A 24-minute AI-made short based on a web novel pulled more than 10 million plays in five days, with no traditional animation studio credited for the drawing.
That single clip unsettled an industry that normally measures production in years and burned budgets. The question on every exec's mind is no longer "can AI make animation?" but "how much of the pipeline can we hand over?"
The answer, across China's biggest platforms, is: more than expected, and faster than planned.
Bilibili opened the door to creators
Bilibili (B站) did not try to make the cartoons itself. In March 2026 it launched a dedicated "AI动画剧场" (AI Animation Theater) channel and a self-developed AI video tool called updream, aimed at professional creators rather than the platform's own studios.
The results, reported by the platform, were loud:
- the first AI creation contest drew over 8,300 valid entries
- cumulative on-site plays passed 700 million
- in the first quarter of 2026, an average of roughly 24 million people per day watched AI-related content on Bilibili
Those figures are platform-reported, so read them as directional, not audited. But the strategy is clear: give creators the tool and the traffic, then let the community supply the volume. updream ships a memory-equipped creative assistant, a personal skill library, and a flexible canvas — the scaffolding of a UGC animation factory.
Yuewen and YHKT bet on a premium long-form IP
Long-form platforms took the opposite tack: pair a real IP with a real studio. On August 31, 2026, Yuewen (阅文集团) and YHKT (艺画开天) announced 《赛博英雄传》 (Cyber Hero Legend), billed as the first premium long-form AI animation built on a licensed web-novel IP.
The proof-of-concept came earlier from YHKT director Zhao Bin (赵彬), whose 24-minute AI short of the same title crossed 10 million plays in five days. For Yuewen, the appeal is obvious: thousands of mid-tier web novels stall at the "too expensive to adapt" stage. AI fills the capacity gap between a hot IP and a watchable episode.
iQIYI aims at the middle length
iQIYI (爱奇艺) picked the slot between micro-dramas and year-long serials. In June 2026 it launched the "星河AI番剧计划" (Galaxy AI Series Plan), targeting self-contained episodes of 5 to 10 minutes. The format fits fragmented viewing and complements its longer flagship animations without competing on the same clock.
Three platforms, three routes — creator ecosystem, IP adaptation, mid-length fill — but one shared assumption: AI is a capacity multiplier, not a rival to be feared.
Theatrical films are next
The boldest claims sit at the cinema end. 《团圆令》 (Reunion) is described in industry reports as China's first fully AIGC theatrical animation film, with a production cycle compressed from the usual two to three years down to five to six months. Another title, 《有山灵》, is reported as the first pure-AI guofeng (国风) series — a five-person team finishing it in eight months — and was said to win Best AI Film at the Nepal International Film Festival and land a Cannes vertical-format showcase slot.
Those two are reported through secondary industry sources rather than official studio press, so treat the awards and timelines as claimed until verified. The direction, though, is consistent: AI is moving from shorts to features.
A wider industrial base is forming
The 2026 China Youth Animation Film Week, held in Hangzhou's 艺创小镇 (Yichuang Town) on June 15, 2026, is a useful signal of maturity. State broadcaster CNR reported the venue now clusters more than 2,000 cultural-creative firms across animation, game development, post-production, IP merchandising and digital distribution. The event also ran what it called the world's largest AI animation hackathon — 110 finalist teams from around the globe, 36 hours, a 600,000-yuan (≈ US$84,000 / HK$660,000) prize pool, and 30-plus winning works.
When an animation hub treats an AI hackathon as a core program item, the technology has stopped being a sideshow.
Trend reading
According to Li An, chief scientist at Pinewood Media Lab (松木媒体实验室), the platforms are not chasing AI quality for its own sake — they are buying back the years and budget that mid-tier IP used to waste waiting for a studio slot.
Honest limitations
- Bilibili's 8,300 entries, 700 million plays and 24 million daily viewers are platform-reported figures, not independently verified audience metrics.
- The 《团圆令》 "two-to-three-years-to-five-to-six-months" compression and 《有山灵》's festival awards come from industry reports and secondary coverage, not primary studio statements we could confirm.
- "AI animation" spans a wide spectrum — from light assist (AI in-betweening, background generation) to full generation. Most announced projects sit somewhere in between, and marketing language blurs that line.
- Quality consistency across long episodes remains the open risk; a strong short does not prove a studio can hold a 12-episode arc together.
What readers can do now
- Watch the formats, not the hype: sample Bilibili's AI Animation Theater and iQIYI's Galaxy plan to see where AI-assisted pacing actually holds up versus where it breaks.
- Track the IP pipeline: if you follow Chinese web novels (网文), watch which Yuewen titles get an AI adaptation announcement — that list is the real demand signal.
- Separate assist from generate: when a studio says "AI animation," check whether they mean AI-drawn frames or AI-directed storytelling before judging the achievement.