The airline CEO putting Starlink on every plane

1. Key Themes

Betting Against the Experts When the Tails Hit

Kirby's COVID fleet order is a template for sizing a bet when consensus is badly wrong. He argues experts "force-fit everything into things that they've already seen," so the tail events, where the value is created, are exactly where they fail. The asymmetry mattered: because rivals were shrinking, he only needed to be "not totally wrong."

"Every single expert in the world said travel demand is going to be down by 50% forever... you should ignore the experts in the outside of two sigma. In the tails, experts are terrible." — Scott Kirby 00:00:32

"This is one of those cases, because everyone else is shrinking, we don't have to be right. We have to be not totally wrong." — Scott Kirby [01:36]

The result was about 200 wide-bodies plus several hundred narrow-bodies, "the biggest aircraft order in history." He got better terms too: "we're the only one in the world. And we sort of launched the market" 00:03:01.

Starlink as a 2-3 Year Window of Competitive Advantage

Kirby calls Starlink "the biggest competitive advantage," though a temporary one. 48% of flights are on Starlink today, rising to 50% within a week. The plan is to use the window to convert other airlines' loyalists. He says customers will "try us for Starlink, and they're going to stay because all the other technology and everything else is better" 00:35:51. He also names the horizon: "we are going to have a few years of being the only airline that has it at scale" 00:35:51.

"I think it is. I think it's the biggest competitive advantage, and it's temporary... Everybody, eventually, is going to have high-speed, low Earth orbit internet connectivity." — Scott Kirby 00:35:38

Los Angeles will be the first 100% Starlink hub, sometime in the first half of next year. The 787 certification, which depends on Boeing, is the bottleneck.

"Fast, Not Free" and the Physics of Legacy Wi-Fi

Kirby rejects "free Wi-Fi" as a value proposition. Legacy systems can't deliver it at usable speeds, and making them free made them worse.

"Free was never the value proposition. Fast is the value proposition. And there's only one option for fast, it's Starlink... It's slower than dial-up internet." — Scott Kirby 00:37:02

"It's physics. There's not enough bandwidth to make it free." — Scott Kirby 00:36:51

The Unglamorous Tech Stack as Compounding Moat

In the depths of COVID, United started a "several hundred million dollar project" to get "everything off of mainframes" with "no ROI. There was no payback, which is why companies don't do it" 00:07:16. Kirby says it now lets United do "stuff that other airlines, frankly, can't do." United is the only airline that owns its reservation system, and it is now taking over software like seatback entertainment from Panasonic. The compounding claim is that "our gap is getting wider and wider with our technology every day" 00:36:12.

AI as Programming Leverage, Not Summarization

Kirby dismisses summarization use cases ("summarize my day. I can read my calendar") but had an "AI epiphany" after three one-hour tutoring sessions. He then built a sports-bet aggregator in 2.5 hours during a gym workout. His conclusion was that United was "all in."

"The power of programming... I could do it with three hours of tutoring, on my own, could write a program to do that was an epiphany for me." — Scott Kirby 00:09:23

He also says AI allows rapid trial: "It takes more time to do a survey or a focus group than just do it, see if they like it, and if they don't, stop" 00:06:10.

Communication Over Expectation Management

United A/B tested weather-delay messages. The plain text version scored about -60 NPS, while the message with a radar map scored +60.

"The people with 1-hour delay, I think the numbers were -60 NPS on the normal way of sending it, +60 when we sent them a picture of the map. 120-point difference!" — Scott Kirby 00:18:37

"It was better than an on-time flight!" — Scott Kirby 00:19:07

Kirby rejects "expectation management" as an excuse framework: "we try to set high expectations and try to meet them" 00:21:55. At a previous employer he banned the word "weather" from the operations team.

The "Wow" Debt: Delight Becomes Baseline

Kirby told his officers "their job is 'Wow'" and acknowledges the treadmill. Customers bank each delight and then expect another.

"We get credit for that one time, and then we owe them another wow. They should put that in the bank and we should owe them, and it's a cultural that means like you can't ever rest." — Scott Kirby 00:16:25

He borrowed the memory-creation idea from the book Unreasonable Hospitality, such as a live countdown to a passenger's 2 million-mile mark.

