What Happened
ByteDance launched Seedance 2.0 in early February 2026 as a direct competitor to OpenAI's Sora 2, Google's Veo 3.1, and Kuaishou's Kling 3.0. The model generates cinema-quality video with synchronized audio from text, image, or video inputs. It's available primarily through ByteDance's Jimeng AI app in mainland China, priced at roughly one yuan ($0.14) per second of generated video.
The model went viral almost immediately - and that's where the problems started.
On the compute side, demand overwhelmed ByteDance's infrastructure. Free-tier users on platforms like Dreamina face queue positions numbering in the thousands, with wait times exceeding two hours for a single generation during peak hours (8 PM to midnight GMT+8). Paid subscribers wait 5-15 minutes. During peak hours, the free tier is essentially nonfunctional.
On the copyright side, Hollywood came down hard. Users generated videos featuring Spider-Man, Darth Vader, Baby Yoda, and convincing likenesses of actors like Tom Cruise and Brad Pitt. The Motion Picture Association's CEO Charles Rivkin accused Seedance 2.0 of "engaging in unauthorized use of U.S. copyrighted works on a massive scale." Disney and Paramount Skydance sent cease-and-desist letters. SAG-AFTRA condemned the "unauthorized use of our members' voices and likenesses."
ByteDance responded by pledging to strengthen safeguards and add content filters. As of early March, those fixes are still in progress.
Why It Matters
If you use AI video tools for content creation, Seedance 2.0 is a preview of what happens when a powerful model launches without adequate infrastructure or content filtering.
The compute bottleneck is the more immediate problem for practitioners. A video tool with multi-hour queue times isn't viable for production work. This is the same pattern we've seen with every viral AI launch - demand spikes far beyond capacity, and the free tier becomes unusable. If you need reliability, you're paying, and even paid tiers get congested.
The copyright issue is more structural. Every major video model will face this pressure. The difference is whether studios target the model maker, the platform, or individual users. ByteDance is getting the Napster treatment - being held responsible for what users create with its tool.
Our Take
Seedance 2.0 is genuinely impressive technically. The multi-modal audio-video generation puts it in the top tier alongside Sora 2 and Veo 3.1. But launching a model this powerful without compute headroom or meaningful content filtering was predictable chaos.
For anyone evaluating AI video tools right now: don't chase the viral demos. Chase the tool you can actually use reliably when you need it. D-ID, Fliki, and Descript aren't generating 60-second cinematic clips, but they deliver results when you hit generate. That matters more for real workflows than theoretical capability behind a two-hour queue.
The copyright fight will define AI video for the next two years. ByteDance is the current target because Seedance made infringement so easy and visible. But this pressure will reach every video model eventually. The companies that build strong content filtering now - not after the cease-and-desist letters arrive - will have an advantage.
The $0.14-per-second pricing is notable. If ByteDance can scale the infrastructure, that's significantly cheaper than Western alternatives. But "if" is doing heavy lifting in that sentence.