Alibaba’s HappyHorse-1.0 ranked first across both text-to-video and image-to-video categories on Artificial Analysis as of April 7. The result signals a rapid shift in competitive positioning within AI-generated video, where benchmark performance often drives enterprise adoption and capital flows.
The model appeared anonymously before developers confirmed it originates from Alibaba’s ATH AI Innovation Unit via X. The disclosure ended days of speculation around ownership and coincided with a 2.12% rise in Alibaba’s Hong Kong-listed shares, following a 6.75% gain earlier in the week.
Dylan Butts
Can Alibaba Sustain Its Lead In AI Video Models?
Alibaba’s advance comes as rivals face operational and legal constraints in video generation. OpenAI has discontinued its Sora platform amid high compute demands, while ByteDance paused Seedance 2.0 following copyright disputes with major studios. Against that backdrop, HappyHorse-1.0’s rapid ascent stands out in a segment where iteration cycles remain costly and uncertain.
The company has tied AI development directly to long-term infrastructure investment and product integration. Chief Executive Officer Eddie Wu has prioritized AI across chips, cloud, and data centers, aligning model development with deployment across e-commerce and media platforms. Can benchmark dominance translate into sustained commercial advantage?
Alibaba is expanding its compute backbone to support that ambition. A joint data center project with China Telecom will deploy 10,000 Zhenwu AI chips, designed to operate as a unified system with roughly four microseconds latency, according to the company. The cluster is built to train and run models with hundreds of billions of parameters.
Capital deployment also reflects sustained competition in video generation. Alibaba Cloud recently led a $275 million funding round for Shengshu Technology, which develops video tools through its Vidu platform, alongside participation from Baidu Ventures and Luminous Ventures. Continued investment across both infrastructure and applications suggests the next catalyst will hinge on real-world deployment scale and enterprise uptake.
