China AI Has “Less Than 20%” Chance to Beat the US in 3 to 5 Years, Says Alibaba Scientist
Published: 12 Jan 2026
What happened?
China AI leaders shared a clear message at a big AI event in Beijing on Saturday. They said China AI is improving fast. But it may not pass the United States soon.
Who said it?
Lin Junyang, a lead engineer on Alibaba’s Qwen AI team, gave a straightforward estimate.
He said China AI has less than a 20% chance to beat top US labs like OpenAI and Google DeepMind in the next 3 to 5 years.
He even called that number “optimistic.”
Why do they think China AI may not catch up fast?
They pointed to one main issue: hardware and resources.
Lin said the US has far more computing power than China right now.
By “computing power,” he meant huge stacks of advanced chips and machines that run AI training.
He also said US labs invest huge amounts of computing into “next-gen” AI research.
At the same time, many China teams use most of their computers to handle daily user demand.
In simple words:
- The US has more powerful chips.
- The US has more machines to run them.
- US labs can spend more power on future research, not just today’s tasks.
Another expert agreed
Tang Jie, co-founder and chief AI scientist at Zhipu AI (Z.ai), supported the safe view.
He joined Lin on the same panel.
But China AI still made progress
The panel also noted something important:
Independent benchmarks show China AI models have closed the quality gap in recent years.
They also highlighted a key difference:
- Many US models stay closed (people can’t freely use or study them).
- Many China AI teams open-source their models, which helps more developers adopt them worldwide.
A real-life example
Think of AI like building a rocket.
Good ideas matter, but you still need fuel and engines.
In AI, that “fuel” is chips + compute. If you have less fuel, you move slower.
Quick questions for you
- Do you think China ai can leap forward with a surprise breakthrough?
- Or will chip limits keep slowing it down?
what to watch next
- Chip supply: Any change in access to advanced chips?
- New data centers: More computers means faster progress.
- Open-source wins: Which China AI models gain global users?
Big research releases: Watch for major new training methods that need fewer chips.
- Be Respectful
- Stay Relevant
- Stay Positive
- True Feedback
- Encourage Discussion
- Avoid Spamming
- No Fake News
- Don't Copy-Paste
- No Personal Attacks
- Be Respectful
- Stay Relevant
- Stay Positive
- True Feedback
- Encourage Discussion
- Avoid Spamming
- No Fake News
- Don't Copy-Paste
- No Personal Attacks
