The United States is ahead of China in advanced artificial intelligence (AI) research, but China has taken the lead in the development and adoption of open-weight AI models, IBM Chairman and CEO Arvind Krishna has said. Speaking to NDTV CEO and Editor-in-Chief Rahul Kanwal in a conversation on the global AI race, Krishna said the US continues to lead in the development of the most advanced AI models. However, he pointed out that China is making significant progress in applying AI across businesses and industries.
Responding to a question on whether the US is maintaining its lead over China in the AI race, Krishna said, "I think that if you look at the very leading edge of the science, very large models, I think the first three models all came out of the US."
He added that he did not believe China currently had a model that could compete with leading American AI systems such as ChatGPT, Claude and Gemini.
Krishna also highlighted the US contribution to some of the most important developments in AI research, including the original work on transformers.
"Great. If I look at the original papers around intent, around transformers, those came out of the United States," he said.
However, he acknowledged that China appears to be ahead when it comes to the practical use of AI across several sectors.
"Usage-wise, China seems to be ahead on many more of the non-leading edge use cases, using it for business, using it across for customer service, using it for autonomy. That's where they seem to be leading."
Despite China's progress in commercial applications, Krishna believes the US will continue to lead in advanced AI research.
"But I think on the actual science, the US is ahead and will likely remain ahead," he said.
China Has An Edge In Open-Weight AI Models
One area where Krishna believes China currently has an advantage is open-weight AI models. Unlike proprietary models, open-weight models allow developers and businesses to access and adapt a model's underlying trained parameters, subject to its licence and terms.
Krishna argued that the US should invest more in this area rather than focus primarily on proprietary AI systems.
"If you're willing to adopt and use smaller models and use more open-weight models, so I actually advocate strongly that open-weight models is an area where China is ahead," he said.
He described China's lead in this segment as being around six months and questioned why Western companies were not investing more aggressively to close the gap.
"It's like a six-month lead. Why shouldn't we invest in the West on open-weight models also so we can catch up in the one area where maybe we are not?"
Watch the video here:
#NDTVExclusive | Who is ahead in the AI race: the US or China? Watch what @ArvindKrishna, Chairman and CEO, IBM, has to say @rahulkanwal https://t.co/kfZeU0neoL
— NDTV (@ndtv) November 17, 2026
80% Of AI Usage Could Shift To Open-Weight Models
The discussion also turned to whether proprietary AI companies such as OpenAI and Anthropic could face greater competition from Chinese open-model developers.
Asked whether the AI race would eventually produce one winner or whether proprietary and open models would coexist, Krishna said he expected both approaches to have a role.
"I think it will be an and. The question is, in what ratio?"
Offering his prediction for the next three to five years, Krishna said open-weight models could account for the majority of AI usage.
"I suspect that in three to five years, and I am more on the inner, the three-year, 80% of all tokens will be on open-weight models and 20% of tokens will be on large proprietary models."
This prediction could have important implications for the business models and valuations of leading AI companies, including OpenAI and Anthropic, which have invested heavily in proprietary AI systems.
What Does This Mean For OpenAI And Anthropic?
Commenting on the possible impact on the valuations of companies such as OpenAI and Anthropic, Krishna said the key question would be how much these companies could charge for their services.
"I'm not a Wall Street guy, so this thing, it depends on what prices they can charge. But they're not going to have 100% of the usage. It's only going to be 20%."
He suggested that while proprietary AI models would continue to have significant commercial value, their share of the overall AI market might be smaller than some current projections suggest.
"I suspect it's not as large as some of the projections of today, but neither is it trivial. It is large, but maybe not as."
Krishna's assessment points to a future in which the AI industry is not dominated by a single company or business model. Instead, advanced proprietary models and more accessible open-weight alternatives could serve different needs.
For the US, maintaining its lead in cutting-edge AI research may not be enough. Greater investment in open-weight models could also become important as competition with China intensifies.
Track Latest News Live on NDTV.com and get news updates from India and around the world