Air Traffic Control Is the Binding Constraint

Kirby's answer to what limits the business is unambiguous: "Air traffic control is 90% of the issue" 00:22:05. He explains the structural reasons: annual appropriations that can't fund 5-10 year projects, congressional district politics ("how many jobs are you going to put in my district?"), 450 facilities where one would do, and a private-jet lobby that pays almost nothing. Capacity growth is therefore capped: "we hit the limit of what the FAA can manage, and then that becomes the limit."

Airline Economics: A Two-Winner Industry

Kirby argues the structure has changed. "Basically two winning airlines in the US, United and Delta" built brand loyalty that goes beyond the "commodity" schedule. Six of ten US airlines will lose money this year, and even with oil up $8 billion for United, it is "solidly profitable."

Discipline: Know What You Are

On Ryanair, Allegiant and the failed low-cost carriers, Kirby's thesis is that companies fail when they abandon their model. Low-cost carriers moved to expensive primary airports, and legacy carriers failed at out-Southwesting Southwest.

"Companies are not good at trying to be something they're not." — Scott Kirby 00:59:20

"In the airline business, discipline is really about where you fly." — Scott Kirby 01:01:35

2. Contrarian Perspectives

Value-Weighted Risk Models and Black-Scholes Are "Dumb" Because the World Has Fat Tails

Kirby calls VaR "the dumbest model in history," and Black-Scholes "such a dumb model for valuing options," because both assume normal distributions. He says things that "shouldn't happen in the history of the universe happen once every five years." The point isn't academic. It underpins his entire investing posture of going big in the tails.

"The world is not normally distributed. The tails are much fatter in everything that matters." — Scott Kirby 00:04:43

Surveys Are the Worst Tool for Predicting the Future

Customers surveyed during COVID said they'd never travel internationally again, and they were sincere and wrong. Kirby uses surveys only retrospectively.

"I like surveys for, how did people feel about stuff we've already done? I ignore surveys for what are we going to do in the future." — Scott Kirby 00:05:37

Make Things That Hurt Wall Street, and Lose Your Job if You're Wrong

Rather than complain about short-termism, Kirby says Wall Street "almost always hates the big things we do, like, and I don't care." His prescription for fellow CEOs:

"If it doesn't work, you ought to lose your job and you ought to be comfortable with that. You get paid a lot of money, like make the big bet." — Scott Kirby 00:40:10

Ben Thompson agreed: "We need more gamblers as CEOs" 00:40:29.

Single-Airline Dominance of a Hub Is Better for the City

Kirby argues that duopolistic competition between two hubs at Chicago wastes slots on duplicative "catchment area" flights (e.g., 16 daily Milwaukee flights vs. 8 on bigger planes). Atlanta and Dallas, smaller cities, generate more long-haul service and economic activity. He claims "Chicago should be the biggest airport in the world. It's not. And it's because of this" 00:29:00, and predicts American will "lose probably a billion and a half in Chicago this year."

Slot Control Is Good for Customers

A pro-market audience would expect an airline CEO to oppose slot constraints. Kirby begged the FAA to impose them at Newark: "What's bad for customers is scheduling more flights than an airport can handle" 00:53:59. Newark is now "the most reliable."

Boeing and Airbus Have a Quiet Duopoly That Stalls Aircraft Innovation

Kirby says the two manufacturers have "quietly agreed, neither one of us will invest in new airframes," and that "we need a third aircraft manufacturer, to be blunt" 00:48:43. He has been "begging them for 20-something years to build a 757 replacement."

3. Companies Identified

Starlink / SpaceX

Satellite internet provider. The central theme of the episode. United is deploying it across its fleet at about three aircraft per day, and Kirby says "SpaceX supported us" in accelerating the rollout.

"Fast is the value proposition. And there's only one option for fast, it's Starlink. That is the only option." — Scott Kirby 00:00:11

United Airlines

Kirby's company, described as "the best airline in the world" on the strength of technology. It is the only airline that owns its reservation system, with 120,000 employees. It will have 53 transatlantic destinations next summer versus 23 ten years ago, with "only two that didn't work."

