I first wrote about AV1 in April 2019 when it was just a baby codec. I covered the monthly ecosytem updates a bunch more times (which you can find in the archives) eventually culminating in the AV1 2019: A Year In Review here.

Half a decade later with widespread adoption (but still not as much as H.264), it’s still hard to find a straightforward analysis of how it performs in the real-world deployments. As a codec enthusiast, I was curious. How is Netflix using AV1?

So I decided to find out. Armed with a Fire TV, a MacBook, and the Android Debug Bridge (ADB), I spent several weekends capturing and analyzing Netflix streams. My methodology involved a level of patience I’ve only previously attained while debugging a newborn’s requirements at 3 AM.

The assumption is that AV1, being newer, is more efficient. It is. But the full story is more nuanced. For some content, Netflix’s AV1 encodes are so focused on quality that they actually use more data than the HEVC equivalent.

TL;DR - The Numbers Don’t Lie

Before we dive deep, here’s the high-level summary.

• Up to 55% bandwidth savings on blockbuster movies.
• Peak bitrate spikes often cut in half during action scenes, reducing buffering.
• This efficiency could translate to ~50% more viewing hours on a mobile data plan.
• The catch: Not all content benefits equally—some animated shows use more data with AV1.

The following table is your cheat sheet to the entire analysis.

How I Accidentally Became a Netflix Data Stalker

Look, I wasn’t planning to spend my evenings decoding Netflix streams like some kind of bandwidth detective. But after seeing the 47th “AV1 is the future” post without actual data, I got annoyed enough to do something about it.

The Setup:

• Fire TV Stick (because it actually reports real bitrates)
  • 4K Max model (2024) for receiving AV1 streams
  • 4K model (2021) for receiving H.264 and H.265 streams
• Python script leveraging Android Debug (ADB) Bridge capturing data every 5 seconds
• 5 viewing sessions per title to account for network conditions
• 30-second windows aggregated into min/max/average values
• The data was collected near the beginning of 2024 (I know! :) life got in the way of publishing) so Netflix might have made their encoding algorithms even more efficient at the time of publishing this blog (Oct 2025).
• The charts were created with Datawrapper

Here’s what the debug info looks like on FireTV

The Victims:

• 🍌 Minions (CGI animation) - AV1 vs H.264
• 🎬 Catch Me If You Can (live-action film) - AV1 vs H.264
• 🎩 Peaky Blinders S1E1 (live-action TV) - AV1 vs H.264
• 🧪 Breaking Bad S1E1 (grainy film) - AV1 vs H.265
• 🐴 Bojack Horseman (2D animation) - AV1 vs H.265

Yes, I watched the Breaking Bad pilot ten times for science. My Netflix recommendations are permanently broken.

Part 1: The Foundational Efficiency Gains

First, let’s establish the baseline. Is AV1 actually more efficient? Yes, and by a significant margin.

Finding #1: Average Bitrate Reduction is Massive

AV1 is competing against the incumbent H.264 and the modern alternative H.265/HEVC. It performs remarkably well against both.

Let’s take Peaky Blinders as an example—a show famous for its cinematic style, complete with smoke and moody lighting. The average bitrate for the AV1 stream was 48.1% lower than H.264. That’s nearly a 50% gain over the most common codec on the planet.

Finding #2: Aggressive Savings in Quiet Scenes

Part of AV1’s efficiency comes from how low it can drop the bitrate during simple, static scenes (like two people talking). It saves bits during the quiet moments to spend them on action sequences.

Look at the average minimum bitrate for each stream. AV1 consistently establishes a much lower “efficiency floor.”

Every bit saved here is a bit that can be reallocated to preserve quality during a high-motion scene.

Finding #3: The Payoff in Gigabytes

Percentages are boring. Here’s what AV1 actually saves you in actual data:

• 🎬 Catch Me If You Can (141 min): 1.94 GB saved
• 🍌 Minions (91 min): 1.48 GB saved
• 🧪 Breaking Bad S1E1 (58 min): 0.93 GB saved
• 🎩 Peaky Blinders S1E1 (59 min): 0.7 GB saved

For a 10-episode season of Peaky Blinders, that’s 7 GB saved. Enough for three additional HD movies.

Part 2: Quality, Complexity, and a Few Surprises

Data reduction is just one part of the story. AV1’s real advantage lies in how Netflix’s encoding strategy uses it to handle visual complexity.

Finding #4: The Action Scene Advantage

You might assume a codec saves data by compromising during chaotic action scenes. AV1 often does the opposite. Its efficiency advantage can actually increase when the on-screen complexity ramps up. To show this, I segmented each stream into four “quartiles” of complexity based on bitrate. Q1 is the quietest 25% of scenes, and Q4 is the most frantic 25%.

Notice how Peaky Blinders actually gets MORE efficient as scenes get harder? A perfect example of this in action is the CGI-animated film Minions. The chart below shows the H.264 stream (red) spiking wildly during action scenes, while the AV1 stream (green) remains remarkably composed.

