In one minute

• Tunnels, not flying cars. Musk says a 3D network of tunnels can absorb any level of urban congestion because you can keep adding levels; cars ride on an electric "skate" at up to 200 km/h (~130 mph), entering via an elevator built into two parking spaces 1:18.
• The cost problem is solvable arithmetic. LA's subway extension ran ~$1B/mile; he claims a 10x cost cut from halving tunnel diameter (26–28 ft → 12 ft, a 4x area reduction), continuous tunneling-plus-reinforcing (2x), and more machine power (another 2–5x) 3:12.
• Autonomy will make traffic worse, not better. Shared autonomy makes point-to-point car travel cheaper than a bus ticket, so total driving goes up 11:33.
• Tesla Autopilot is cameras-only (plus radar), no LiDAR. Eight cameras per car; his target was LA-to-New York fully autonomous with no controls touched by Nov/Dec 2017, and safe hands-off sleeping in ~2 years 14:03.
• Solar roof economics: cost of solar glass roof + electricity is claimed to be less than a normal roof + electricity; he expects it to be unusual to have a non-solar roof ~15 years out 23:43.
• Roughly 100 Gigafactories would move the world off fossil fuels; locations for 2–4 more to be announced "later this year," probably four 27:39.
• Why Mars: sustainable energy is "largely inevitable," a space-faring civilization is not. "I'm not trying to be anyone's savior... I'm just trying to think about the future and not be sad" 39:45.

What was covered

Why dig tunnels: the 3D network under LA 0:31

Musk called traffic "one of the most soul-destroying things," affecting people everywhere and particularly bad in LA 0:31. He showed the first visualization of the concept 1:18:

• Entrances and exits integrate into the city using two parking spaces and an elevator carrying a car skate.
• No speed limit in the tunnel; designed for 200 km/h / ~130 mph.
• Claimed trip time: Westwood to LAX in five or six minutes (the transcript renders the destination as "Ali-X" — unclear, but LAX from context) 2:07.
• Likely initially operated on a toll-road basis, which also pulls traffic off surface streets 2:17.

His key rebuttal to induced demand: unlike building up, there is no limit to how many tunnel levels you can add — "the deepest mines are much deeper than the tallest buildings are tall" — so a 3D network can alleviate "any arbitrary level of urban congestion" 2:26.

Making tunneling 10x cheaper 3:12

Musk agreed digging is currently very expensive: the LA subway extension was roughly a 2.5-mile extension for $2 billion, about $1 billion per mile, and "this is not the highest utility subway in the world" 3:12.

His path to at least an order-of-magnitude improvement:

1. Cut tunnel diameter. Regulations put a single road-lane tunnel at 26–28 ft (crashes, emergency vehicles, ventilation for combustion engines). A 12-ft tunnel is plenty for an electric skate. Halving diameter cuts cross-sectional area by 4x, and tunneling cost scales with cross-sectional area — "roughly a half-order of magnitude" 3:55.
2. Continuous tunneling and reinforcing. Current machines tunnel half the time and stop to install wall reinforcement. Doing both continuously gives a factor of 2 — combined, a factor of 8 4:23.
3. More power. Machines are "far from being at their power or thermal limit"; jacking up power could add a factor of 2 to 4 or 5 on top 4:52.

The internal benchmark is a pet snail called Gary — he attributed the name to SpongeBob SquarePants after misspeaking "South Park" — which is currently 14 times faster than a tunnel-boring machine. "Victory is beating the snail" 5:34.

Why not flying cars 6:10

Musk said he's "in favor of flying things" and has no inherent bias (he does rockets), but flying cars would be quite noisy and generate very high wind force. Large numbers of them overhead is "not an anxiety-reducing situation" — you'd be wondering whether someone serviced their hubcap or whether it's going to "come off and guillotine" you 6:10.

Hyperloop and the tunnel tie-in 7:14

• SpaceX built a Hyperloop test track next door for a student competition to encourage transport innovation. By volume it ends up as the biggest vacuum chamber in the world after the Large Hadron Collider. He framed it as "kind of a hobby thing" 7:14.
• They built a pusher car for student pods and planned to see how fast it goes unloaded — "cautiously optimistic" it beats the world's fastest bullet train in a 0.8-mile stretch. "It's either going to smash into tiny pieces or go quite fast" 8:08.
• Key technical point: tunnel walls are typically designed to resist the water table at about five or six atmospheres. Vacuum is only one atmosphere, so a tunnel built to resist groundwater automatically holds vacuum 8:23.
• Length: no real limit. For something like DC to New York, he'd go underground the entire way because it's high-density, under buildings and houses 9:13.
• Detectability: dug more than three or four tunnel diameters beneath your house, you cannot detect it at all. If you had a device that could, "you can get a lot of money for that device" from the Israeli military (detecting Hamas tunnels) and US Customs and Border Patrol (drug tunnels). Earth is "incredibly good at absorbing vibrations"; maybe a very sensitive seismic instrument could pick it up 9:13.

