A Ferrari sold at Pebble Beach for around $17.8 million weeks after the same model failed to clear €8 million at a European auction, and Mitchell Green says the gap is the American wealth effect showing up in an asset you can photograph.
Most investors read that market through art. Green says the art market has not moved, and that the money is going into collector cars, trading cards, private jets and houses in a six-block radius of San Francisco.
"Like, the amount of wealth creation that's happening is just astonishing."
Green founded Lead Edge Capital and invests in growth-stage software, backed Duo Security and has been buying into public software companies for long enough to have a rule for reading the sector: watch what the listed ones say on their earnings calls.
The full interview is covered here so you can skip it. 28 minutes of audio, 13 minutes of reading.
Here are the 9 arguments that matter.
👤 Guest: Mitchell Green, founder and Managing Partner of Lead Edge Capital
🎙️ Hosts: John Coogan and Jordi Hays, who run TBPN's daily live tech show
👥 Also on: Nico Wittenborn of Adjacent, Scott Keogh of Scout Motors, Ben Gilbert and David Rosenthal of Acquired, and Faraj Aalaei of Cognichip, in separate segments of the same episode
📰 Published: 14 September 2026 on YouTube (TBPN)
🔴 YouTube | 🟣 Apple Podcasts | 🔗 Episode page | ⏱️ 28 min | ✅ Time saved: 15 min
Key Takeaways
The same car costs about twice as much in America as in Europe, and the buyer cannot arbitrage it for 25 years
A Ferrari at Pebble Beach near $17.8 million against under €8 million in Europe a month earlier
A heavy debt load, not a weak product, is what stops a company answering AI
His example: Ford can invest through a robotics shift, Stellantis is paying down debt
The best read on software demand is what listed software companies say on their earnings calls
Formula One's halo is the argument for pacing AI: it adds weight, slows the car and saves lives
Buying a no-growth software company at three times revenue can pay for itself out of cash flow
A $100 million revenue business bought for $300 million and run at $60–70 million of earnings
Voice fraud is the next security problem and nobody has built the company for it
1. The AI-Built Racetrack
Green came on tired, and the reason turned into a test of what the current models can do.
He has a new racing simulator and cannot get off it. Friday and Saturday nights after the children are in bed, he tells himself he will stop at ten and it is past midnight before his wife texts to ask whether he is seriously still in there
"But it's the most it's the most addictive." He comes out of it sweating, with his brain fried by light, and then tries to sleep
One host pointed out the scale of the screen problem. People are told not to look at a phone before bed; Green is looking at a wraparound triple-monitor setup
The evaluation they built is a racetrack. A host used GPT six Astra to find satellite footage of Thermal and build the circuit inside Assetto Corsa, which gives them something they can actually drive and judge
Green has already tried it, badly. He drove the generated track on a keyboard rather than a full rig, and passed it on for a proper demo
His prediction for real cars is that an AI would beat the professionals. He has argued it with GT3 drivers who say it is impossible, and answers them with SpaceX landing rockets. Given enough money, he thinks a machine-driven car would lap faster than any professional
The commercial idea attached to it is a racing league in which each model company sponsors a team, which the hosts noted is already half-happening through Formula One sponsorship
The counter-argument was chess. Once the machine is superhuman you still watch the humans
2. Watch Software Earnings
Asked what has changed in the real economy since he was last on, Green started with the listed software companies.
His method is to read the public companies' results as the best available picture of what is happening in software globally, and he said those results have been pretty strong
He pointed at Workday's disclosure of AI-attributed revenue as an example of the numbers now being put on the table, and hedged the figure himself as he said it
Large enterprises are not ripping vendors out. His read is that big companies want their existing vendors to build solutions with them rather than replace them with new suppliers
That does not make anyone safe. The pace of innovation is only going to increase because of AI and agents, and every company carries a risk of being disrupted
What he keys on in a listed software company is the person running it. A founder or chief executive who is locked in, aware, early but not repeating talking points, and thinking about the capabilities rationally, makes him more optimistic
3. Debt Decides Who Innovates
The host asked whether capital structure is under-discussed as a variable, and Green said the leverage debate is being had in the wrong place.
He rejected the blanket version of the argument. People bucket all private equity assets together as over-levered and in trouble, and he does not think that is right
The distinction he draws is about what debt prevents. A very highly levered company finds it harder to disrupt itself and keep innovating, because more and more money is going out as interest
His example was two carmakers. If robotics, humanoids and AI dramatically change manufacturing, Ford can invest in it and Stellantis probably cannot, because it is paying its debt load down
He said the same test applies to every sector. "the people that are not innovating are gonna get left behind"
The cost cuts both ways. It is easier to start a company now, and it is also easier for incumbents to develop new products
4. Ferrari Costs Double Here
The conversation turned to where the money from this cycle is actually landing, and Green used the collector market as the evidence.
