All podcasts / My First Million / Summary

The oldest trick in the book to make your first million

2026-07-23 - 51 min - source - Read full transcript

Sam Parr (host)Shaan Puri (host)

Key insights

A middleman business can out-earn everyone else in a supply chain when it occupies a physical position nobody else can replicate.

Cargill's origin story is building grain elevators right next to new railroad lines so farmers had nowhere else to go to store and ship their crops. That unreplicable physical position, not superior farming or trading, is what let a pure logistics middleman grow into the largest private company in America.

middleman-value-creation

Cargill has stayed dominant for 40+ years by pairing a strict reinvestment discipline with professional management.

The family enforces an 80/20 rule (80% of profits reinvested, 20% paid out as dividends) and has handed day-to-day control to professional CEOs for roughly the last 20 years rather than insisting on family operators, which the hosts frame as the reason the compounding never stopped.

family-business-dynasties

Multi-generational family wealth tends to survive on formal governance, not goodwill.

William Randolph Hearst set up an ironclad legal trust for the Hearst family with fixed board seats and a rule that profit distributions can't be argued over - dispute it and you're cut out of the will. The hosts contrast this with most American families, where money is simply never discussed, and argue the explicit structure is what let Hearst's empire outlast him.

family-business-dynasties

Deliberately manufactured family rituals substitute for conversations most households never have.

Sam describes running recurring family meetings modeled on board meetings, plus inventing personal 'gift' narratives for each of his kids and toying with designing a family crest, as ways to make values and expectations explicit instead of implicit - which he ties directly to why some family dynasties (Rockefeller, Hearst) held together across generations while others didn't.

family-business-dynasties

The Jevons paradox: making a resource more efficient to use increases total consumption of it, it doesn't reduce it.

Named for the 1865 book The Coal Question, which predicted Britain's coal use would rise, not fall, as the steam engine made coal use more efficient - because efficiency unlocks demand nobody had accounted for. The same pattern shows up with the cotton gin: it made cotton processing 50x faster, and instead of shrinking the demand for enslaved labor, cotton got so cheap and so dominant ('King Cotton') that the US imported far more slaves to keep up.

jevons-paradox

The hosts argue AI-driven code efficiency will explode demand for code and coders rather than shrink it, mirroring Nvidia's real-time Jevons paradox.

As AI makes writing code dramatically cheaper, they argue total demand for code will grow far faster than jobs disappear, the way electricity became a base unit you can apply almost anywhere rather than a fixed need (unlike hot water, which plateaued once households had 'enough'). Jensen Huang's claim that AI inference demand will vastly outstrip any efficiency gains from more efficient model training is cited as the same dynamic playing out live in GPU markets.

jevons-paradox

Every technology that boosts efficiency creates a losing subgroup that resists it, and history treats that resistance as understandable but ultimately futile.

The Luddites smashed textile machines under penalty of death; automatic switchboards eliminated ~800,000 telephone operator jobs in the 1970s; ATMs were feared as bank-teller killers but coincided with more bank branches opening. The hosts extend the parallel to today's backlash against AI and data centers (college graduation crowds booing a Goldman Sachs exec, a New York official banning data centers), arguing the pattern - and the eventual outcome - repeats.

ai-labor-disruption

How fast a new technology's economic boom arrives depends on breadth of impact, intensity of impact, and how much co-invention it requires.

Railroads took roughly 50 years to fully pay off because they needed newly invented steel and labor-intensive track-laying alongside them. The hosts speculate AI's boom could compress into a much shorter window because it needs comparatively little physical co-invention to diffuse.

ai-labor-disruption

New technology repeatedly produces industries too large to have been predicted by anyone sizing the market beforehand.

Nobody estimating demand after Gutenberg's printing press could have conceived of something like Twitter; the hosts use this to argue that trying to bound AI's job-creation potential using today's categories will systematically undercount what actually gets built.

ai-labor-disruption

'Have you tried solving the problem?' is framed as the single most common piece of advice a YC partner gives founders who over-engineer.

Shaan recounts his former Twitch boss Emmett Shear (later a YC partner) repeatedly cutting off elaborate 15-minute plans with that one line, and says Shear told him it's the most common advice he gives YC startups - a reminder that the 'only way out is through,' not around.

productive-avoidance-patterns

Small shifts in self-talk are framed as levers for reclaiming agency, both in business and daily habits.

The hosts single out two phrases as traps: 'it is what it is' (framed as a total surrender of agency, to be replaced with 'it is what I make of this') and 'I might as well' (used to rationalize a bad decision, e.g. eating poorly, as if it were already decided).

productive-avoidance-patterns

Books referenced

The Coal Question - William Stanley Jevons - 1865 book that originated the Jevons paradox - the argument that Britain's coal use would rise, not fall, as steam engines got more efficient.
The Medici Effect - Frans Johansson - Book Shaan's producer Diego was reading; source of the Frederick Tudor 'ice king' story used to close the episode.
Americana - Bhu Srinivasan - Older book on breakthroughs in American capitalism that Shaan says is popular again; used to argue that fears of new technology repeat throughout history.

