Will Chinese AI labs produce frontier models that surpass leading US models in the near term?

Alexandr Wang · Jun 12, 2024

Chinese LLM and AI capabilities are now roughly neck and neck with US capabilities, having meaningfully caught up in just the past few months

Zero One AI's Yi-Large is one of the best models in the world, sitting just behind GPT-4o, Gemini and Claude 3 Opus on the leaderboards, whereas two years ago China was nowhere near GPT-4

34:14 20VC: Scale's Alex Wang on Why Data Not Compute is the Bottleneck to Foundation Model Performance, Why AI is the Greatest Military Asset Ever, Is China Really Two Years Behind the US in AI and Why the CCPs Industrial Approach is Better than Anyone Else's

Anton Osika · Aug 18, 2025 · hedged

There is roughly a 50/50 chance that China will have the best foundation model, and the leading model has not been created yet

Scope: stated as a probability, not a certainty

0:26 20VC: Lovable CEO Anton Osika on $120M in ARR in 7 Months | The Honest Truth About Defensibility and Unit Economics for AI Startups | The State of Foundation Models: Long Grok, Short OpenAI, Why | Replit vs Lovable vs Bolt: What Happens

Anton Osika · Aug 18, 2025 · hedged

The leading model has not been created yet and will come from China — roughly a 50/50 chance the best model in use at some point is Chinese, which he finds somewhat concerning

Chinese companies aren't as good at understanding users, so they're not a near-term product threat, but on raw model quality they may lead

Scope: 50/50 probability; not worried about Chinese companies on user understanding

58:15 20VC: Lovable CEO Anton Osika on $120M in ARR in 7 Months | The Honest Truth About Defensibility and Unit Economics for AI Startups | The State of Foundation Models: Long Grok, Short OpenAI, Why | Replit vs Lovable vs Bolt: What Happens

David Cahn · Aug 5, 2024 · hedged

China will catch up in AI but America has a good shot at staying ahead, and no scenario where America falls completely behind is likely

America has strong immigration, a capitalist economy, freedom, a good business environment and Silicon Valley culture, and the last two years of evidence show America performing well

Scope: don't underestimate competitors; China will also do well

40:30 20VC: Sequoia's David Cahn on AI's $600BN Question | Why the Data Centre is the Most Important Asset | Servers, Steel and Power: The Core Pillars Powering the Future of AI

Nick Frosst · Sep 1, 2025 · hedged

China will not produce models that beat leading US models over the next two years

They haven't done it yet — they have made definitely good models, but not models that beat the frontier

Scope: "I'm not sure"

59:35 20VC: Cohere Founder on How Cohere Compete with OpenAI and Anthropic $BNs | Why Counties Should Fund Their Own Models & the Need for Model Sovereignty | How Sam Altman Has Done a Disservice to AI with Nick Frosst

Harry Stebbings · Jun 12, 2024

The claim that China is two years behind the US in AI is wrong, and any gap will close quickly

The Chinese government will be willing to do far more on data access, data provisions and regulation than Western governments

33:50 The two year gap narrative is wrong china is closing fast