China's prime-time TV slot now belongs to an all-AI drama
For years, AI-generated video lived on phone screens — short dramas, feeds, ads. This month it moved onto the biggest stage Chinese television has, and the numbers behind it explain why broadcasters are moving fast.
Mango TV's Hou Xiyou Ji ("Journey to the West: The Sequel") became China's first AIGC long-form drama to air in satellite-TV prime time on August 31, and it held its own. The 30-episode series, with 40-minute episodes and not a single human actor, drew a 0.27% average viewership rating and a 0.33% peak — good enough for a top-three prime-time slot among satellite channels on debut day, per Huan.TV data. The production math is the real headline: roughly RMB 900,000 (about $126,000) per episode, around RMB 20,000 a minute — less than a tenth of a comparable traditional animated series — with a crew of about 100, more than half of them "AIGC directors" prompting models instead of pointing cameras. Total director Li Dongshen's framing: "we used 100 people to do the work of 2,000."
Two things made this possible, and neither is the model itself. First, policy: the broadcast regulator's August 2025 "21 measures" scrapped the 40-episode ceiling and piloted review-while-airing, compressing approval from months to days — a rule change that happens to suit AI productions, where revising an episode costs a re-render, not reshoots. Hou Xiyou Ji is the first series to run through that pipeline end to end. Second, tooling: ByteDance's Seedance 2.0 launched in February, followed quickly by Kuaishou's Kling 3.0 and Shengshu's Vidu Q3, and this generation's multi-reference character consistency and synced audio-visual output pushed AI drama quality past broadcast standards. Broader cost data backs the shift: a provincial satellite-TV executive told Beijing Youth Daily that per-minute production costs fell from RMB 1,500–2,500 to RMB 400–1,000 after Seedance 2.0, making purchased AI dramas roughly 20% or less of the price of live-action short dramas.
The caveats are real. Audience reaction split between "I couldn't tell it was AI" and "soulless acting"; early outsourced AI dramas got mocked on big screens for stiff faces and the tell-tale "high-frequency AI face," flaws that phone viewing forgives and 65-inch sets don't. Selection is still conservative — every AI drama that has aired leans on pre-validated IP (mythology, heritage stories, award-winning sci-fi), not the original romance and urban genres that dominate short-drama spending. And the labor picture is strained the way every gold rush is: DataEye counted over 19,000 AI-drama job listings across three hiring platforms on a single August day, with under 4% asking for 5+ years of experience, and studios reporting they fill only about 30% of demand. Mango's own answer is institutional — a standalone AIGC content center created in February, an AI industrialization center in May, and its Lingchuang platform aggregating 40-plus multimodal models with built-in broadcast-grade review. The market echoed it: Mango Excellent Media hit its 20% daily limit on premiere day before media stocks pulled the whole sector up.
Watch the next window: the remaining 25 episodes land over the 2027 winter break, and Mango has said an AI "pseudo-live-action" long drama benchmarked against S-tier human productions arrives between this quarter and Q1 2027. That is the first true head-to-head — same slot, same length, AI against the industry's best — and Chinese regulators already require prominent AI-generated labels on this content, so viewers will know exactly what they're comparing.
What to watch: whether review-while-airing plus AI economics spreads beyond Mango-affiliated channels — Guangdong TV's six-person in-house team already shipped a fully AI-post-produced micro drama that topped its local time slot, the cheaper copyable path.
Is an AI drama holding prime time a production revolution or a discount trick broadcasters will outgrow? Tell us in the comments.