"That's why I now believe that United, in totality, is the best airline in the world, and it's because of technology." — Scott Kirby 00:17:52

Delta Air Lines

Kirby's main competitor and the other "winning" airline. He credits Delta as "the first airline to start down this building a brand loyalty" path. He believes Delta's gradual Starlink rollout (they picked Viasat early) was a mistake: "It was a mistake."

"I took inspiration for what they... They were the first airline to start down this building a brand loyalty." — Scott Kirby 00:30:37

Ryanair

The only successful low-cost carrier left in the world, according to Kirby. It flies to secondary airports and stays disciplined.

"There's really one still low-cost carrier successful airline in the world, it's Ryanair." — Scott Kirby 00:57:40

Allegiant Airlines

Profitable ultra-low-cost carrier flying small cities to Vegas, one of the four airlines profitable this year. Held up as a model of knowing what you are.

"They've done a great... they're one of the four airlines that's going to be profitable this year." — Scott Kirby 01:00:31

JetZero

Blended wing body aircraft developer in which United is an investor. Kirby thinks blended wing bodies are the next big airframe advance.

"We're an investor in a company called JetZero, who's trying to..." — Scott Kirby 00:49:38

Boom Supersonic

Supersonic aircraft company working with United, mentioned by John Collison ("obviously you guys are working with Boom") as the speed play.

Copa Airlines

Panama-based partner of United and the only other airline using United's reservation system.

"Copa Airlines in Panama uses it because they're a partner with us. We're the only two airlines in the world that use it." — Scott Kirby 00:11:01

ANA

Japanese joint-venture partner. Thompson notes passengers who previously chose ANA over the Pacific will now choose United because of Starlink.

Samsung

Mentioned by Thompson as a model of counter-cyclical investment: "Samsung came to dominate memory by systematically investing into every downturn" 00:02:08.

Berkshire Hathaway

Mentioned by John Collison as having owned airline stocks recently, in the context of airlines' improving economics.

CVS Health

Kirby serves on its board.

Apple

Cited by John Collison as an example of building without focus-group testing: "Apple famously does not do a lot of focus group testing" 00:05:46.

In-N-Out

Collison's example of disciplined growth within a distribution constraint.

"They famously do everything within a one day's drive of their refrigeration centers." — John Collison 00:58:52

Tesla

Collison's example of de-vendorizing and owning the tech stack.

Panasonic

Current provider of United's seatback entertainment software, which United is taking over.

Sabre / American Airlines Decision Technologies

Dominant reservation-system provider. Kirby worked at its predecessor, and every other airline outsources to Sabre or Apollo.

Spirit Airlines

Liquidated low-cost carrier used as the cautionary tale. Average fare was $58 while New York airport fees were $52 per passenger.

American Airlines

Rival losing "a billion and a half" in Chicago. Kirby says it lost gates through inattention and a Sun Belt focus.

Southwest and Alaska

Competitors against which United gained share in Denver and the Bay Area. Kirby says Southwest is "now trying to be like us, and they're not going to be very good at that either" 00:59:40.

Kalshi

Prediction market that Kirby included in his sports-betting aggregator.

Viasat

Legacy in-flight Wi-Fi provider Delta chose early, mentioned by Thompson.

Stripe

Collison's company, which published Stripe Economics research on AI and productivity.

4. People Identified

Patrick Quayle

United executive who runs international route planning. He found Nuuk and the Spain/Portugal pilgrimage routes by seeking out data no one else uses.

"We got a guy named Patrick Quayle and a great team that are really creative, and they go look for things that are different and outside the norm, not in the data." — Scott Kirby 00:44:32

Maggie Schmerin

United VP who runs marketing. She launched the Starlink ad campaign without asking Kirby for approval.

"She didn't ask me, she knew she didn't ask me, like no approvals. She's awesome, I trust her, and that creates speed." — Scott Kirby 00:52:28

Michael O'Leary

CEO of Ryanair. Described by Kirby as "entertaining" and "disciplined," and by Collison as "a very impressive CEO."