Companies

Cargill - Episode's 'billy of the week' - largest private company in America for 40 years, 88% family-owned, ~$150B in annual revenue, built from grain-elevator middleman logistics into food, shipping, and meatpacking.
Garda Capital Partners - $10B+ hedge fund owned by the Cargill family, used to hedge commodity exposure for the family and for farmers.
Hearst Corporation - Cited as another century-plus family empire; William Randolph Hearst set up an ironclad legal trust with fixed board seats and equal profit distribution to keep heirs from fighting.
Rapha - D2C cycling brand cited as an example of a low-profile family (the Waltons) quietly buying into a niche business most people wouldn't associate with them.
HubSpot - Episode sponsor; built and gave away a free database of every business idea MFM has ever discussed.
Nvidia - Referenced via Jensen Huang's comments that AI inference demand, not training efficiency, will define GPU demand - a real-time Jevons paradox example.

Techniques and frameworks

80/20 profit rule - Cargill reinvests 80% of profits back into the business and pays out only 20% as family dividends, funding a century-plus compounding streak.
Family meetings - Recurring (monthly/quarterly/annual) explicit family sessions modeled on board meetings - values, results versus goals, and money openly discussed - proposed as the mechanism that lets wealth and culture survive across generations.
Jevons paradox - Economic principle: making a resource more efficient to use increases total consumption of it rather than decreasing it, because falling costs unlock latent demand.
Have you tried solving the problem? - Shaan's former Twitch boss Emmett Shear's go-to line for founders who over-architect a solution instead of taking the direct, unglamorous first step.

Summary

Sam Parr and Shaan Puri run the episode as a two-part riff with no outside guest: a "billy of the week" deep dive into Cargill, and a long historical detour into what they keep calling the Jevons paradox as it applies to AI. The Cargill segment opens with a Reddit thread on billion-dollar industries nobody's heard of, and unspools into the story of a company that started by building grain elevators next to new railroad lines - a pure middleman play that grew into the largest private company in America, 88% owned by one family, doing roughly $150 billion a year in revenue with a hand in nearly every packaged food product in the US. The hosts use Cargill's 80/20 reinvestment rule and multi-generational family structure to pivot into a broader conversation about what actually keeps family wealth and family businesses intact across generations: explicit governance (Hearst's ironclad legal trust, recurring family meetings, even a designed family crest) rather than the implicit, money-avoidant culture most American families default to.

The second half is Shaan's extended monologue on the Jevons paradox - the 1865 idea, from William Stanley Jevons's The Coal Question, that making a resource more efficient to use increases total consumption of it rather than shrinking it. He walks through the steam engine (more efficient coal use led to more coal burned, not less), the cotton gin (cheaper cotton processing led to more enslaved labor imported to keep pace with exploding demand, not less), and the Luddites (textile workers who rioted, under threat of execution, against machines that ultimately created more work than they destroyed). The through-line is a direct argument about AI: the hosts believe cheaper, AI-generated code will explode total demand for code and the people building with it, rather than eliminating jobs net-negative, and they point to Jensen Huang's public comments that AI inference demand will vastly outstrip any efficiency gains from cheaper model training as the same dynamic playing out live in the GPU market.

They spend real time on the emotional and political backlash to that argument - college graduation crowds booing a Goldman Sachs executive over AI comments, a New York official proudly banning data centers - framing today's anti-AI sentiment as a direct rerun of the Luddite movement and arguing it's understandable (particularly for a generation that graduated into a hollowed-out post-COVID job market) but ultimately futile, since "the AI freight train has left the station." The episode closes on two smaller threads: a riff on self-sabotage through over-engineering (anchored by Shaan's old Twitch boss Emmett Shear's line, "have you tried solving the problem?") and the Frederick Tudor "ice king" story - the entrepreneur who built a 19th-century business shipping New England ice to South America by manufacturing demand for cold drinks among local bartenders before people even understood what ice was for.

Throughout, the episode functions less as a single organized thesis and more as two hosts thinking out loud - jumping from Cargill's logistics moat to family trust law to 1800s economic history to AI labor anxiety to personal productivity traps - with the connective tissue being a shared belief that new efficiency-unlocking technology reliably gets under-forecast, over-resisted by whoever it displaces first, and ultimately net-positive for total demand and opportunity.

Notable Quotes

"I'm not selling ice, I'm selling winter." - Shaan Puri (retelling 19th-century ice merchant Frederick Tudor's pitch)

"The AI freight train has left the station. It is happening... You have zero chance of slowing down AI." - Shaan Puri

"It is what it is is the complete surrender of power, of agency, of anything. Just replace it with 'it is what I make of this.'" - Sam Parr

"Have you tried solving the problem?" - Shaan Puri, quoting his former Twitch boss Emmett Shear's standard response to founders who over-architect a plan instead of starting