ByteDance readies real-time world model under Zhang Yiming

ByteDance founder Zhang Yiming is personally directing the development of an AI model capable of generating real-time spatial video, with a launch possible as soon as next month, Bloomberg reported Monday, citing people familiar with the matter.bloomberg+1

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. The effort places the TikTok parent company in direct competition with Meta ↘ 2.43% and Alphabet ↗ 0.21% in the emerging field of world models — AI systems that simulate physical environments in response to user input.theedgemalaysia+1

Cloud-Rendered Worlds for Pico Headsets

Zhang has been coordinating efforts across ByteDance's business units and deploying AI resources and computing capacity behind the project, according to Bloomberg's sources, who cautioned that the timing is not settled and plans could change. The model targets users of Pico, ByteDance's extended reality hardware arm, generating virtual worlds that respond to headset users' voices and movements.bloomberg+1

The reported specifications call for on-demand video at roughly 20 frames per second with latency of about 0.05 seconds, rendered in the cloud rather than on the device. That approach would reduce the processing power required in headsets, opening a path for cheaper hardware and shifting competition away from device specifications toward AI models, cloud infrastructure, and content distribution.theedgemalaysia+1

A Strategic Bet on Video-Centric AI

The project reflects ByteDance's broader strategic pivot. Earlier this year, 36Kr reported that ByteDance 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, according to the outlet's sources. Internal testing early in 2026 placed ByteDance about 10 percent behind the global state of the art, with Google's Genie serving as the benchmark.cryptobriefing+1

Bloomberg framed Zhang's involvement as placing him alongside researchers such as Fei-Fei Li and Yann LeCun, who have argued that models grounded in visual and physical understanding are the path to systems that can act in the world. ByteDance's Seed research team has built toward this moment with a series of releases, including Seedance 2.5 in July, which introduced 30-second single-pass audio-video clips and enhanced spatial controls.thenextweb+2

Implications for Meta and Apple

For Meta and Apple ↘ 0.26%, which have invested heavily in virtual and mixed-reality headsets that have struggled to reach mainstream adoption, ByteDance's cloud-first approach poses an indirect challenge. Rather than competing on hardware fidelity, the model seeks to make high-quality spatial content accessible through lower-cost devices fed from data centers.theedgemalaysia+1

ByteDance secured a $30 billion loan last week and has been weighing capital expenditure of as much as $70 billion on its AI build-out, Bloomberg reported. Seedance already underpins CapCut and Doubao, China's most popular AI chatbot, and has become a tool of choice for independent film studios and creators.thenextweb+1