“Should we switch to AV1 for our video platform?” we’ve heard this question at least a dozen times in the past six months from product managers looking to optimize their WebRTC applications.
The answer keeps surprising them: probably not yet.
Despite years of hype, better compression numbers that look amazing on paper, and backing from tech giants like Google, Netflix, and Meta, AV1 isn’t dominating real-time video in 2026. VP8 and H.264 — codecs that are over a decade old — still power the majority of WebRTC sessions.
Why? The answer comes down to something unglamorous but critical: CPU cost and the stubborn reality that “good enough” codecs don’t get replaced overnight.
Here’s what you need to know about AV1’s actual state in 2026, based on what’s happening in production systems, not marketing materials.
The Promise vs. The Reality
Let’s start with why everyone got excited about AV1 in the first place.
What AV1 Was Supposed to Deliver
AV1 (AOMedia Video 1) arrived with serious credentials. Developed by the Alliance for Open Media — a consortium including Google, Mozilla, Netflix, Amazon, Microsoft, and Apple — it promised to be the successor to both VP9 and H.265.
The technical advantages are real:
- 30-50% better compression than VP9 and H.264 at equivalent quality
- Royalty-free licensing with no patent concerns
- Built-in SVC (Scalable Video Coding) for flexible quality adaptation
- Optimized for screen sharing with special coding tools for text
- Modern features like HDR support and 8K capability
For streaming giants like Netflix and YouTube, these advantages translate to massive savings. Netflix reports that AV1 uses one-third less bandwidth than both H.264 and H.265 while achieving higher quality scores. By the end of 2025, they had converted 85% of their HDR catalog to AV1.
Sounds like a no-brainer, right?
Why It’s Not Taking Over WebRTC
Here’s where theory meets reality.
Around 90% of WebRTC video sessions in 2026 still use VP8. That’s not a typo—ninety percent. Google Meet uses AV1 in some scenarios, Jitsi has experimental support, but the vast majority of video calls, virtual events, and real-time applications stick with the old guard.
The gap between “better on paper” and “dominant in practice” comes down to three hard truths:
1. CPU Cost Is Brutal
AV1 encoding is computationally expensive — about 3-5x more intensive than VP9 on typical desktop CPUs, and VP9 already requires more processing power than VP8 or H.264.
What does that mean in practice?
- Most devices can’t encode AV1 in real-time without hardware acceleration
- Battery life takes a significant hit on mobile devices using software encoding
- Server-side encoding costs spike for platforms that need to transcode video
- Older laptops and budget devices struggle or can’t handle it at all
One platform engineer told us, “We tested AV1 for our group video product. The compression was beautiful, but half our users’ laptops sounded like they were about to take off. We rolled it back.”
2. Hardware Acceleration Is Patchy
Hardware acceleration solves the CPU problem — when it’s available. The challenge is that AV1 hardware support is still rolling out unevenly across devices.
Here’s the landscape in 2026:
Desktop/Laptop:
- Intel: 11th gen and newer (Tiger Lake+) have AV1 decode; Meteor Lake (2024+) added encode
- NVIDIA: RTX 30/40/50 series support AV1 decode; RTX 40/50 added encode
- AMD: Ryzen 7000+ series support AV1 decode and encode
- Apple Silicon: M3 and A17 Pro chips (2023+) have AV1 hardware decode—but no encode yet
Mobile:
- High-end Android: Most flagship phones from 2023+ (Snapdragon, MediaTek flagships) support hardware decode
- Mid-tier Android: Limited support; expected to improve by late 2026 but not there yet
- iPhone: iPhone 15 Pro and iPhone 16+ have hardware AV1 decode; older models rely on software
- Low-end devices: No hardware support
Smart TVs and Streaming Devices:
- Most 2020+ models from Samsung, LG, Sony support AV1 decode
- Older devices: Software decode only, if at all
The problem? For a codec to become the default in WebRTC, it needs to work well everywhere. AV1’s hardware acceleration is getting there for decode, but encode support — critical for real-time video — remains limited to newer, high-end devices.
3. VP8 and H.264 “Just Work”
This is the unsexy reality that kills codec transitions: the old stuff works fine.
