Zhang Yiming tells ByteDance AI staff catching up with OpenAI and Anthropic will take until 2029 or 2030

ByteDance co-founder Zhang Yiming presented a new AI plan at a company-wide meeting for the AI team in recent days, according to people familiar with the meeting cited by Bloomberg. The report, relayed by IT Home on November 11, 2026, says the company plans to launch an AI assistant similar to Meta's Muse within weeks. The exact date of the meeting was not specified in the material reviewed.

The central claim is about timing. Zhang reportedly told staff that ByteDance can compete with frontier labs such as Anthropic and OpenAI, but that matching them will take until 2029 or 2030 at the earliest. That is a internal target as relayed by sources, not a public commitment.

He also pointed to data as a constraint. According to the same sources, the pretraining dataset for ByteDance's Seed large language models is several orders of magnitude smaller than those used by leading US rivals. The reporting does not include the underlying figures, so that comparison rests on the sources' account. Zhang also reportedly argued that ByteDance should not rely too heavily on model distillation.

The trade-off is visible in the reporting. ByteDance has a clear lead in video generation through Seedance, which the report ties to its large video data holdings and commercial traction. Its language models, by contrast, have not reached the top of major global leaderboards, and its weights are not open. The report says the company is weighing up to $70 billion in investment, a figure that should be read as a reported plan rather than a confirmed budget. Whether that spending goes to a consumer assistant, data acquisition or the Seed language models is not established by the reporting.

The assistant's launch timing and scope are reported by Bloomberg and have not been confirmed by ByteDance in the material reviewed. The reporting also does not say whether the assistant will get dedicated hardware, as Meta plans for Muse. The sharpest open question is whether a 2029-2030 horizon is a realistic goal for the Seed team, or a benchmark set to motivate it.