The Boring Company as a "hobby" 10:29

Musk said the tunneling project is 2–3% of his time 10:43, staffed "basically interns and people doing it part-time," using second-hand machinery, "puttering along, but it's making good progress" 10:54.

Why autonomy makes congestion worse 11:33

A lot of people assume autonomous cars go faster and reduce congestion, and "to some degree that will be true." But with shared autonomy, point-to-point car travel becomes cheaper than a bus ticket, so total driving rises and "traffic will get far worse" 11:33.

State of electrification across the industry 12:13

Almost every automaker has some EV program, but "they vary in seriousness" — some are serious about transitioning entirely, some are "just dabbling," and some are "amazingly" still pursuing fuel cells, which he thinks "won't last much longer" 12:39. Asked why he doesn't declare victory and move on, he said he intends to stay with Tesla "as far into the future as I can imagine," citing Model 3 and the Tesla Semi 13:13.

Autopilot: cameras, not LiDAR 14:03

• The demo video used only cameras and GPS — no LiDAR, no radar — i.e. passive optical, "essentially what a person uses." The whole road system is meant to be navigated with passive optical, so "once you solve cameras or vision, then autonomy is solved. If you don't solve vision, it's not solved" 14:03.
• Tesla's focus is a vision neural net effective for road conditions. On others going the LiDAR route: he wants cameras plus radar as most of it, and says "you could probably do 10 times better than humans with just cameras" 14:44.
• Cars being sold then had eight cameras 14:55.
• Timeline claim: still on track to drive LA to New York fully autonomous by the end of the year — "November or December of this year" — parking lot in California to parking lot in New York, no controls touched at any point 15:08.
• He was "fairly confident" the route could be changed dynamically in real time (LA to Toronto instead of New York; Seattle, Florida), because once on a highway it should handle the whole highway system of a country — not just one memorized route 16:03.

When can you sleep in the car 16:50

About two years, technology aside from regulation 17:07. His framing: the hard part isn't 99.9% — if a car crashes 1 in 1,000 times "you're probably still not going to be comfortable falling asleep... certainly you shouldn't be." No system will be perfect; the bar is something like "unlikely to crash in 100 lifetimes or 1,000 lifetimes" 17:07. He added that vehicle safety is probabilistic — the chance a human driver has an at-fault accident is "never zero" — so the real question is how much better autonomy needs to be than a person before you rely on it. He expects the autonomy system to "at least mitigate the crash, except in rare circumstances" 18:05.

Shared autonomy fleet 18:40

Asked whether you could buy a car, send it out Uber-style and effectively cover your lease: "Yeah, absolutely this is what will happen." Owners will choose to use the car exclusively, restrict it to friends and family, restrict to other drivers rated five-star, or share it sometimes and not others. "That's 100 percent what will occur. It's just a question of when" 19:01.

Tesla Semi 19:34

• Planned announcement in September; only a teaser shot was shown 19:34.
• Heavy-duty, long-range — highest weight capability with long range — aimed at heavy-duty trucking loads, which people don't think is possible for electric (power and range doubts) 20:10.
• Claim: it will out-torque any diesel semi, and in a tug-of-war it will pull a diesel semi uphill 20:10.
• Short term these are trucks drivers want to drive. The electric motor gives a flat torque-RPM curve (vs. a "hill" for internal combustion), making it "a very spry truck" you "could drive around like a sports car," with no gears — single speed 21:10.
• Musk said he'd already driven the test prototype around a parking lot: "it's really weird, because you're driving around, and you're so nimble, and you're in this giant truck" 21:40.

Solar roof and Powerwall 22:18

The picture of the future he showed: electric car in the driveway, three Powerwalls stacked against the side of the house, and an actual solar glass roof — on what he conceded was "a real fake house" 22:18.

• Solar glass tiles allow texture and color adjustment at a fine-grained level 22:57.
• Microlouvers in the glass mean that from street level all tiles look identical whether or not there's a solar cell behind them; from a helicopter you could see which ones have cells 22:57. Cells go in the tiles likely to see a lot of sun 23:43.
• Economics: "very confident" that solar glass roof + cost of electricity < normal roof + cost of electricity — "economically a no-brainer" 23:50.
• Durability: they considered an infinity warranty but worried it would "sound like we were just talking rubbish." It's toughened glass — "well after the house has collapsed... the glass tiles will still be there" 23:50.
• Rollout: starting with two roofing types, with the second two early next year 24:45.
• Adoption timescale: eventually almost all houses, on the order of 40 or 50 years, since roofs are replaced every 20 to 25 years on average. Fast-forward 15 years and "it will be unusual to have a roof that does not have solar" 24:59.
• Most houses in the US have enough roof area to power all the needs of the house, though it depends on location and house size relative to roof area 25:56.

