ByteDance’s AI Ambitions Are Being Hampered by Compute Restraints and Copyright Concerns
A 15-second video costs ~$2, but users face 90k-person queues and 4+ hour waits for generation.
ByteDance has launched Seedance 2.0, a major upgrade to its AI video generation model that has sent shockwaves through China's tech scene. Praised by industry figures like Game Science founder Feng Ji and video producer Pan Tianhong for its 'director-like' capabilities, the model represents a significant leap in quality. However, its rollout is severely constrained, available only through ByteDance's domestic Chinese apps (like the chatbot Doubao) with no international access, leading to a black market for accounts. The company has disclosed a proposed API price of slightly over $2 for a 15-second video, signaling future broader availability, but has not yet opened access to third-party developers.
Despite the hype, Seedance 2.0 faces two critical roadblocks. First, a severe compute bottleneck has resulted in massive queues; one test showed a waitlist of 90,985 users with a 4-hour estimated wait for a 5-second clip, even for paying subscribers on plans costing up to $70/month. Second, ByteDance is confronting legal challenges, having received cease-and-desist letters from major Hollywood studios including Disney, Netflix, and Paramount, which allege the model's outputs infringe on copyrighted works. Analysts like Afra Wang note this highlights the diverging US-China AI paths, where China leads in video generation but lags in tools like coding assistants. The model's future scalability and global impact now hinge on resolving these compute and copyright crises.
- Seedance 2.0 generates 15-second videos for ~$2 via API, but access is currently restricted to ByteDance's Chinese domestic apps like Doubao.
- Compute bottlenecks cause extreme delays, with users reporting queues over 90,000 people long and wait times exceeding 4 hours for a short clip.
- Major studios Disney, Netflix, and Paramount have sent cease-and-desist letters, alleging the model's outputs infringe on their copyrighted works.
Why It Matters
Highlights China's lead in generative video AI but exposes critical scaling and legal vulnerabilities that could limit global adoption.