Full Breakdown
Chinese AI Progress and the Role of Distillation
By Drooid · · How we work
Core Event: Accelerating Chinese AI Amid U.S. Accusations
Chinese AI laboratories such as Alibaba, Moonshot and DeepSeek have released models that, on several benchmarks, outperform leading U.S. systems. U.S. AI firms and the Cybersecurity and Infrastructure Security Agency (CISA) allege that these advances rely heavily on “industrial-scale” knowledge-distillation of proprietary U.S. models. Chinese officials and some industry leaders contest the extent of that reliance, arguing that homegrown innovation—spurred by chip restrictions—drives the progress.
Background: Innovation Under Chip Constraints
China’s tech sector has expanded from electric vehicles to robotics despite limited access to cutting-edge semiconductors from companies like Nvidia. This environment, they argue, has cultivated independent breakthroughs rather than mere imitation.
Official Statements & Responses
- models as a core development strategy.
- China’s Ministry of Commerce called the U.S. allegations “groundless and legally unsound.”
- Anthropic, creator of Claude, reported that Alibaba, Moonshot and DeepSeek attempted “illicit distillation” to train their models.
- Aidan Gomez, CEO of Cohere and co-author of the 2017 “Attention Is All You Need” paper, described Chinese models as “world class” and said the U.S. lead is “evaporating very quickly,” while acknowledging some distillation occurs but is not the sole driver of progress.
- Sriram Krishnan, former senior White House AI advisor, emphasized that distillation has long been a fundamental part of computer-science development.
Data & Statistics
- Anthropic’s report this month identified specific Chinese labs engaged in alleged illicit distillation.
- Nvidia plans to double its chip sales next year, reflecting heightened demand for AI infrastructure.
- A new report projects a shortage of up to 157,000 semiconductor workers in the United States by 2030.
- Chinese AI startup behind the Kimi model reports adoption by major financial institutions.
Verbatim Quotes
- “There was like a lot of talk about the Chinese are just copying, they're just distilling, they're cheating. A lot of these tropes were applied, and that was definitely true to an extent,” — Claude. Gomez, the CEO of Cohere
- “So the idea of distilling has always been a core part of how computer science works,” — Sriram Krishnan, a former senior White House policy advisor on artificial intelligence
- “Because of chip restrictions, Chinese developers were forced to build leaner, smarter architectures and do more with less compute,” — Neil Shah, a partner at Counterpoint Research
- “Dismissing that as mere imitation might make for convenient restrictive policy making, but it fundamentally misjudges the competition,” — Neil Shah, a partner at Counterpoint Research