"He knows his business exceptionally well." — John Collison 01:00:59

Secretary Sean Duffy

Transportation Secretary. Kirby says he can text him and "get a response within 30 seconds."

"Secretary Duffy is great." — Scott Kirby 00:22:22

Bryan Bedford

FAA Administrator, described as "the best FAA administrator we've ever had" 00:22:40.

Clarence "Kelly" Johnson

Lockheed Skunk Works founder, cited via the book Impossible Factory as an example of how fast aircraft development used to move (SR-71, U-2 "in a couple of years, from start to finish").

Elon Musk

Founder of SpaceX/Starlink. His push to require airlines to make Wi-Fi free was met with Kirby's "We were going to make it free anyway." Kirby also noted the public "fight" between Musk and Delta CEO Ed Bastian "I love it for us."

Ed Bastian

Delta CEO. Kirby says "I kind of feel bad for Ed," and that he won't lose his job over Viasat.

Juan Trippe

Pan Am founder, referenced by Collison as the ego-driven era of aviation.

Warren Buffett

Quoted by Collison for the line about shooting down Orville Wright at Kitty Hawk.

Orville Wright

Referenced in the Buffett line.

5. Operating Insights

Pre-Authorize Your Best People to Move Without Approval

Kirby's marketing VP launched the Starlink ads without asking, and he praises that as the source of speed.

"Most companies, especially big, traditional companies, have 16 layers of approvals. You can't do stuff fast, or creative, or fun, or interesting because it just gets watered down." — Scott Kirby 00:52:28

The tactic is to hire people you trust and make clear they have standing support, so approval isn't the default step.

Test Customer Communications With Real A/B, Not Surveys

The radar-map experiment (-60 vs +60 NPS) is a template: swap the content of a mandatory message, not the feature. A 120-point swing came from showing evidence instead of asserting a cause.

Stage Risky Cutovers by Deliberately Cancelling Exposure

For a critical mainframe migration, United canceled "several hundred flights" so a failure wouldn't strand customers. He says: "We just don't schedule flights during that window, so that in case something goes wrong, we're not putting customers in harm's way" 00:10:18. Pre-shrinking the blast radius lets you do the cutover at all.

Use Incremental Windows to Retrofit Big Programs

Kirby explains the easiest way to install Starlink is during a heavy check: "You've got an airplane that's going to be down... put Starlink on." Piggyback on scheduled downtime instead of creating new downtime.

Make Data-Light Bets Through Creative Hires, Then Kill Fast

The Nuuk and pilgrimage-route playbook is: hire people who read outside the data, validate with "broken itineraries," try the route, and stop quickly when it fails. This is possible because "our assets are fungible... we can try stuff and stop" 00:45:09 (Bergen, Norway was dropped after a small loss).

Start With Anecdotes Before Committing to Scope

Kirby planned Starlink on 65% of the fleet but changed course after conversations with passengers: "I'm switching to flying a regional flight because it's got Starlink." The operational lesson is to keep the plan open until real customer behavior shows up after the first unit ships.

6. Overlooked Insights

Owning the Legacy Reservation System Was an Accident That Became a Moat

Kirby mentions almost offhand that "we lucked into it" owning the mainframe reservation system. When EDS and then "someone else" didn't want to keep supporting it, "we had to support it on our own, but it's been huge" 00:11:24. This is the enabling condition for the whole stack-ownership strategy (seatback software, personalization, ConnectionSaver-style tools). A thing everyone else outsourced as a cost center, once forced in-house, became an asymmetric capability that can't be bought. The Copa partnership means only two airlines on Earth run it, which also means no vendor ecosystem can replicate it for competitors.

The Gate Reallocation Rule Is How United Took Chicago Share

Ben Thompson mentions in passing that Chicago gates are reallocated yearly based on prior flying. Kirby's offhand answer reveals the mechanism: American focused on the Sun Belt after COVID and "just didn't add back their schedules," so it "lost six gates to us." United "didn't do anything special" except return to its pre-COVID schedule. A bureaucratic use-it-or-lose-it rule converted a strategic inattention into a structural, hard-to-reverse share transfer. The same pattern may apply wherever allocation is based on historical usage.