The price gap is his headline observation. A famous Ferrari sold at Pebble Beach for around $17.8 million; the same car in Europe had never sold for more than about €8 million, or €7.5 million, at an auction a month earlier
The gap cannot be closed by shipping the car. He pointed out for the audience that a European car cannot be brought into the United States for 25 years
He reads it as a statement about the economy, not about cars. He said it shows the strength of at least some segment of the American economy against anywhere else in the world. "Like, the amount of wealth creation that's happening is just astonishing."
Trading cards have gone the same way and art has not. He named Michael Jordan cards as an example of the collector-card market going crazy, and said the art market has not
His explanation is equities plus mortality. People who were wealthy in America three years ago owned stocks, are now far wealthier, and have realized they cannot take it with them — so they are buying for children and grandchildren who are into cars or watches
He listed the sources together: equity markets, AI, secondaries and venture, conflated into one wave of spending that is more concentrated in America than anywhere else
Family offices are now treating cars as an allocation. He has heard of second or third generations putting 3% of a large fortune into cars and going shopping with a $50 million list
The same pressure is visible in aircraft. He was told a new customer cannot currently get a NetJets or Flexjet plane at all
The stat he was given, and flagged as someone else's: if Anthropic is worth a trillion or a trillion and a half, OpenAI is worth something similar, and SpaceX trades where it does, those three companies have created more gains than the entire internet bubble
Housing is the same auction with fewer lots. A buyer who has just made a great deal of money quickly is close to price-agnostic about a particular neighborhood, the seller knows there is no inventory, and homeowners locked into 2020 and 2021 mortgage rates have no incentive to sell
The poster effect explains which cars. The car someone had on their bedroom wall as a teenager is the car they buy when they make money in their thirties
5. Track Clubs And Cold Starts
Southern California now has several private track-and-housing developments, and the hosts asked how many the market can carry.
Green is a member of one of them. He named Thermal, and said Tim Rogers has been running it for fifteen years
His view on the newcomers was polite and unconvinced. He said the incumbent would probably wish them luck, and that it is a lot harder than people think
The problem is the cold start. You have to build a track, then get enough members to hold member race weekends, and nobody wants to be among the first five houses
One host had checked the marketing against the map. A development borrowing the Nürburgring idea for people who will never get to Germany turned out to be about a fifth of the length
6. Street Cars Kill On Track
The safety argument was the one Green pushed hardest, and it is also a cost argument.
His position is unambiguous. "I truly believe that it is extremely dangerous to drive streetcars on a racetrack."
The regulatory difference is the reason. European rules let a manufacturer fit four-point harnesses to a fast road car; in America, he said, you cannot
So the belt does not match the speed. "Driving a car down a racetrack at a 170 miles an hour with a lap belt, I think it's totally insane."
The comparison he offered is the crash footage. People walk away from purpose-built race cars; hit a wall in a street car at 100 miles an hour and you are dead
The economics point the same way. "If you're looking for like a specific style of car, the track only version of the car will be half the price."
His conclusion for the clubs is a membership rule. He thinks these facilities should not really allow street cars on the track at all — if you want to be a member, get a race car
7. The Halo Slows The Car
The hosts brought the weekend's argument about pacing AI development to him, and he answered it in racing terms.
What struck him was the agreement, not the fight. "the three guys that all run the companies actually all agreed" — which is exactly what makes people call it collusion
He was careful about his standing to judge it. He said he does not know Sam Altman, does not know Dario Amodei and has never met Elon Musk, and would still bet they are genuinely concerned about what happens if the technology advances too fast
He wants open-source models to exist so that the power does not sit with three companies
The analogy he chose is a safety device that costs lap time. "The halo in f one does slow the car down. It is extra weight, and yet it saves so many lives." He cited the crash where a wheel hit Lewis Hamilton's halo
The argument from authority is the point. If the fastest drivers in the sport all said a rule was needed even though it would slow the racing, you would listen — and he thinks the same applies to the three people running these companies
He thinks regulation is coming regardless. His stated fear is that governments overregulate rather than that they do nothing
He credited them for moving first because China is already at it, thinking about regulatory frameworks for AI
On the day's selloff, he thought the market misread the message. Nobody is saying stop innovating; the message is that the industry does not want to hurtle off a cliff
The incident that made it concrete was the OpenAI model's attack on Hugging Face. "Like, can you imagine if you read, like, oh, sorry. Like, JPMorgan accidentally hacked Goldman Sachs."
