There are also network effects. “The Loop product is pretty good and it’s obviously getting better,” Davis says, “but the system’s usefulness is limited by the number of tunnel miles we have and are open.” A hotel connected only to the convention center is worth less than one connected to restaurants, arenas, the airport, and every other hotel. Each new station gives riders more places to go and gives property owners another reason to join. And as the tunnels get cheaper, more destinations will make sense.
A Factory on Wheels
After Line-Storm came Prufrock, the first machine Boring designed entirely in-house, named after the T.S. Eliot character who spends an entire poem wondering whether he dares disturb the universe. It had been in development since 2018, and the goal was a machine that would finally be an order of magnitude faster and cheaper than a conventional one.

The company builds these machines in Bastrop, Texas, where it has been headquartered since 2023. Ten years ago, Bastrop had just over 8,000 residents. Today the town also holds a SpaceX Starlink factory and the headquarters of X. Davis lives on site, in one of the 18 homes on the campus. “I would always ask my wife and all the families: What do we not have here?” he says. Whatever they named, he added to the campus. The nearest grocery store was half an hour away, so now there’s one on campus (the Boring Bodega) inside a metal building called Hyperloop Plaza, along with the Prufrock Pub, a barbershop, and a café where employees make pottery. Outside are pickleball and tennis courts with a view of the factory, and a Montessori school is on the way.

Across the road is a 420-by-169-foot factory that’s long enough for an entire boring machine to fit in a line. “It took all of one minute to size this building,” Davis jokes. About 200 people work there in two shifts, seven days a week. The factory builds a boring machine every two months and is working toward a cadence of one a month. On Prufrock-1, the original prototype, Boring made about 30% of the parts itself. On Prufrock-6, it makes about 99%.
Everyone works under the same roof – a lesson Davis learned from Musk at SpaceX. Design engineers sit next to the machinists who make the parts and the technicians who assemble them. “Before the part is even released, the engineer brings a preliminary drawing and says, ‘Hey, what do you think?’” says Mike Baier, who runs production, electrical, and software engineering. He points out that if that part were sent to an outside shop instead, every question about cost or delay would have to pass through a chain of three or four people. Each machine has roughly 176,000 parts and carries about a thousand design changes from the previous version (e.g. Prufrock-4 to Prufrock-5), which is why Baier calls them “late-stage prototypes.”
“Imagine building a two megawatt factory, which is hard enough,” Davis says, “and then it moves 100-200 feet every day and it just has to keep running.”
A boring machine does not leave Bastrop until it has dug a test tunnel there. “We’re the only company that I’m aware of that won’t ship a machine until it’s dug a test tunnel,” Davis says, a habit he brought from SpaceX, where no rocket launches without a static fire. During those tests, the engineers who designed each subsystem of the machine operate it themselves, watching what works and making small changes on the spot.
The fourth step of Musk’s Algorithm is to accelerate, and for a boring machine that means eliminating every moment it isn’t moving forward.
The way a Prufrock machine works is it uses twelve hydraulic propulsion cylinders to push itself forward five feet, builds a ring of concrete lining, then pushes again. “I consider [building the concrete lining] to be a failure because you’re not pushing,” Davis says. To mine while building the ring, it has to retract some cylinders to place a segment while the rest keep pushing, using force controllers to keep the thrust pointed exactly where it should go. Davis calls it a dance of the cylinders. “On paper it all works immediately,” he says, but making it work safely and reliably took about two years and is an important step from a machine that mines 5 to 10% of the time to one that mines more than 90% of it.

Dirt and muck are extracted with a continuous conveyor belt that runs the length of the tunnel and back, about five miles of rubber for the 2.3-mile tunnel Boring just finished in Las Vegas. Every 500 feet, the crew has to splice in a new spool of belt, which used to take eight hours. When Davis looked into why, he found that most of the delay was due to human factors and everyone being packed on top of each other. The new conveyors now have big splice platforms with every tool within reach, and a team called “fast splice” got the job down to nine minutes.
“Every time you stop, ask why,” Davis says. Even the rollers that keep the belt on track were rethought from first principles. Watching crews knock them into place with hammers, Davis realized that the industry’s method had been adopted “because of inertia and heritage instead of engineering,” and the team is now building a belt that steers itself. Every foot the machine mines fills about one dump truck with muck.