Gigafactory and the 100-factory math 26:11

• The Gigafactory is a bet on falling lithium-ion battery prices, roughly doubling world supply — described as almost Tesla's core competency 26:11.
• Physical detail: overall diamond shape when fully done, aligned on True North 26:44.
• Capacity: 100 gigawatt-hours of batteries a year, "probably more" 27:05. Cells are already in production; the video shown was "the slowed-down version" — at full speed "you can't actually see the cells without a strobe light" 27:19.
• How many to get the world off fossil fuels: about 100. "It's not 10, it's not 1,000, most likely 100." The interviewer put the cost at roughly $5 billion, "maybe the next one, five to ten billion dollars" 27:39.
• Announcement plan: locations for between two and four Gigafactories later this year, probably four. On geography he would only say, "We need to address a global market" 28:25.

The one politics question 28:55

Asked about advising a president who says he doesn't believe in climate change, Musk said he is on two advisory councils whose format is "going around the room and asking people's opinion," with a meeting every month or two — "that's the sum total of my contribution." He said he has used the meetings to argue in favor of immigration and in favor of climate change, which "wasn't on the agenda before": "maybe nothing will happen, but at least the words were said" 29:32.

SpaceX reusability 30:26

• Footage shown was a booster returning from high and fast in space after delivering the upper stage at roughly Mach 7 30:50.
• On how many landings: the interviewer said five or six, Musk said "four or five," then "eight or nine or something" — the transcript is unclear on the count 31:20.
• They landed a booster, prepped it and flew it again — "the first reflight of an orbital booster where that reflight is relevant" 31:37.
• Core principle: "reusability is only relevant if it is rapid" — rapid and complete. "You do not send your aircraft into Boeing in between flights" 31:37.

The Mars vehicle and interplanetary transport system 32:09

• Scale: roughly a 40-story skyscraper, "maybe a little more" 32:35.
• Thrust: about four times the thrust of the Saturn V 32:42. His comparison: a 747 is about a quarter-million pounds of thrust, so this is "the thrust equivalent of 120 747s with all engines blazing" (the intermediate arithmetic in the transcript — "for every 10 million pounds of thrust, there's 47" — is garbled) 33:00.
• Payload: it could take a fully loaded 747 — maximum fuel, passengers and cargo — to orbit as cargo 33:26.
• Timeline: for people walking into the rocket, "I'm hopeful it's sort of in the eight- to ten-year time frame. Aspirationally, that's our target. Our internal targets are more aggressive" 34:09.
• He expects future spacecraft to "make this look like a rowboat" — "truly enormous" 34:09.

Why Mars, and why technology is not automatic 34:43

• Motivation: "it's important to have a future that is inspiring and appealing... there have to be reasons that you get up in the morning and you want to live." A future that doesn't include "being out there among the stars, and being a multi-planet species" he finds "incredibly depressing" 34:56.
• On the "distraction" objection: he looks at the future as "a branching stream of probabilities," where actions accelerate or slow particular outcomes 36:00.
• Sustainable energy is inevitable — "tautological": unsustainable energy runs out and economics drives civilization to sustainable energy. So Tesla's fundamental value is how much it accelerates that, hopefully "by a decade, potentially more than a decade" 36:37.
• Becoming a space-faring civilization is NOT inevitable. The trend has gone backwards: 1969 the US sent someone to the moon; the Space Shuttle could only reach low-Earth orbit; after retirement the US could take no one to orbit — "the trend is like down to nothing" 37:04.
• "People are mistaken when they think that technology just automatically improves. It does not... It only improves if a lot of people work very hard to make it better," and by itself it will degrade. Examples: ancient Egypt forgot how to build pyramids; the Romans forgot how to build aqueducts 38:00.

Motivation: humanity's good plus something exciting 38:40

The interviewer summarized Musk's pattern as a double motivation — long-term good for humanity plus doing something exciting — citing sexy electric cars, beautiful solar roofs, and "your newest thing," a brain-machine interface to guard against bad AI and give "infinite memory and telepathy" (the interviewer raised this; there was no time to discuss it, and Musk did not address it) 38:40.

Musk's response: "I think the value of beauty and inspiration is very much underrated, no question. But I want to be clear. I'm not trying to be anyone's savior... I'm just trying to think about the future and not be sad" 39:45.

Questions from the room

All questions came from the interviewer (labelled Speaker 4 in the transcript; no separately identifiable audience questions were captured).

