TikTok's CEO sees AI cracking physics and biotech within three years: "very, very, very possible"

TikTok CEO Shou Zi Chew told DBS Bank’s Asia Future Forward conference in Singapore on Monday that he expects artificial intelligence to deliver "real scientific breakthroughs" — in physics, biotech, chemistry and material science — within two to three years, arguing the technology’s long-term potential is underestimated.

TikTok CEO Shou Zi Chew expects artificial intelligence to deliver "real scientific breakthroughs" within two to three years — in physics, biotech, chemistry and material science — and says the technology's long-term potential is being underestimated. He made the remarks on Monday at DBS Bank's Asia Future Forward conference in Singapore, in a session on unlocking human potential in the AI era. (CNBC, Forbes Europe)

"In the next two to three years, we're going to have a real scientific breakthrough: physics, biotech, chemistry, material science, because of AI. It's very, very, very possible," Chew said. He added that rapid advances in AI were opening up possibilities far beyond the technology's current applications.

ByteDance's own bet on science

The prediction has some grounding in Chew's own house. ByteDance, TikTok's parent company, has been expanding its AI research beyond consumer applications: in April, its Seed research team unveiled Protenix-v2, a model for biomolecular structure prediction and design with potential applications in drug discovery.

"Silicon is just superior to carbon"

In a dialogue with DBS group head of consumer banking and wealth management Shee Tse Koon, Chew left the door open to a more radical reading of the trajectory: there is a "non-zero chance" that AI surpasses humans in certain aspects, he said. "You can see a world where silicon is just superior to carbon; you can see a world where it will last longer; you know it's more scalable."

Alignment is the open question

But where others frame the debate around timelines, Chew put the emphasis on control: "If that happens, then you know the question would be: how do you align the models? How do you align the machines to be doing things that are obvious to us as human beings?"

He drew a line between today's machines and tomorrow's. Right now, AI takes over routine, low-stakes tasks, while frontier models can handle higher-stakes activities with appropriate guardrails — and humans will likely remain necessary for tasks that involve high stakes or require differentiated judgment. But he cautioned against underestimating the long term: "It's risky to think that just because it's slow in the shorter term, that it will not become very important in the long term — that's the wrong conclusion to take away."

Why it matters

The remarks sharpen the industry's timeline debate. Chew's two-to-three-year window sits close to Anthropic CEO Dario Amodei's 2025 Davos prediction that AI could outdo a Nobel laureate within two to three years — and cuts against more conservative framings, like DeepMind CEO Demis Hassabis's ten-to-fifteen-year horizon for an AI-driven scientific renaissance. Either way, the head of the world's most-downloaded app is staking a public position: the science phase of AI could arrive fast, and the alignment question arrives with it.