Newsletter Subscribe
Enter your email address below and subscribe to our newsletter
[forminator_form id="25163"]

bloombergtheedgemalaysiathenextwebByteDance founder Zhang Yiming is personally overseeing the development of an AI model designed for real-time spatial video generation, with a launch possible as soon as next month, Bloomberg reported on Monday, citing people familiar with the matter. The move positions the TikTok parent company as a direct challenger to Meta and Alphabet in the fast-emerging field of world models.bloomberg
The model, built on ByteDance's existing Seedance video generation system, would let users create interactive virtual worlds for live streams, short-form dramas and games, according to the report. Zhang has been coordinating efforts across business units and directing AI resources and computing capacity toward the project, the people said. One person cautioned that the timing is not settled and plans could change.theedgemalaysia+1
The reported specifications call for on-demand video at roughly 20 frames per second with latency of about 0.05 seconds, generated in the cloud rather than on-device. That architecture could lower the hardware requirements for ByteDance's Pico extended reality headsets, reducing cost barriers to VR adoption and shifting competition away from device specs toward AI models, cloud infrastructure and content distribution.thenextweb+1
Both Meta and Apple have invested heavily in headsets that have struggled to reach mainstream audiences. A cloud-rendering approach would not match Apple's Vision Pro on fidelity, but it would offload the computational burden from the hardware itself.thenextweb
The spatial video model fits within a broader strategic shift at ByteDance. Earlier this year, 36Kr reported that the company had designated world models as its top AI priority for 2026, commanding an eight-figure training data budget in renminbi — three to four times what rivals were spending. The stated benchmark is Google's Genie, which lets users interact with real-time rendered environments.theedgemalaysia+1
ByteDance is pursuing two parallel tracks: a vision-language-action approach aimed at robotics and embodied intelligence, and 3D simulation for entertainment and games. The spatial video model belongs to the latter, which already has a user base through platforms like CapCut and Doubao.thenextweb
The financial backing is substantial. Bloomberg reported last week that ByteDance secured a $30 billion loan and has been weighing capital expenditure of as much as $70 billion on its AI build-out. Seedance itself has become a widely used tool among independent film studios, creators and AI startups, underpinning several of ByteDance's consumer products.theedgemalaysia+1
Zhang's direct involvement places him alongside researchers such as Fei-Fei Li and Yann LeCun, who have argued that models grounded in visual and physical understanding — rather than language alone — represent the path toward AI systems capable of acting in the world.bloomberg+1