A company sells hardware at a total loss. How do they build a multi-billion dollar monopoly?
Look at Nespresso.
The machine sits on your kitchen counter for $150.
The manufacturing cost is actually higher than the retail price.
They lose money on every physical unit shipped.
The aluminum capsule costs cents to make and sells for $1.
The margin on the coffee is over 80%.
You are not buying an appliance. You are buying access to a closed ecosystem.
Hardware is just the bait. Consumables are the trap.
The most boring business in real estate is a goldmine. A simple parking lot.
It’s just an empty patch of asphalt. No buildings, no interior design, no expensive engineering.
The startup cost is minimal: pave the dirt, throw up a fence, and install an automatic gate.
The operating cost is practically zero. No large staff, no inventory, no customer support. Security is just a camera.
The math is brutal: you pay a fraction for upkeep, but charge drivers $15 to leave a hunk of metal on your land for two hours.
Margins easily hit 70% to 80%. Meanwhile, the urban land underneath it appreciates every single year.
You don’t need to code a revolutionary app or build a rocket.
Sometimes, the ultimate wealth is just owning a scarce square meter where people have nowhere else to go.
Uber is the biggest taxi company on earth.
They own 0 cars.
Airbnb is the biggest hotel chain.
They own 0 rooms.
Traditional businesses sink billions into factories, fleets, and mortgages.
Platform giants skipped the heavy lifting entirely. They don’t buy the asset. They build the digital tollbooth.
Drivers pay for the car, gas, and repairs. Hosts pay the mortgage and cleaning fees.
The platform just takes 20% to 30% off the top of every single transaction.
Zero physical liability. Infinite scale.
Owning physical assets is a financial anchor. Owning the marketplace where everyone is forced to trade is an empire.
Would you rather own the fleet, or the software that taxes every ride?
A small can with a plain logo sits on a desk in a warehouse somewhere in Austria. Inside: water, sugar, caffeine, taurine. About ten cents to make.
Everything around the can costs a billion dollars.
Two F1 teams. Football clubs in three countries. A media studio that films BASE jumpers and cliff divers. Hundreds of athletes on exclusive contracts. An aerospace program that sent someone to the edge of space.
Red Bull spent thirty years building a world around a ten-cent product, and then watched people pay three dollars to enter it.
Most companies see this and assume the drink must be massively profitable to fund all of it. They argue about flavor profiles, caffeine content, can design.
None of that is the business.
Red Bull owns zero factories. Every can is produced by partners — primarily an Austrian co-packer called Rauch. No plants, no equipment to depreciate, no production lines to maintain. That decision freed up the entire balance sheet.
A normal consumer goods company spends most of its money upstream — factories, raw materials, logistics. Whatever's left goes to marketing. That's the whole game.
Red Bull flipped the order. The drink was not the business. The drink was the receipt.
The business is the attention.
You can sell a hundred million cans, or you can sell the same cans while owning an F1 team, and the second version of the company is worth ten times the first. Not because the cans changed. Because the world around them did.
Most companies compete on price. They fight over the same shelf at the same gas station for the same dollar from the same tired commuter.
Red Bull competed on culture. They didn't sell energy. They sold the lifestyle that energy belongs to — the late nights, the dirt bikes, the stadium jumps, the wingsuit flights. The drink became a passport. Buy it, and you're in the world Red Bull built.
That's why the can costs three dollars when production costs ten cents. You're not paying for the liquid. You're paying for the membership.
Watch the edit again. Notice how the drink barely appears. The whole piece is motion — skydivers, rally cars, motocross, a guitarist on a rooftop. The product is an afterthought.
That's not an accident. That's the business model.
Somewhere between thirty and thirty-five percent of Red Bull's revenue goes back into sports, content, and events every year. PepsiCo spends about seven percent on marketing. Red Bull spends five times that — but it's not marketing. It's structural. It's the cost of running a parallel empire — media, sports, events — that makes the main business impossible to compete with.
Every dollar poured into an F1 car is also a dollar poured into a content engine that keeps the brand connected to the only demographic that actually pays premium prices for anything: young people who want to feel alive.
The product is the cheapest part of the business.
What you're selling is the world the product sits inside.
Anyone can make a can of sweet water for ten cents. Almost no one can make the world around the can worth three dollars. Red Bull spent thirty years doing it while their competitors argued about the recipe.
The recipe was never the point.
You paid $1,200 for a phone. The most expensive part costs $3 to make. Want to know why?
That part is the chip. Jensen Huang runs the company that designs it.
Nvidia doesn't make chips. Nvidia designs them. Taiwan Semiconductor makes them. That's TSMC. A Dutch company called ASML builds the only machine on Earth that can physically print the most advanced ones.
Three companies. Three countries. One bottleneck.
Here's what actually happened.
In the 1980s, Morris Chang built TSMC in Taiwan on a bet. His bet: chip design companies would rather pay someone else to make their chips than build their own factories. Building a fab costs tens of billions. TSMC would build them once and rent the capacity to everyone. TSMC controls about70% of the global chip manufacturing market today.