When the screen is filled with smoke, rain, and fast action, AV1’s advantage widens. It thrives under pressure.

Finding #5: The Peak Bitrate Anomaly

Peak bitrates are what cause your stream to buffer. Yet, the single highest bitrate spike across all five titles didn’t come from an older codec. It came from AV1.

The hall of fame (or shame):

• AV1’s Wildest Moment: The AV1 stream for Breaking Bad hit a staggering 22,573 kbps spike.
• HEVC’s Response: For the same scene, the HEVC stream topped out at a more controlled 19,428 kbps.
• The Norm: In contrast, during the most chaotic action scenes in Minions, H.264 peaked at over 10 Mbps, while AV1 stayed calmly under 4 Mbps.

This isn’t a bug. It’s a deliberate choice to prioritize visual quality above all else, which leads to the next finding.

Finding #6: The Breaking Bad Paradox

Why would a “more efficient” codec need a 22 Mbps spike? This is a direct result of Netflix’s shot-based encoding strategy. Their system analyzes the show scene-by-scene and identifies its signature 35mm film grain as a crucial, and very data-intensive, part of its artistic identity. The encoder’s directive is clear: preserve that grain at all costs.

This results in a smart trade-off:

• It aggressively compresses simple scenes (like quiet dialogue) to save bits.
• It then spends those saved bits on complex scenes to render the heavy film grain with absolute fidelity, even if it causes a massive, temporary spike.

Finding #7: The Bojack Horseman Mystery

The fact that AV1 used slightly more data for Bojack Horseman is the other side of the same coin: Netflix’s focus on perceptual quality. For this show, the shot-based analysis isn’t worried about film grain, but about preserving the clean lines and solid colors of 2D animation.

The system recognizes that a tiny extra investment in bitrate can deliver a disproportionately large improvement in visual quality, eliminating the subtle artifacts that older codecs might introduce around character outlines. HEVC gets to “good enough” at 1.0 Mbps, but Netflix has directed AV1 to achieve “pristine” at 1.1 Mbps. It’s a worthwhile trade-off to ensure the animation looks exactly as intended. It’s the difference between a decent burger and a gourmet one for just a few cents more

Part 3: The User Experience - A Smoother, More Stable Stream

What does this all mean when you hit “play”?

Finding #8: Winning the Buffering Battle

Imagine a 10 Mbps internet connection. Let’s count how many times each stream would cause buffering by exceeding that limit.

For Catch Me If You Can, the H.264 stream crossed the 10 Mbps line 16 times. The AV1 stream? Zero. For two of the titles, AV1 effectively eliminated buffering on this hypothetical connection.

Finding #9: Faster Start-Up, Less Waiting

A high initial bitrate can lead to a long wait before a video starts on a slow connection. Across the board, AV1 requires less data to get going.

Real-world impact: That hotel WiFi that struggles to load email now has a better chance of starting a Netflix stream without a long initial buffer.

Finding #10: A Wider “Safe Zone” for Mobile Users

For users on mobile data or congested networks, the goal is often to stay below a low bitrate threshold (e.g., 2 Mbps). AV1 spends significantly more time in this “safe zone.”

AV1 makes HD streaming viable on connections that would have struggled with older codecs. Your mobile data plan just became a lot more useful.

The Bottom Line: What This Actually Means

Let’s bring this all together.

For Viewers:

• On Mobile: A 10GB data plan that gave you ~12 hours of Netflix with H.264 could now deliver closer to 20 hours with AV1.
• On Bad Wi-Fi: AV1’s lower bitrate floor and fewer spikes make streaming on unstable connections more reliable.
• On a Great Connection: You get a higher-fidelity picture, particularly in complex scenes, because the codec and encoding strategy are using your bandwidth more intelligently.

For Netflix: My back-of-the-napkin math, using conservative industry estimates for bandwidth costs, puts Netflix’s potential savings in the ballpark of $25M/year. That’s money that can be invested back into content.

The Catch (There’s Always a Catch)

AV1 isn’t a magic bullet. It has two major hurdles:

1. Encoding Cost: AV1 is computationally expensive to encode. This requires a massive, ongoing investment in server infrastructure for any streaming service.
2. Device Support: Hardware decoding for AV1 is still not ubiquitous. This is a classic chicken-and-egg problem that will take a few more years of device refresh cycles to resolve.

The Verdict: A Revolution in Progress

Despite the hurdles, the real-world data is clear. AV1 delivers on its promise with impressive results. It’s smarter in quiet moments, more composed during action, and is guided by an encoding strategy that makes intelligent trade-offs between efficiency and perceptual quality. The transition will be slow and expensive, but it’s no longer a question of if AV1 will dominate streaming, but when.

Contact

Reach out if you have any questions! Feel free to follow me on

• Twitter - @singhkays
• LinkedIn - https://www.linkedin.com/in/singhkays/

Found this useful? Share it with anyone who’s ever cursed their internet connection during a Netflix binge. And yes, I realize I spent a few weekends proving that Netflix’s engineers know what they’re doing. No regrets.