The city skyline was a matte painting. The shark was rubber and it barely worked. The crowd of 40,000 was 400 extras and a lot of patient duplication. Audiences have never actually wanted the truth from a screen. We want the trick to hold.
Hollywood has always been comfortable with that bargain. What it has never been comfortable with is a change in who performs the trick.
That is the fight that shut the industry down in 2023, and it is the fight that has kept studios careful about how they discuss artificial intelligence (AI) ever since. Everyone has been running pilots. Almost nobody has been willing to say how many. The safe play was to call it experimentation, name a single project, and move on before anyone asked a follow-up question.
The streamer told shareholders in its second-quarter letter that roughly 300 of its titles used generative AI somewhere in production this year, with most of that work landing in post-production, according to Variety.
The letter described the technology showing up across the full production process, from concept and previsualization through post-production and release. Netflix said the tools help it deliver “higher-quality output more quickly and at a lower cost,” according to Variety.
Three titles got named on the second-quarter earnings call: the Indian sports thriller “Glory,” the Brazilian soccer miniseries “Brasil 70: A Saga do Tri,” and the American Revolution docuseries “The American Experiment.”
Co-CEO Ted Sarandos said the tools handled “really complicated shots and sequences,” including enhanced crowds and historical battle scenes, according to Deadline.
The most specific claim involved that docuseries. It contains 17 minutes of AI-assisted footage produced “twice as fast and at half the cost,” per Fortune.
Netflix does have guardrails on paper. Production partners must flag intended AI use, and generating performances or replacing union-covered work requires consent and the right agreements, according to Digital Trends.
What struck me working through the earnings coverage was the order of the argument. Sarandos led with creative scope and reached cost second. AI “will give creatives better tools to bring their visions to life,” he said, according to IndieWire.
That sequencing is a choice, and it is the tell. Cost is the part shareholders are paying for.
Why the 300 number matters more than the earnings miss
Netflix reported second-quarter revenue of $12.56 billion, up about 13% year over year, with earnings of 80 cents per share. The stock still fell as much as 9% after hours.
Investors were not reacting to the AI disclosure. They were reacting to deceleration. Revenue growth has slid from 16% in the first quarter to 13% in the second, with roughly 12% guided for the third, according to Fortune.
That context is what makes the 300 figure load-bearing. A company growing 20% a year does not need to talk about production efficiency. A company guiding to 12% does.
It also explains the timing. A year ago the streamer was pointing to a single production that used generative footage. Getting from one title to about 300 took four quarters, faster than any tooling change Hollywood has absorbed in decades.
The path there was short:
Netflix acquired InterPositive, the AI filmmaking company co-founded by Ben Affleck, in March to put production tools in filmmakers’ hands, according to Variety.
Chief product and technology officer Elizabeth Stone said in June that generative AI would make choosing what to watch more personalized and immersive, according to Bloomberg.
Roughly 300 titles used generative AI workflows this year, concentrated in post-production, according to Deadline.
Content spending is set to rise about 10% in 2026 even with those savings, according to Deadline.
Netflix tells shareholders that roughly 300 of its titles used generative AI in production this year.Vertigo3d / Getty Images
What the AI savings actually buy Netflix subscribers
Sarandos said the savings will most likely be reinvested in more programming rather than used to shrink the content budget. That budget is heading toward about $20 billion this year, up from $17.1 billion in 2025 and $16.2 billion in 2024, according to Fortune.
Read that plainly. Your subscription does not get cheaper. The output gets bigger.
That is the unglamorous version of the AI efficiency story in almost every industry. Savings rarely reach the customer. They get spent on more of whatever the company already wanted to buy, which for Netflix means more hours of programming to defend against YouTube and everything else competing for the same evening.
It fits a pattern subscribers can already see, since the company has been building generative tools into the ad experience itself, and it explains why Netflix keeps circling assets that help people decide what to watch, as TheStreet reported.Â
What Netflix’s AI disclosure signals for your portfolio
The AI trade so far has been a story about sellers. Chipmakers, cloud providers, and everyone supplying the shovels.
Netflix is one of the first large consumer companies to put a number on what it gets back as a buyer. That matters because the bull case for AI outside semiconductors rests on an assumption nobody has been able to price, which is that ordinary companies eventually convert these tools into margin.
Three hundred titles and one documentary produced at half the cost is not proof. It is the first serious data point from a company with every incentive to undersell a technology its own creative partners resent.
More Entertainment:Â
In my analysis, the disclosure itself is the tradable signal. Companies quantify things in shareholder letters when those things turn material. Netflix would not have printed this number in 2024. It printed it now.
Two things are worth watching. The first is whether Netflix’s operating margin, already above 30%, starts pulling away from its content spend in the third quarter, which is the quarter analysts setting price targets will be scoring.
The second is whether Disney (DIS), Warner Bros. Discovery (WBD) and Amazon (AMZN) start disclosing numbers of their own, because once one competitor quantifies AI savings, silence from the rest stops reading as caution and starts reading as a gap.
There is one more thing the 300 figure buys, and it is not for investors. Post-production is skilled, expensive, deadline-bound work that most people assumed was safe. It turned out to be the easiest place to start.
The next time a crowd looks a little too full, you will not be able to tell. That was always the point.