• Why are you boring? — To dig under LA and begin a 3D tunnel network to alleviate congestion, the "most soul-destroying thing" 0:31.
• Isn't digging incredibly expensive — doesn't that block the idea? — "They're right": ~$1B/mile in LA, so he needs a 10x cost cut via smaller diameter, continuous tunneling/reinforcing, and more machine power 3:05.
• Why not flying cars, saving all the tunneling cost? — Noise, high wind force, and constant anxiety about things overhead 6:10.
• Is there a tie-in with Hyperloop? — Yes; tunnels built to resist the water table (5–6 atm) automatically hold vacuum (1 atm) 7:14.
• How much of your time is the Boring Company? — "Maybe two or three percent"; interns and part-timers with second-hand machinery 10:43.
• Is one motivation that autonomous EVs mean more cars on the road? — Yes; shared autonomy is cheaper than a bus ticket, so driving increases and traffic worsens 11:33.
• Could you declare victory on electrification and move on? — No; he intends to stay with Tesla "as far into the future as I can imagine" 13:13.
• When will the eight-camera cars do what the video showed? — Cross-country LA to New York fully autonomous by November/December 2017, no controls touched 15:08.
• When can you take your hands off, sleep, and arrive safely? — About two years; the bar is "unlikely to crash in 100 or 1,000 lifetimes," not 99.9% 17:07.
• Will buyers really be able to earn money renting the car out Uber-style? — "100 percent what will occur. It's just a question of when," with owner-set sharing restrictions 19:01.
• Anything to show on the Semi? — Only a teaser shot; heavy-duty long-range, out-torques any diesel, would win an uphill tug-of-war, drives like a sports car, single speed 19:34.
• Are solar roofs affordable? — Solar glass roof + electricity should cost less than a normal roof + electricity; "economically a no-brainer" 23:43.
• How many houses could end up with this? — Eventually almost all, over 40–50 years; in 15 years a non-solar roof will be unusual 24:59.
• How many Gigafactories to get off fossil fuels? — About 100; 2–4 new locations to be announced later that year, probably four; location details withheld beyond "global market" 27:39.
• Should you be advising a president who doesn't believe in climate change? — He's on two advisory councils that meet every month or two and go around the room; he has used them to argue for immigration and climate change 28:55.
• Why build a city of a million people on Mars? — Because a future among the stars is inspiring, and unlike sustainable energy it is not inevitable 34:43.
• Why not just fix problems here and now? — Sustainable energy will happen regardless; Tesla's value is accelerating it by ~a decade. Being space-faring is not inevitable, and space capability has been trending down since 1969 36:00.

Things to look up

• The Boring Company — Musk's tunneling venture; ~2–3% of his time 10:29
• Gary the snail — pace benchmark for the tunnel boring machine; from SpongeBob SquarePants (Musk first said South Park), currently 14x faster than a TBM 5:34
• LA subway extension — ~2.5 miles for ~$2 billion, his cost baseline 3:12
• SpaceX Hyperloop test track — built for a student pod competition; largest vacuum chamber in the world by volume after the Large Hadron Collider 7:14
• Tesla Model 3 — production "looking quite good" for starting in July 13:34
• Tesla Semi — heavy-duty long-range electric truck, announcement planned for September 19:34
• Tesla Autopilot / Model S — cameras + GPS, passive optical, eight cameras on new cars 14:03
• Powerwall — three shown stacked against the side of the demo house 22:18
• Solar Roof (solar glass tiles) — microlouver glass tiles, two types at launch, two more early next year 22:39
• Gigafactory — diamond-shaped, aligned on True North, 100+ GWh/year 26:44
• Interplanetary Transport System — ~40-story vehicle, 4x Saturn V thrust, 8–10 year aspirational timeline 32:09
• Saturn V — thrust comparison point 32:42
• Space Shuttle — cited as evidence capability regressed after Apollo 37:04

Cautions

• Don't assume 99.9% is good enough for autonomy. Musk: if a car crashes 1 in 1,000 times, "you're probably still not going to be comfortable falling asleep... certainly you shouldn't be" 17:07.
• Risk of premature trust: the interviewer raised the concern that people get "seduced too early" into thinking it's safe and a horrible incident sets things back; Musk answered that vehicle safety is probabilistic and the system is likely to at least mitigate crashes "except in rare circumstances" 17:55.
• Autonomy on the Semi: "That's definitely a case where we want to be cautious about the autonomy features" 19:52.
• Adding one layer of tunnels won't fix congestion on its own — it gets used up; the argument only works if you can keep adding levels 2:26.
• Autonomous cars will not by themselves reduce traffic — shared autonomy makes traffic "far worse" 11:33.
• Technology does not automatically improve and "by itself [will] degrade" without a lot of people working hard — pyramids and Roman aqueducts as examples 38:00.
• Musk declined to say where new Gigafactories would be located 28:37.