Former Google CEO Eric Schmidt says China’s leading artificial intelligence models are now approximately six months behind those developed in the United States. His assessment reflects the rapid improvement of Chinese AI systems and the narrowing gap between the two countries’ leading models. Speaking on the In Good Company podcast, hosted by Nicolai Tangen, Schmidt warned that the United States could lose its global advantage if Chinese companies continue distributing capable AI models at little or no cost.
Schmidt’s argument extends beyond technical performance. While US companies have invested heavily in frontier models and computing infrastructure, Chinese developers have increasingly competed through lower prices, open-weight models and broad distribution. His comments suggest that the global AI race will depend not only on which country builds the most capable systems, but also on which companies make their technology easiest to adopt. The six-month estimate is Schmidt’s assessment rather than a definitive measurement of every AI capability.
Key Takeaways
- Six-month gap: Eric Schmidt estimates that China’s leading AI models trail the US frontier by roughly six months.
- Changing assessment: Schmidt previously estimated the gap at one to two years, according to comments reported earlier in 2026.
- Pricing advantage: He argues that free or cheaper Chinese AI products could attract users in developing economies.
- Open-weight distribution: Models that can be downloaded and deployed by developers may help Chinese AI companies expand internationally.
- Technical progress: Chinese developers have continued improving models despite restrictions on access to some advanced semiconductor technology.
- Competition beyond benchmarks: Pricing, computing costs, availability and real-world adoption will influence the industry’s long-term direction.
Why Eric Schmidt Says China Is Six Months Behind
Schmidt shared his assessment during an episode of In Good Company, the podcast hosted by Nicolai Tangen, CEO of Norges Bank Investment Management. His remarks, reported by Benzinga on October 22, 2026, described a narrowing gap between the leading AI systems developed in China and the United States.
The estimate is notable because frontier AI capabilities change quickly. Companies regularly release updated models, and competitors can sometimes reproduce or approach particular capabilities within months.
Schmidt’s assessment also reflects the progress of Chinese developers such as DeepSeek, Alibaba and other AI companies working on language models, coding systems and reasoning capabilities.
However, there is no single measure that captures every aspect of AI performance. A model may be competitive in mathematics or programming while trailing in other areas, such as complex reasoning, reliability or particular enterprise tasks.
The six-month figure should therefore be understood as a broad estimate of the gap between leading models, not a guarantee that every Chinese model is six months behind every US model.
Why Schmidt Thinks Free AI Models Could Win More Users
One of Schmidt’s central arguments concerns the economics of AI adoption. In his view, the United States could remain ahead in developing frontier models but still lose ground in the global market if Chinese providers distribute their systems more widely.
Chinese AI services can be free or comparatively inexpensive, lowering the barrier for developers, small businesses and users who cannot afford premium subscriptions or expensive API access.
For companies in emerging markets, the cost of running AI can determine whether a product is commercially viable. A capable model with lower operating costs may make it possible to build customer-service tools, educational applications, coding assistants and business automation products on tighter budgets.
This is the distinction Schmidt emphasized: technical leadership does not automatically translate into market leadership.
The argument is not that price alone determines success. Reliability, language support, privacy, infrastructure, documentation and the availability of developer tools also influence which models businesses choose. But if several models perform adequately for a task, the less expensive option may be attractive.
How Open-Weight AI Models Strengthen China’s Position
Open-weight AI models are another important part of the competition. Their trained parameters can be downloaded and run by developers, subject to the terms of the relevant licence.
This differs from a closed model that can be accessed only through a provider’s website or API. With open-weight systems, developers may be able to run models on their own infrastructure, adapt them for specific tasks and avoid paying for every request to an external service.
Chinese developers have released models that have attracted international attention for their capabilities and accessibility. Such distribution can help a model spread through developer communities, startups and organizations that want greater control over deployment.
| Competitive factor | Potential advantage |
|---|---|
| Free or low-cost access | Makes experimentation more affordable |
| Open-weight availability | Gives developers more deployment flexibility |
| Local deployment | Can help organizations retain control over their infrastructure and data |
| Model customization | Allows adaptation for specialized tasks |
| Global distribution | Helps products reach users across different markets |
Open-weight availability does not automatically mean a model is completely open source. Licences can impose conditions on use, redistribution or commercial deployment. Organizations must review the terms before integrating a model into a product.
How China Has Advanced Despite Chip Restrictions
China’s progress has occurred alongside US restrictions on exports of some advanced semiconductor technologies. These restrictions have limited access to certain high-performance chips and related equipment, which are important for training and operating frontier AI models.
