Is China Quietly Pulling Ahead In The Global AI Race?
On the surface, Pinterest looks like a digital mood board for dreamers and trend hunters. Every month, hundreds of millions of users scroll through the platform in search of fashion ideas, interior inspiration and quirky creative concepts. One popular page showcases intentionally bizarre designs: Crocs turned into flower vases, burger-shaped makeup palettes, and even a vegetable-built “gingerbread” house.
What most users don’t see is the technology quietly shaping what appears on their screens. Behind Pinterest’s recommendation engine, Chinese-built artificial intelligence models are increasingly doing the heavy lifting.
Pinterest CEO Bill Ready describes the platform today as something far more advanced than a visual discovery app. “We’ve effectively turned Pinterest into an AI-powered shopping assistant,” he says. And since the release of China’s DeepSeek R-1 model in early 2025, Chinese AI tools have become a core part of that transformation.
Ready refers to the release as a turning point. DeepSeek’s decision to open-source its model triggered a surge of similar tools from Chinese tech firms, including Alibaba’s Qwen, Moonshot’s Kimi, and models under development at ByteDance, the parent company of TikTok.
Why U.S. Companies Are Turning To Chinese AI
For Pinterest’s engineering team, the appeal isn’t political or symbolic – it’s practical. Chief Technology Officer Matt Madrigal explains that open-source Chinese models can be freely downloaded, adapted and retrained internally, something most US-built alternatives do not allow.
According to Madrigal, Pinterest’s customised models now outperform many off-the-shelf systems by a wide margin. “The open-source techniques we use to train our in-house models are about 30% more accurate,” he says. The cost savings are even more striking, with expenses sometimes as much as 90% lower than using proprietary AI systems from major American providers.
Pinterest is far from alone. Across corporate America, Chinese AI tools are gaining ground. Airbnb CEO Brian Chesky revealed last year that the company relies heavily on Alibaba’s Qwen model to power parts of its customer service operations. His reasoning was simple: the technology is effective, fast, and affordable.
Airbnb maintains that it runs all AI systems within its own infrastructure, ensuring company data is never shared with model developers – Chinese or otherwise.
The Open-Source Advantage
The growing influence of Chinese AI is also visible on Hugging Face, the leading platform for developers to download and share machine learning models. Jeff Boudier, who works on product development there, says cost is a major factor driving adoption.
When developers browse the most popular and most downloaded models, Chinese-built tools frequently dominate the rankings. In some weeks, four of the top five training models originate from Chinese labs. Last September, Alibaba’s Qwen overtook Meta’s Llama family to become the most downloaded large language model on the platform.
Meta’s Llama models, released in 2023, were once the default choice for developers building customised AI applications. But enthusiasm cooled after the release of Llama 4, which many developers felt failed to live up to expectations. Meta has since reportedly leaned on external open-source models, including those from Alibaba, as it prepares a new generation of AI tools.
A Shift In The Global AI Narrative
At the start of 2025, many analysts believed Chinese companies were catching up to US AI leaders despite America’s massive financial investments. According to Boudier, that framing is now outdated. “The conversation has changed,” he says. “The strongest models today are open-source—and many of them are coming from China.”
A recent Stanford University report reinforces that view, suggesting Chinese AI models have matched or even surpassed their global competitors in both performance and adoption. The study also points to strong government backing as a factor behind China’s rapid progress in open-source development.
Former UK deputy prime minister Sir Nick Clegg believes strategic focus plays a role too. In his view, US firms have been overly fixated on the pursuit of artificial general intelligence – systems that might one day outperform humans – while China has concentrated on practical, accessible tools.
“There’s a deep irony here,” Clegg notes. In the competition between China and the US, he argues, China may be doing more to democratise the very technology both sides are racing to dominate.
Different Pressures, Different Paths
American AI firms face a different set of constraints. Companies like OpenAI are under mounting pressure to generate revenue and justify enormous infrastructure costs. That has pushed them toward closed, proprietary systems – and even advertising models – to sustain growth.
OpenAI CEO Sam Altman has acknowledged the trade-off, explaining that the company continues to invest heavily in computing power and future model training. While OpenAI released a pair of open-source models last year, the bulk of its resources remain tied to commercial products.
As global companies quietly integrate Chinese AI into everyday tools – from shopping recommendations to customer support – the question is no longer whether China is competing in the AI race. It’s whether the most influential part of that race is happening in plain sight, powered by open-source code and driven by practicality rather than hype.