The slowest part of starting a tunnel, though, was the launch. Conventional boring machines are lowered into enormous pits, like the 100-foot shaft Musk and Davis descended into in Washington. Boring’s machines arrive on a transporter called “the Monster,” which tilts them nose-down so they bore straight into the ground. “It saves millions of dollars on every launch,” Davis says. At the other end of the tunnel, the Monster tilts the other way, and the machine climbs out of the earth onto it, “Dune style.” The fastest launch so far went from an empty site to mining in six and a half days.

Avoiding downtime is also a crucial factor in tunneling speed. A few weeks after Mike Baier joined the company, Davis told him he wanted to be able to cut any wire on a Prufrock and have it keep mining. So the team redesigned the electrical system with redundancy, and “for the most part,” Baier says, “Prufrock now passes the scissor test.”
The final step in Musk’s Algorithm is automation. One of the first questions Musk had as he learned more about tunneling was: Why are there so many people? Oil and gas companies drill for miles, thousands of feet below the surface, with zero people in the hole. Once a person is inside, Davis says, “your entire machine becomes a safety system.” Deleting people from the process lets Boring delete other things too. For example, with zero people in the tunnel, you don’t need the ventilation system that feeds them air.
Boring’s first machine needed about a dozen people, but the newest Prufrock needs only two. Crews used to haul barrels of grease into the tunnel by hand, so the team copied the quick-disconnect fittings SpaceX uses on its launch pads, and now a driverless truck plugs in, delivers the grease, and leaves. “Getting to two is hard, but not that hard,” Davis says. Getting to zero will be much harder, but once it’s figured out, “you can just unleash a fleet of these things.”
In May 2025, the company demonstrated “Zero-People-in-Tunnel” (ZPIT) continuous mining, advancing the machine and erecting a full tunnel ring at the same time with no crew inside the tunnel. By August 2026, ring building was fully autonomous: six concrete segments of about 3,750 pounds each were placed with millimeter precision in less than a minute.
All of it is operated from the company’s Global Operations Control Center (GOCC) in Texas, which the team pronounces “gawk.” Even a machine being tested outside the factory is run over a network loop routed through Houston, so operating it is identical to operating one anywhere in the world. Each machine carries about 80 cameras and more than 1,700 sensors, and the team operates each subprocess of tunneling – from steering the boring machine to the liner trucks feeding the concrete segments – by remote control before automating any of it. That gives every automation a graceful fallback to a human, so it doesn’t need to be perfect on day one. For the liner trucks, the operator today mostly holds down a button and watches the cameras.

The result of all these improvements is a tunneling cost that’s below $10 million a mile, and Davis expects that to converge on $4 million to $5 million. For context, adding a single lane to a big-city highway can cost tens of millions of dollars per mile. Los Angeles, for example, spent $1.1 billion and five years adding a single 10-mile carpool lane to the 405.
Beyond Nevada
In July 2025, the company and the State of Tennessee announced the Music City Loop, a privately funded system connecting downtown Nashville with Nashville International Airport. It wasn’t an easy place to dig – the rock is extremely hard, and the depth to the top of the rock changes every few hundred feet. But after receiving its permit from the Tennessee Department of Transportation on February 25, 2026, the company started tunneling the same day. By August, two machines were mining simultaneously, and a third was in final assembly in Bastrop.
The hard rock turned out to be a gift. Nashville’s limestone tests at about 14,000 psi, stronger than the 8,000-psi concrete Boring uses to line the tunnels. “Nashville and the evolution of the planet has basically built the liner for us,” Davis says. In many of the parking garages beneath Broadway’s high-rises, there’s nothing between the support columns but bare limestone – the rock is the wall. So the team designed a new machine, the Prufrock Main Beam, and deleted the most complicated systems from their soft-ground machine: no screw conveyor, no grout, no concrete segments, and none of the machinery that was required to move and place them. “That machine looks nothing like our normal Prufrock 1s,” Davis says. “It’s a way simpler machine.” With no lining to push against, it braces itself against the tunnel walls with more than four million pounds of force – Davis compares it to an acrobat hovering inside a tube by pushing out against both sides – then mines forward, releases, and grips again.