Schmidt has pointed to the role of advanced computing resources and techniques such as knowledge distillation in China’s AI progress. Distillation involves using outputs from a more capable model to help train another model to reproduce or learn similar behavior.
This technique is widely used in AI development, although its legality and legitimacy depend on the methods, permissions and terms involved. Its use should not automatically be equated with unauthorized copying.
Chinese AI developers have also pursued software and engineering improvements that can make models more efficient on available hardware. These approaches can reduce some computing requirements, although they do not eliminate the importance of access to advanced chips, data centers and energy.
The broader issue is that hardware restrictions alone do not determine a country’s AI trajectory. Software improvements, research talent, engineering efficiency and access to global users all influence how quickly AI capabilities advance.
Does the Six-Month Gap Mean the US Has Lost Its AI Lead?
No. Schmidt’s estimate does not establish that the United States has lost its technical lead, nor does it prove that China will overtake it.
AI leadership is multidimensional. Frontier model performance is one measure, but the broader ecosystem includes semiconductor design, computing infrastructure, research, investment, developer adoption and commercial deployment.
The Associated Press reported in September 2026 that Chinese companies had made significant advances and released models competing with leading US systems, while the United States retained advantages in several areas, including advanced models and research.
The report also described US allegations that some Chinese developers had used large-scale distillation to extract capabilities from American models. Chinese authorities have rejected accusations of wrongdoing, while researchers have noted that distillation itself is a common technique in AI development. The claims about specific alleged activities remain contested.
Consequently, the six-month estimate is best viewed as an indicator of how closely leading systems may be competing, rather than a complete assessment of either country’s AI industry.
What the AI Race Means for Businesses and Developers
For businesses, the narrowing capability gap could expand the range of models available for commercial applications. Companies may increasingly compare models based on performance, cost, speed, deployment options and the rules governing their use.
A startup developing an AI-powered application might use a closed model for complex tasks and a cheaper open-weight model for routine processing. An enterprise with strict data requirements might prefer a model that can run on its own infrastructure.
These choices also create new considerations around security, intellectual property, compliance and support. A model that is inexpensive to access may still require substantial computing resources, monitoring and engineering work before it is ready for production.
For developers, the growing range of competitive models could reduce dependence on a small number of providers. For AI companies, it increases pressure to improve model quality while controlling the cost of inference and distribution.
The Bigger Picture
Schmidt’s comments point to a changing definition of AI leadership. The first phase of competition focused heavily on training increasingly capable models and securing the computing infrastructure needed to develop them. The next phase is likely to place greater emphasis on deployment: how easily models can be integrated into products, how much they cost to run and how widely developers and businesses adopt them.
The United States and China have different strengths across research, hardware, software, commercial ecosystems and distribution. A narrow performance gap between leading models does not erase those differences, but it makes competition over price and accessibility more consequential. The companies that combine useful capabilities with affordable and reliable products may gain influence even when they are not first to achieve a particular technical milestone.
FAQs
1. What did Eric Schmidt say about China’s AI progress?
Schmidt said China’s leading AI models are roughly six months behind the US frontier. He also warned that low-cost Chinese models could gain widespread adoption even if American systems remain technically ahead.
2. Why does Schmidt believe free AI models matter?
Free or inexpensive models reduce the cost of experimenting with AI and building products. This can make them particularly attractive to developers, small businesses and users in markets where premium AI services are less affordable.
3. What are open-weight AI models?
Open-weight models make their trained parameters available for download, allowing developers to run them on their own infrastructure or adapt them within the relevant licence terms. They are not necessarily unrestricted or fully open source.
4. How has China continued advancing its AI models?
Chinese developers have pursued model research, software optimization and techniques such as distillation while facing restrictions on access to certain advanced chips. The relative contribution of each factor varies by model and company.
5. Does China’s six-month gap mean it will overtake the US?
No. The estimate is Schmidt’s assessment, not a prediction that China will become the overall AI leader. Leadership depends on multiple factors, including model capabilities, hardware, research, costs, adoption and commercial execution.
Looking Ahead
The next stage of the AI competition will depend on how quickly leading companies improve their models and turn those capabilities into widely used products. Developers and businesses will be watching not only benchmark results but also pricing, reliability, open-weight licensing, hardware availability and the practical costs of deploying AI at scale.
Schmidt’s remarks underscore why the race cannot be measured by technical performance alone. If Chinese companies can deliver sufficiently capable systems at lower prices and make them easy to adopt, they may expand their global footprint even while US laboratories continue pushing the frontier. Whether that translates into a lasting commercial advantage will depend on adoption, product quality and the evolving economics of AI.
Get the day’s top stories in your inbox
One concise email. No spam, unsubscribe anytime.