VP8 has been in WebRTC since the beginning. It’s:
- Universally supported across all browsers and devices
- Stable and thoroughly debugged after a decade in production
- Computationally manageable on virtually any hardware
- Royalty-free with no licensing concerns
H.264 is even more entrenched:
- Hardware encoding and decoding on almost every device made in the last 15 years
- The default codec for cameras, security systems, and IoT devices
- Required by Safari and iOS for WebRTC (Apple won’t negotiate with VP8)
- Licensing is free for most use cases (under 100,000 concurrent users)
As one senior architect put it, “In two years, VP8 will still just work. Why would we break something that works to adopt something that only works on some devices?”
The Adoption Pattern We’re Actually Seeing
If AV1 isn’t dominating, where is it gaining traction?
Where AV1 Is Winning: Streaming and VOD
For non-real-time video, streaming services, video-on-demand, archived content, AV1 is making serious inroads.
YouTube, Netflix, and Meta have all deployed AV1 extensively for playback. The use case is different:
- Content is encoded once, server-side, with time to optimize
- Playback only requires hardware decode, which is more widely available
- The bandwidth savings pay off immediately at scale
- Latency isn’t critical, so buffering can absorb processing delays
Netflix went from 0% to 85% AV1 coverage of its HDR catalog in just two years. That’s impressive adoption—but it’s for pre-encoded, one-to-many delivery, not interactive video calls.
Where AV1 Struggles: Real-Time Interactive Video
For WebRTC applications — video conferencing, telehealth, live collaboration — the constraints are different:
- You need fast, lightweight encoding on the sender’s device
- Both sender and receiver need to support the codec
- Latency requirements are strict (sub-second, ideally under 300ms)
- Users have diverse hardware, from high-end workstations to budget laptops
Many WebRTC services haven’t even fully adopted VP9 yet, despite it being available since 2013. VP9 offers better compression than VP8, but it also requires more CPU. Most platforms stick with VP8 for broad compatibility and predictable performance.
Based on this pattern, industry experts predict that if AV1 becomes dominant in WebRTC, it won’t happen until 2028 at the earliest—and even that assumes hardware acceleration becomes ubiquitous.
The Real Numbers: Codec Usage in 2026
Here’s a snapshot of codec distribution across different video applications in 2026:
| Use Case | Dominant Codec | AV1 Adoption | Reason |
| WebRTC video calls | VP8 (~90%) | Experimental (~5%) | CPU cost, device compatibility |
| Browser streaming (YouTube) | AV1 majority | 70%+ | Decode-only, hardware support |
| Netflix, on-demand | AV1 majority | 85%+ | Server-side encode, broad decode support |
| Live broadcasting | H.264, VP9 | Limited (<10%) | Encoding performance matters |
| Security cameras | H.264 | Minimal (<1%) | Legacy hardware, “just works” |
| Screen sharing (WebRTC) | VP8, VP9 | Growing (~15%) | AV1’s text coding helps, but CPU cost limits |
VP8 and H.264 remain the workhorses because they balance quality, performance, and compatibility in ways that matter for production deployments.
The Hidden Factor: AI Upscaling Changes the Game
There’s another twist that makes the codec race even more interesting: AI-based upscaling is improving fast.
YouTube recently rolled out “enhanced bitrate” features using AI to upscale lower-resolution videos on playback. The result? You can upload at 720p, and viewers see something approaching 1080p quality—without the bandwidth or storage cost of sending full HD.
This raises a fundamental question: Should we invest in better compression on the sender side, or better enhancement on the receiver side?
If AI can upscale efficiently, the value proposition for ultra-advanced codecs like AV1 shifts. Instead of burning CPU to send perfect 1080p or 4K streams, you could send optimized 720p with VP8 and let the receiver’s device upscale intelligently.
One engineer working on video optimization told us, “We’re not sure anymore if the endgame is AV1 or just smarter post-processing. The calculus is changing.”
This uncertainty means teams are less willing to bet heavily on codec transitions when the best practice might shift entirely in the next few years.
When Should You Actually Consider AV1?
Despite everything we’ve said, there are scenarios where AV1 makes sense in 2026.
You Control the Client and Target Modern Devices
If you’re building a native app where you can specify hardware requirements—think enterprise tools, premium services, or controlled deployments — AV1 becomes viable.
