Wan 2.7 של Alibaba מוביל את גל וידאו-ה-AI הסיני — בזמן ש-OpenAI נסוגה

A Forbes deep-dive argues that Chinese labs are winning the video-AI race that OpenAI effectively stepped back from, with Alibaba's open-weight Wan 2.7 leading the charge. By publishing model weights, Alibaba runs an open-source flanking strategy analogous to Meta's Llama play — using openness to seed adoption, build developer goodwill, and set de facto standards while commercial rivals keep their best video models closed.

The strategic logic is that video generation is compute- and data-intensive but commercially uncertain, so an open-weights approach lets Alibaba capture mindshare and ecosystem lock-in without betting everything on a hosted product. It fits the week's broader China-AI narrative: Huawei post-training DeepSeek-V4 on domestic chips, a Hong Kong DeepSeek-based model for local hardware, and Qwen3.7-Plus's agent push.

Separately, CNBC reports China may move toward a US-style talent-poaching path as firms compete aggressively for AI researchers and ramp basic-research investment — a sign the Chinese AI sector is maturing from fast-following toward original research leadership.

Competitively, the contrast with OpenAI's retreat from consumer video is striking and reframes who 'leads' in generative video. Caveats: 'winning' is partly a function of OpenAI's deprioritization rather than pure technical superiority, and open-weight video models raise the same misuse concerns (deepfakes, watermark stripping) surfacing elsewhere. What to watch: Wan 2.7 adoption outside China and whether Western labs re-enter the video race.