Netflix

With 301.6 million subscribers across 190+ countries, Netflix isn’t just a streaming platform anymore. It’s a data company that happens to produce entertainment.

Think of your patterns on Netflix.

Every time you hit play, pause, rewind, or scroll past a title, Netflix is learning from your behaviour. The company processes billions of events daily from user interactions.

This dynamic data + real-time learning and testing make Netflix one of the biggest entertainment enterprises.

In today's AI at the Top, we will learn how Netflix uses AI to personalise your experience, decide what shows to produce, and deliver content without buffering.

Netflix has been obsessed with data long before AI became mainstream.

Remember when Netflix was mailing DVDs? Even then, they tracked which movies you rented, how long you kept them, and what you rated them.

When streaming launched in 2007 (it was called Watch Now back then), that data collection exploded. Today Netflix tracks what you watch, how long you watch before stopping, what thumbnails you click on, and whether you binge or spread episodes out.

This real-time behavioral data is what makes Netflix’s AI effective. Similar to how Starbucks used its mobile app to understand customer preferences, Netflix built its entire business on knowing what you want to watch before you do.

Over 80% of what people watch on Netflix comes from recommendations, not from search.

Most of us open Netflix, scroll for 90 seconds, and move on to another OTT or YouTube if nothing catches our attention. Netflix knows this, so they’ve built an AI system that makes sure you find something interesting fast.

There’s more.

Netflix doesn’t just recommend what to show you. They also decide how to show it.

The same show gets different thumbnail images for different users. For House of Cards, viewers who prefer movies with strong female leads see thumbnails featuring Robin Wright. Those who watch political dramas see Kevin Spacey.

Netflix tests these thumbnails across millions of users. The artwork personalization (thumbnails, synopsis, trailers, and so on) increased CTRs by ~30%.

The homepage you see is completely unique to you. No two Netflix users see the same interface. This personalization saves Netflix over $1 billion annually in reduced churn.

Netflix spends around $17 billion on content every year, making it important to decide what to produce.

Before Squid Game became Netflix’s biggest show ever, the data showed strong interest in Korean content globally. They greenlit the show based on this insight, and it led to 1.65 billion viewing hours in its first 28 days.

Stranger Things has a similar story. Netflix’s data revealed an interest in 80s nostalgia and sci-fi. The show reached 14.07 million viewers within one week of its release.

In our opinion, this is a double-edged sword. In art, there will always be scripts that challenge patterns, trends, and make it to the top.

Relying on the AI to greenlight shows may not always be the best process, but the data will at least show what productions will likely fail.

Netflix recently started using generative AI in actual production.

In The Eternaut, an Argentine sci-fi series, they used AI to generate a building collapse scene. It was created faster than traditional VFX. For Happy Gilmore 2, Netflix de-aged characters with AI.

Netflix understands how users feel about buffering videos, and it’s bad for business and CX. To solve this, Netflix developed its proprietary CDN (content delivery system) called Open Connect.

Most streaming services rely on third-party CDNs. Netflix built its own. They place cache servers directly inside internet service providers’ data centers around the world. Then your content is streamed from a server physically close to you.

Think of it like warehouses. Your order delivers faster if the warehouse is in the same city as yours.

Netflix delivers all its content with Open Connect. That’s over 250+ million viewing hours per day.

This infrastructure + AI integrations adjust video quality based on your internet speed, predict which content you’ll watch next and pre-cache it, and optimize video encoding for each scene.

Among many use cases, Netflix runs agents that act autonomously and share data with each other to improve the platform overall.

Netflix will continue expanding generative AI use in production workflows, especially for VFX sequences and pre-production visualization.

They have set up clear guidelines on how to use and not use AI for its partners, ensuring AI turns into a powerful tool for storytellers but not a threat to their jobs.

Netflix’s recommendation systems will get even smarter to show the best content based on its users’ experience. Experiments and testing continue.

The company reported 17% YOY revenue growth to $11.5 billion in their Q3 2025. With this kind of momentum, their AI investments will only compound.

The streaming wars are far from over. But Netflix’s data advantage and AI infrastructure give them a lead that competitors will struggle to match.

YOUR TURN