Examples:
- Internal company video systems where you know everyone has recent hardware
- Gaming platforms targeting high-end PCs with RTX 40/50 series GPUs
- Premium streaming services where you can require iPhone 15+ or recent Android flagships
Your Primary Use Case Is Broadcasting, Not Peer-to-Peer
If you’re distributing a single stream to many viewers — webinars, live events, game streaming — you can encode once on powerful server hardware and benefit from AV1’s superior compression.
The server handles the encoding cost, and viewers only need decode support, which is more widely available.
You Have a Long Planning Horizon
If you’re building a platform that won’t ship for 18+ months, betting on AV1 makes more sense. By late 2027 or 2028:
- Hardware support will be more pervasive, especially for mid-tier devices
- Browser and SDK implementations will be more mature
- Tooling and best practices will have solidified
Building with AV1 in mind now — while keeping VP8/H.264 fallbacks — positions you well for the transition.
Screen Sharing Is a Core Feature
AV1 has specialized coding tools for screen content (text, UI elements, line art) that make it exceptionally efficient for screen sharing compared to VP8 or H.264.
If your application is built around screen sharing — remote desktop, collaborative design, pair programming — AV1’s advantages are more compelling.
The Realistic 2026-2028 Timeline
Here’s what the codec landscape likely looks like over the next few years:
2026 (Now):
- VP8 and H.264 dominate real-time video
- AV1 usage grows in streaming and VOD
- Experimental AV1 in WebRTC for controlled use cases
- Hardware support expands but remains incomplete
2027:
- Mid-tier Android devices start getting AV1 hardware encode
- More platforms add AV1 as an option alongside VP8
- Hybrid architectures emerge: VP8 for compatibility, AV1 where supported
- Safari might add better codec negotiation (maybe)
2028:
- AV1 reaches majority adoption in high-end devices
- Becomes the preferred codec for new platforms
- VP8 remains as fallback for legacy devices
- H.264 still required for Apple ecosystem interop
The transition won’t be a clean cutoff. It will be gradual, messy, and driven by hardware deployment cycles more than technical superiority.
What Trembit Recommends for 2026
If you’re building or optimizing a real-time video application right now, here’s our practical advice:
For Most WebRTC Applications:
- Stick with VP8 as your primary codec in 2026
- Design your architecture to make codecs swappable
- Monitor AV1 hardware penetration in your target audience
- Test AV1 in controlled scenarios but don’t bet the business on it yet
For Streaming/Broadcasting:
- Use AV1 for server-side encoding where you control the hardware
- Implement adaptive codec selection based on device capability
- Keep H.264 or VP9 as fallback for older devices
For Enterprise/Controlled Environments:
- If you can mandate device requirements, AV1 is viable
- Test real-world performance on your target hardware
- Plan for a phased rollout with monitoring
For Long-Term Projects:
- Build flexibility into your codec pipeline now
- Assume hybrid codec strategies will become the norm
- Watch hardware acceleration deployment, not just codec specs
The Bottom Line: 2026 Isn’t AV1’s Year (Yet)
AV1 is a better codec than VP8 and H.264 by almost every technical measure. The compression is superior, the features are more advanced, and the licensing is cleaner.
But technology adoption isn’t just about technical superiority — it’s about ecosystem maturity, hardware reality, and the stubborn fact that “good enough” solutions stick around longer than we expect.
VP8 has been “good enough” for a decade. It works on every device, in every browser, without burning your CPU or draining your battery. For most real-time applications in 2026, that’s still the winning formula.
AV1 will get there. The hardware is catching up, the tooling is maturing, and the industry momentum is real. But we’re not there yet, and rushing to adopt before the ecosystem is ready is a risk most teams can’t afford.
The smartest teams aren’t choosing sides. They’re building flexible architectures that let them use VP8 for broad compatibility now while positioning to adopt AV1 selectively as hardware support expands.
That’s not as exciting as “we’ve switched to the next-gen codec,” but it’s what actually works in 2026.
At Trembit, we’ve built real-time video systems across multiple industries, and we understand the gap between codec specs and production reality. Whether you’re evaluating AV1, optimizing your current WebRTC stack, or planning for future codec transitions, we can help you make the right technical decisions based on your specific hardware constraints, user base, and business requirements — not just what looks good in a press release.