In 1984, Philips spun off a small Dutch project called ASML. For years it was a money-losing afterthought. Then it spent decades building the only machine that can etch circuits at the scale modern chips require. Today, ASML holds about 83% of the global lithography market. Its most advanced machine: extreme ultraviolet lithography. EUV for short. Each one costs up to $200 million. Weighs 180 tons. Needs three Boeing 747s to ship.
In 1993, Jensen Huang founded Nvidia in a Denny's diner to make graphics chips for video games. He bet that graphical processing would become the future of computing. For20 years, the company mostly sold chips to gamers. Then AI happened. Nvidia's chips turned out to be exactly what training large AI models requires.
Nvidia's net income last fiscal year: $120 billion. Its market cap briefly crossed $5 trillion in 2025.
The "silicon" itself? The raw material in a chip costs a few dollars. What you're actually paying for is the30-year head start on solving problems nobody else has solved.
Three companies each control something that cannot be substituted. ASML: the machine. TSMC: the factory. Nvidia: the design.
You don't beat a chip by making it cheaper. You beat a chip by being the only place that can make it.
You spend 6 months building a product. A platform you don't control takes 30% of every sale. And somehow Jeff Bezos became the richest person on earth.
Bezos figured out the rule everyone else missed.
In 2006 he launched Fulfillment by Amazon. 3rd-party sellers could now store products in Amazon's warehouses and ship with Prime's badge. Amazon would handle the entire logistics chain.
Today ~60% of everything sold on Amazon comes from 3rd-party sellers. Amazon didn't build most of those products. Amazon takes15–45% of each one.
The maker carries the inventory risk. Amazon carries nothing.
Here's the trick.
Distribution is harder than production. A factory in Shenzhen can build a $5 cable in a week. Getting it in front of 200 million Prime subscribers in 2 days takes billions.
So the bottleneck moves. Whoever owns the bottleneck owns the margin.
Same play, different industry. Alphabet reported $403B in revenue in 2025. ~86% came from advertising.
Google built the most powerful attention-distribution machine in history. Then sold access to every business that wants attention. Businesses make the products, run the ads, take the risk. Google takes a cut of every impression.
The casino doesn't need to win at the table. It rents the chairs.
Visa processed233B transactions in 2024. Took ~3% per swipe. The card is a marketing expense. The rail is the business.
The pattern is identical. Own the rails. Charge a toll. Never touch the cargo.
This is why "make a great product" stopped being enough. Value migrated from the product to the position the product gets sold from. From the factory to the platform. From the maker to the middleman.
If you want to get paid like Bezos, don't try to win on the widget. Try to win on the position every widget-maker has to pass through.
The best business isn't the one that makes the best thing. It's the one nothing reaches the customer without.
Costco has kept its famous hot dog combo at $1.50 for over 40 years.
In 2013, the CEO asked founder Jim Sinegal if they could raise the price.
Sinegal's answer:
“Raise the effing price and I’ll kill you. Figure it out.”
So they figured it out.
The price stayed at $1.50.
Through inflation.
Higher meat costs.
Higher wages.
Higher transportation costs.
And in 2025, Costco sold more than 245 million of them.
The hot dog isn't really about the hot dog.
It's a signal.
Every time a customer sees that $1.50 price, Costco reinforces the same idea:
You can trust us to give you a deal.
Most companies optimize every transaction.
Costco optimized what customers expect from the next one.
The hot dog isn't the product.
The trust is.
Jensen Huang says linking AI to job loss is "too lazy."
Here's why he's only half right.
His logic: AI makes one engineer way more productive. If one engineer can now create a crazy amount of value, why hire fewer of them, you hire more. The output is too good to stop at a few.
But here's the second-order problem nobody talks about.
If 1 person using AI can match what 5 used to do, companies don't necessarily need to hire the other 4. AI doesn't have to replace you directly. It just has to make fewer people necessary.
And that's a much bigger problem.
His whole argument only works if the company actually wants to keep making more and more forever. Plenty of them don't. They just want the same thing cheaper.
So AI can make the best engineers way more valuable AND cut how many people you need for the same job at the same time.
That's the bit people mix up.
Being more productive and hiring more people were never the same thing.
So how many engineers do you think a company actually needs when one of them can ship in 1 day what used to take a team a month? Really.
⚡️Tesla just showed us the inside of Terafab.
A render of the new factory where Optimus and Robovan appear in the same frame for the first time.
Robovan gliding on an elevated track through the building. Optimus on the floor. Cybercab and Semi parked nearby.
It's a render. Not a photo. Not a video of a real facility.
But here's what makes it interesting: the render isn't trying to sell us a factory. It's selling us a workflow.
Robots moving products. Robots moving people. The whole building as one machine.
If even half of this lands by 2027, Tesla stops being a car company and becomes a robotics manufacturer with its own internal customer - a factory that builds the next version of itself.
Who else has a factory big enough to be the first buyer of its own robots?👇