“For all these systems,” Davis says, “you’re taking Elon’s Algorithm – the best part is no part – and applying it.” Tunneling in Nashville costs about half what it does in Vegas, and stations are simpler too. “All I have to do is pop a hole in the rock I’m drilling through anyway,” he says, “and I’m in their parking garage.”
Davis thinks about the project in terms of “pushes.” Each push is five feet, and twin nine-mile tunnels to the airport plus twin two-mile tunnels under Broadway add up to more than 20,000 of them. If you can shave a minute off every push, you save two weeks of schedule. “Whenever we see something go wrong – which is all the time – we write it in an Excel spreadsheet, assign an engineer to it, and we fix it,” Davis says. The machines are already about five times faster than when they started, and he expects another tenfold gain. His bet is that the first mile of Nashville, which includes the convention center and four major properties, will carry real passengers in the first quarter of next year while the machines keep boring toward the airport. The U.S. Department of Transportation puts the first phase, about 10 miles from downtown to the airport, at roughly $200 million, or about $20 million per mile of route (each route has two one-way tunnels, so about $10 million per mile of tunnel). Boring expects to finish all 22 miles of tunnel in two to three years. The average American light rail line costs about eight times as much per mile of route and takes about twice as long per mile to build.

Internationally, the company signed a construction contract with Dubai’s Roads and Transport Authority for the pilot phase of the Dubai Loop – about four miles of tunnel and four stations that will cut the trip between the city’s financial district and Dubai Mall from 20 minutes to three. Tunneling is planned to begin in late 2026, and its broader partnership with the UAE is for more than 150 kilometers (~93 miles) of tunnel.

Closer to home, Dallas and New Orleans won the company’s Tunnel Vision Challenge this spring – Boring offered to build a free, one-mile tunnel to whoever pitched the best idea – and Orlando is working through early approvals.
Each new city should be easier than the last. When Las Vegas signed on, Boring had never completed a commercial project. Now any city considering Loop can see that it works, which takes much of the risk out of saying yes. Permitting is getting easier too. Las Vegas’s full system requires more than 600 permits, but in Nashville, Davis explains, “the entire thing was permitted in one shot.”
In September, the company announced a $3 billion Series D at a $23 billion valuation, which a16z participated in.
The Future Underground
In 1964, Isaac Asimov toured the New York World’s Fair and imagined its successor fifty years on. He called an underground house at the fair a sign of the future and predicted that exhibits would one day show off entire underground cities. He also foresaw vehicles with “robot-brains” that could be told where to go and would get there without the slow reflexes of a human driver. Asimov imagined them as two separate predictions, but the Boring Company believes tunnels and autonomous cars will go hand in hand.
“Tunnels under cities with self-driving electric cars will feel like warp drive,” Musk tweeted in 2020. Today, 5 to 10% of Loop rides already run without a driver. A sealed, one-way tunnel with no pedestrians, no cross traffic, no weather, and no left turns is about the friendliest place an autonomous car will ever drive. And as tunneling gets cheaper, there will be more of them, and the whole network will get more useful. Davis believes that in the extreme, you could have a station in every driveway – a robotaxi picks you up anywhere, drops into the nearest tunnel portal, and comes back up at your destination.
A city like that would also be far safer. Worldwide, car crashes kill more than a million people a year and are the leading cause of death for children and young adults. We think the most promising way to bring that number down is autonomous cars and tunnels. Autonomous cars remove human error, the most common cause of crashes, and tunnels remove many of the most dangerous parts of driving, like pedestrians, cross traffic, and bad weather.
The obvious worry is what happens if something goes wrong in one of the tunnels. Because the tunnels are one-way, head-on crashes are impossible, but if a car does stall or catch fire, cameras will see it, and fire detectors will sound an alarm. Water lines run the length of the tunnel for firefighters, ventilation can push smoke in either direction to keep an escape route clear, and a dedicated communication system keeps police and firefighters in contact underground. Davis says the company runs about one safety drill a week with firefighters and police. Clark County’s rules cap how far a tunnel can run between exits. “I’m often told we don’t have [exits],” Davis says, laughing. “But we have dozens and dozens of them.” And since normal fire engines are too big for the tunnels, Boring has outfitted the county fire department with Cybertrucks it converted into custom fire trucks. So far, more than 4 million passengers have ridden the Loop without a single passenger-related safety or security incident.

