LibTV Will it be the next Cursor?

Last night I saw that LaterPoint published a major article on YanYu Technology, with 8 months and two funding rounds raising 500 million USD, Chen Mian coming from JianYing to continuously achieve product-market fit for three AI products. LaterPoint calls it: the "big player" in domestic AI applications.
I'm a bit curious, so I casually asked my lobsters for some insights, and it turns out the main contributor is LibTV — an AI video creation platform that only launched in March. According to the information my lobsters found, LibTV's first-month daily revenue exceeded one million USD, serving nearly a thousand teams. The company's overall ARR surpasses 300 million USD, with May revenue up 3000% year-over-year.
The revenue explosion is due to catching the wave of AI short dramas/animations. In Q1 this year, the entire industry launched 128k micro-short dramas, with AI accounting for 122k of 👉 them. LibTV just provides video production capabilities 👈 for these short drama teams, capturing the real budgets from professional content producers on the B side, not small change from C-side subscriptions.
This script feels a bit familiar, looking like how Anthropic used Claude Code to overtake ChatGPT... no, more like the script from Cursor back in the day.
If LibTV is the Cursor of the video track, then LibTV is very likely to face the same subsequent difficulties as Cursor.
We all know that Seedance2.0 is currently leading in the video generation track, with over 90% usability, but Veo and Alibaba’s HappyHorse are not yet at production level. It’s easy to infer that LibTV’s core video generation capability mainly relies on Seedance 2.0.
And ByteDance itself has JianYing — since February this year, it has deeply integrated Seedance 2.0. JianYing + Seedance is ByteDance’s own end-to-end solution, and LibTV is likely to become an external player that helps verify market demand, but could be pulled back at any time once validated.
Under the same script, LibTV is more passive than Cursor:
1️⃣ Cursor still has options after being betrayed by Claude — besides Claude, there’s GPT, Gemini, and the IDE ecosystem and user habits are real barriers. For LibTV, it nominally connects with multiple models, but only Seedance2.0 can be used for production; models can’t be swapped (I sincerely hope Alibaba’s HappyHorse can put in effort — competition is good, referencing CC and Codex).
2️⃣ Looking at the harness layer — JianYing itself is a mature video editor used by hundreds of millions of users, and ByteDance can easily replicate the workflow orchestration barrier. Cursor at least has its own IDE ecosystem as a moat, but LibTV doesn’t even have that thick a layer.
3️⃣ The parent company behind LibTV, Evoken, is a typical pure application-layer company, likely without self-developed video generation models. Meanwhile, Cursor is already working on self-developed models — even if it can’t catch up with Claude in the short term, at least there’s a path to gradually reduce dependency.
No choice of models, easy to chase the harness, no self-developed capability, and vendors are direct competitors. Chen Mian himself came from JianYing, so he should be well aware of this situation.
LaterPoint’s articles are always very insightful, and I guess they will definitely ask this question — out of the $300 million USD ARR, how much is paid to ByteDance for Seedance?
They didn’t specify, I think it’s most likely not because they don’t want to say, but because they can’t.