iQiyi says AI will slash film costs and break TV's centralized gatekeepers — but can China get the compute?

iQiyi (爱奇艺) founder and CEO Gong Yu told industry audiences that artificial intelligence could be the single biggest cost disruptor in Chinese film and television. Speaking at the 16th Beijing International Film Festival industry forum, Gong warned that traditional production costs have ballooned — actors and crew have pushed budgets up so far that losses are now more common than profits — and argued AI could reverse that trend by driving production costs down by “several times.” It has been reported that a short shown at the festival would have cost five to ten times more if produced by conventional live-action methods.

From centralized commissioning to decentralised creation

Gong has repeatedly argued — most recently at the 13th China Online Audio‑Visual Conference, it has been reported — that long‑form video platforms in China have long operated a centralized model in which a small group of decision‑makers control what gets made. AI, he said, offers a path to “decentralization”: lower unit costs mean more creators can experiment, more niche formats (short series, animation, micro‑dramas) can find audiences, and platforms could shift from gatekeepers to enablers. He also reportedly described iQiyi as “begging” for high‑quality AI‑generated content and said the company urgently needs hybrid talent who understand both AI technology and artistic creation.

The catch: distribution and compute constraints

Gong acknowledged big caveats. Even if production costs fall, offline marketing and distribution remain expensive and unpredictable. And there is a geopolitical dimension few in China’s tech and media sectors can ignore: high‑end AI model training requires advanced chips and cloud compute. U.S. export controls and broader trade tensions have constrained access to some top‑tier semiconductors, forcing Chinese firms to accelerate domestic hardware and software innovation. Will AI‑driven production scale in time to rescue a cash‑strapped industry? That depends not just on algorithms, but on who controls the compute and the money to deploy it.