OpenAI’s retreat from video opens a runway for Chinese firms, says Zhou Yahui

Big bet, bigger machines

At a Zhongguancun forum session, Kunlun Wanwei (昆仑万维) rolled out three large generative models — Matrix-Game 3.0 for games, SkyReels V4 for video and Mureka V9 for music — underscoring a Chinese push into AIGC verticals. Zhou Yahui, chairman and CEO of Tiangong AI (天工AI) and a well‑known early‑stage investor turned operator, argued that OpenAI’s decision to shutter Sora’s standalone entry has not signalled the death of video generation but rather created an opening for Chinese companies. He told Phoenix Tech reporters that video models can be profitable and that OpenAI’s withdrawal reflected internal choices — and, he added, it has been reported that much of Sora’s video team was poached by Meta — not a structural market failure.

Spending to lead, and a different pricing logic

Zhou laid out a blunt arithmetic: to stay in the global first tier you need sustained heavy R&D — “at least RMB 100 million a month,” he said — and should expect multi‑year operating deficits before platform revenue catches up. He is explicit about the business case: enterprise customers will pay, margins matter, and model pricing could rise in relative terms as services crystallise — he predicts relative price increases of 10–50x over the coming decade even if absolute inference costs fall. The thesis is straightforward: invest now in models and IP, then monetise via APIs, platform revenues and high‑margin vertical applications for businesses.

Talent, chips and geopolitics

This competition plays out against a geopolitical backdrop: talent mobility, US export controls on high‑end chips and broader trade frictions have shifted the battleground from pure compute arbitrage to domestic stack resilience and differentiated product roadmaps. Zhou openly endorsed a strategy of following ByteDance (字节跳动) into international markets — “they eat imported beef, we eat domestic beef,” he quipped — while warning that long‑term leadership requires deep, persistent technical investment. The question now is whether China’s cohort of AIGC specialists can convert a temporary opening into durable global positions — Zhou is betting they will.