Netflix

The mechanism is a subscription, and everything else is a consequence of it. Around 300 million households pay between about $8 and $25 a month, every month, for access to whatever is on the service. Because the fee is fixed and the content cost is fixed, every additional member costs Netflix close to nothing to serve and adds almost their whole fee to profit. That is the flywheel: more members fund more content, more content makes the service harder to cancel, harder to cancel means more members. It ran in reverse in 2022, when growth stopped for two quarters, the share price fell about 70% from its peak, and the whole industry declared the model broken.

What fixed it was two decisions Netflix had spent a decade refusing to make. The first was charging for password sharing: an estimated 100 million households were watching on someone else's account, and from 2023 Netflix asked them to pay for an extra member or their own plan. The second was an advertising tier, launched in November 2022 at a lower price, which Reed Hastings had publicly ruled out for years. Together they added over 40 million paid memberships in 2024 alone and pushed operating margin from 18% in 2022 to 27% in 2024.

The cost side is the content budget: about $17 billion of cash spent in 2024 on programming, most of it now originals Netflix owns outright rather than licenses. Owning the content is what allows the service to be the same in 190 countries, to keep a show forever, and to stop paying studios a toll that rises with Netflix's own success. It is also what turned Netflix from a distributor into a studio — with a studio's costs and a studio's hit-rate problem.

For most of its streaming life Netflix had no moat and knew it. Content was licensed, so studios could pull it — and did, once Disney, Warner and NBCUniversal decided to launch their own services. Switching costs were nil: cancelling takes two clicks and churn has always run around 2% a month. The genuine moat only formed once Netflix became the largest buyer of programming in the world and could spread $17 billion of content across 300 million members. Disney+ has to amortise its shows across a base a third of that size; Paramount+ and Peacock across far less. At that scale the arithmetic is decisive — Netflix can outspend everyone and still earn a 27% margin while its rivals lose money on the same activity.

Underneath the scale sits process. Netflix has spent twenty years learning what 300 million people watch, second by second, and uses it to decide what to commission, how to cut a trailer, and which of forty thumbnails to show you. That data is not decisive for any single show — hits remain hits for reasons nobody can model — but it is decisive for the portfolio, and the portfolio is what a subscriber is buying.

The brand is the third layer and the easiest to underrate: in most of the world Netflix is the default verb for watching television at home, which is why it can raise prices almost every year and lose fewer members each time than it gains. The moat's honest limit is not another streamer. It is YouTube, which is free, has more viewing hours in American homes than Netflix does, and is now the thing teenagers mean when they say television.

Netflix's capital allocation has been one long argument with its own shareholders about content. From 2013 it borrowed — eventually about $16 billion of high-yield debt — to fund original programming that produced negative free cash flow for a decade, on the theory that owned content was the only durable asset in streaming. The theory was right and the market spent years doubting it. By 2024 the same content budget, now around $17 billion a year, sits inside a company earning a 27% margin and generating nearly $7 billion of free cash.

The second phase has been discipline about what not to buy. Netflix has consistently declined to acquire a studio, a cinema chain or a sports league, on the stated logic that it can build cheaper than it can buy; it repurchased about $6 billion of stock in 2024 and pays no dividend. The exceptions are revealing. It has bought video-game studios and spent selectively on live rights — the Jake Paul–Mike Tyson fight, two NFL Christmas Day games, a ten-year deal for WWE Raw at roughly $5 billion — where the content is genuinely scarce and the audience arrives on a schedule that advertisers will pay for.

The December 2025 agreement to acquire Warner Bros. Discovery's studio and streaming assets breaks that pattern, and the case has to say so. It is a bet that owning HBO and a century of film library is worth more than the discipline of not buying. Whether it is right depends on a price and a regulatory outcome I cannot see from here; what I can say is that it changes the company from one that grew by building into one that has decided building is no longer fast enough.