Moving traffic underground also frees up land for better use. By the company’s estimate, roads take up 20% to 30% of the land in many major cities, and as much as 60% including parking.
Davis would put utilities underground too. In his view, the pipes and cables under city streets – water, sewer, power, fiber – all belong in tunnels. When a water pipe breaks, it would flood a tunnel instead of a street, and repair crews could walk in to fix it instead of tearing up the road. Davis says the company has never had to close a road to build its tunnels, and while a machine bores three miles away, the only sign of it on the surface is a launch site with dump trucks of dirt coming and going. “You basically have this major infrastructure project that nobody even knows is happening,” he adds.
Then, once tunnels are cheap, they can start connecting cities. The other big question for a hyperloop was how well a tunnel can hold a vacuum. “We didn’t know,” Davis says. “We could do a bunch of fake analyses, or we could just do it, and that’s what we did.” The former SpaceX engineers sealed a tunnel, hooked up the vacuum pumps from their old California hyperloop test track, and pumped it to see how well it worked. “You can easily get it below 1 psi,” he says. “When you start getting to 0.5 it starts to struggle.” For context, you can still reach very high speeds at 7 to 10 psi, compared with about 14.7 psi at sea level. And the leak rate isn’t too bad, which makes sense given that they already seal the tunnels to keep out groundwater.

In September, Musk wrote that the company is working on a “simple, precursor Hyperloop tunnel” between Austin and San Antonio that can travel at speeds greater than 200 miles per hour. “Due to traffic congestion, that journey can currently take up to 2.5 hours,” he continues, but the Boring Company can reduce that to “a consistent <30 mins.” And if Nashville goes as well as Davis expects, it could be linked by hyperloop to Knoxville or Chattanooga – or even as far as Memphis or Atlanta. “With the mining speeds and the permitting and regulatory structure we have now, there’s a very clear path to doing that,” he says. “Most of the risk is us executing, which is great from the hyperloop perspective.”
The most important consequence of this underground future may be the cost to access a city. “The big win for helping people, in my opinion,” Davis says, “is one of the fundamental issues with cities besides traffic is rent. It’s super high because no one wants to live two hours away from their job.” But if tunnels can shorten a two-hour commute to ten minutes, affordable land in the middle of nowhere becomes land near everything. Las Vegas is already becoming an “eight-minute city.” Every trip on the Loop takes eight minutes or less, and each new tunnel expands the area you can reach in that time. “You can solve the rent and cost of living problem by making all of these places very desirable to live with just transportation,” Davis says. Boring and some developers are already betting on it. In Las Vegas, Boring is planning a six-story, 132-unit apartment building about a mile from the Strip with a Loop station on its ground floor. In Nashville, a developer has signed on for a private station serving residents of three of its towers on Church Street. Farther out, Boring says developers are starting to consider projects 10 to 20 miles from the urban core as tunnels put downtown within reach.
Las Vegas will be the first test. Once the core of the Vegas Loop is built, Davis plans to run tunnels out to the neighborhoods around it. “The only reason we can do that is because our cost of tunneling is low,” he says. A pair of tunnels – to and from – a neighborhood 10 miles away would cost $20 billion at a billion dollars a mile, but at $10 million a mile, it costs $200 million. “That’s a very different game.”
Somewhere at the end of all those tunnels is a half-hour commute from D.C. to Vegas, and until it exists, Steve Davis isn’t going anywhere.


This newsletter is provided for informational purposes only, and should not be relied upon as legal, business, investment, or tax advice. Furthermore, this content is not investment advice, nor is it intended for use by any investors or prospective investors in any a16z funds. This newsletter may link to other websites or contain other information obtained from third-party sources - a16z has not independently verified nor makes any representations about the current or enduring accuracy of such information. If this content includes third-party advertisements, a16z has not reviewed such advertisements and does not endorse any advertising content or related companies contained therein. Any investments or portfolio companies mentioned, referred to, or described are not representative of all investments in vehicles managed by a16z; visit https://a16z.com/investment-list/ for a full list of investments. Other important information can be found at a16z.com/disclosures. You’re receiving this newsletter since you opted in earlier; if you would like to opt out of future newsletters you may unsubscribe immediately.
Source: a16z News · a16z.news