TikTok CEO Shou Zi Chew has joined the growing list of tech executives putting a clock on artificial intelligence. According to CNBC, Chew said AI will drive major scientific breakthroughs within the next three years.
He made the remarks on August 1, 2026, in a speech to PhD graduates at Nanyang Technological University in Singapore.
What Chew told the graduates
Chew described the present moment as an era of innovation without precedent. His central argument was that AI is pushing technology and society through change that compounds on itself rather than arriving in steady increments.
Notably, Chew did not spell out which discoveries he expects or which scientific fields will move first. His forecast was about the pace and scale of change, not a list of specific milestones with dates attached.
Chew has run TikTok since 2021. Before that, he held roles at smartphone maker Xiaomi and at Goldman Sachs, giving him a background that spans consumer hardware, finance and now one of the world’s most closely watched social platforms.
Advertisement
A crowded field of three-year forecasts
Chew is far from alone in setting an aggressive timeline. OpenAI CEO Sam Altman said in 2025 that he expected a major scientific discovery within two years.
Anthropic CEO Dario Amodei has made a similar case from a different angle. He has said AI systems could outperform humans at most tasks within a two-to-three-year window.
Taken together, these leaders are pointing at roughly the same horizon. Altman and Amodei both anticipate major AI-driven discoveries and capabilities arriving by 2028, which lines up neatly with Chew’s three-year framing from a 2026 stage.
ByteDance’s own AI push
Chew’s comments carry extra weight because of the company behind TikTok. ByteDance, TikTok’s parent, has been building out its AI business aggressively throughout 2026.
Its Doubao AI assistant reached 345 million monthly active users by March 2026. A large part of that growth came from integrating Doubao with Douyin, the Chinese version of TikTok, which put the assistant directly in front of an enormous existing audience.
ByteDance has also released multimodal AI models called Seed2.0 and Seed2.1. Multimodal simply means the models can work across different kinds of input and output, such as text, documents, images and video, rather than handling only one format.
The company has positioned these models for demanding work. Their stated uses include managing workflows, processing and analyzing documents, generative design and video production.
What this means
For ByteDance specifically, Doubao’s user base offers a meaningful foundation. An assistant with 345 million monthly active users opens potential paths to revenue through advertising and subscriptions, two models ByteDance already understands well from its social platforms.
The Seed2.0 and Seed2.1 releases signal ambitions that extend well beyond recommendation algorithms and short-form video.
There is also a credibility risk baked into these forecasts. Three years is short enough that the predictions will be tested within the current product cycle. If major AI-driven scientific discoveries fail to materialize by 2028 or 2029, the executives who made these calls will face harder questions about how much of the optimism was analysis and how much was marketing.
Disclosure: This article was edited by Diego Almada Lopez. For more information on how we create and review content, see our Editorial Policy.