The version of the risk now being argued is narrower than extinction. The scenario put on the table in the conversation was a small chance of a botnet causing a massive internet outage within a year, which is a far more tractable thing to argue about than a doomsday scenario
8. Voice Fraud Is Next
Green has made money in security before and thinks the obvious next company has not been built.
The gap he keeps pointing at is synthetic voice. "I think voice fraud is going to explode." If a voice can be recreated easily, it can be used to commit every kind of fraud
The live example is call spoofing. He said Apple still has not fixed caller-ID spoofing, so a scammer can appear on the screen as a company you trust and ask you to reset a password — and he credited Ryan Peterson with raising it that morning or the day before
His track record here is the reason he is watching. Lead Edge backed Doug Song at Duo Security, which he said became a huge business, and also backed Andrew Peterson
He now rates a cyberattack as the largest global risk, on the grounds that people lose patience when they cannot buy an airline ticket for three hours — and an outage measured in days would be a different order of problem
The precedent both sides reached for was the CrowdStrike update that took Windows machines down, as the nearest thing to a preview
9. An American Bending Spoons
The last stretch was about who buys the software companies venture capital cannot exit, and Green thinks the answer is coming.
The host framed it as a grievance. Venture funds put hundreds of millions into these companies and then watch an Italian buyer take them for pennies on the dollar
The precedent Green named was Trilogy Software and its founder, who built a large business buying software in the 1990s and 2000s
He expects an American version. "Somebody's gonna do it in The States."
The reason the current buyer prints money is the absence of competition. "I think that Bending Spoons is gonna print on a lot of these deals is that there aren't a lot of buyers." There are not many people willing to spend a billion dollars on a company that is not growing anymore
The arithmetic he laid out is the whole case. Buy a $100 million revenue business for $300 million, run it at $60 million or $70 million of earnings before interest, taxes, depreciation and amortization, and you have paid yourself back
"I mean, that's an amazing free cash flow yield"
What stops other investors doing it is temperament. Venture funds, buyout funds and private equity firms cannot let go of an asset they have held for fifteen years
Why this buyer gets noticed is the shelf it is buying from. Trilogy bought smaller businesses nobody had heard of; Bending Spoons is buying names that are household names in technology and were backed by top-tier venture funds
He expects more of it, and more failures alongside it. There are companies funded in 2012, 2015 and 2018 that have been sitting around for years with nobody knowing what to do with them, and founders who want to move on to a fresh team
One deal already shows the mechanism. A company the buyer acquired spun out its AI and experimentation division, taking the founders and some key employees with it while others stayed with the business
Green's bottom line is that the returns in this cycle are being made on the balance sheet — who can spend through an AI transition, who can buy a finished company cheaply — rather than on picking the winning model.
Bonus Insights
His verdict on the greatest companies of the next twenty years is that we do not know their names. Sitting in 1999, he said, nobody would have named ByteDance or Facebook, and social media is a $5 trillion market today
He expects most of the new companies to fail, and said the prize is gigantic for the ones that win
On whether AI racing would be watched, the hosts' test was whether a humanoid robot could put in a lap in a simulator first and a real car second
The chat produced a name. A viewer suggested Octave Capital as a candidate American acquirer, and Green said there are others potentially working on it
Products, Companies & Tools Mentioned
Lead Edge Capital (Green's firm; he invests in growth-stage software and reads the sector off public earnings)
Bending Spoons (The acquirer he says will print on these deals because there are almost no competing buyers)
Workday (Cited as an example of a listed software company now disclosing AI-attributed revenue)
Ford and Stellantis (His illustration of debt deciding who can invest through a technology shift)
Duo Security (Lead Edge backed its founder; the security investment he points to as evidence)
CrowdStrike (The Windows outage both sides used as the nearest preview of a real internet failure)
NetJets and Flexjet (He was told a new customer cannot currently buy into either)
Assetto Corsa and iRacing (The simulators; one now running an AI-generated copy of a real circuit, the other with AI cars he rates for practice)
Ferrari (The price gap between the American and European auction markets)
Anthropic, OpenAI and SpaceX (The three private companies in the stat he was given about gains exceeding the internet bubble)
Trilogy Software (The 1990s precedent for buying up software companies, and the model he expects America to repeat)
If this was worth your time, send it to someone who has to have a view on this.
Get the latest market chatter as it happens: