Is China quietly winning the AI race?

Every month, hundreds of millions of users flock to Pinterest seeking inspiration for the latest styles. While many may be unaware, the underlying technology powering these personalized recommendations isn’t exclusively American. Pinterest, a prominent US-based platform, is actively experimenting with Chinese artificial intelligence (AI) models to enhance its recommendation engine. "We’ve effectively made Pinterest an AI-powered shopping assistant," stated Bill Ready, the company’s chief executive.

While San Francisco-based Pinterest has access to numerous American AI labs, the landscape has shifted significantly since the January 2025 release of China’s DeepSeek R-1 model. This marked a pivotal moment, often referred to as the "DeepSeek moment," which Ready described as a breakthrough. "They chose to open source it, and that sparked a wave of open source models," he explained. This strategic move by DeepSeek has spurred a flurry of innovation from Chinese competitors such as Alibaba’s Qwen and Moonshot’s Kimi, with TikTok’s parent company, ByteDance, also developing similar advanced AI technologies.

Matt Madrigal, Pinterest’s Chief Technology Officer, highlighted a key advantage of these Chinese models: their open-source nature. This allows companies like Pinterest to freely download and customize them, a stark contrast to the proprietary models offered by many US rivals, including OpenAI, the creator of ChatGPT. Madrigal noted, "Open source techniques that we use to train our own in-house models are 30% more accurate than the leading off-the-shelf models." Furthermore, these enhanced recommendations come at a significantly reduced cost, sometimes up to ninety percent less than employing the proprietary models favored by US AI developers.

These "fast and cheap" Chinese AI models are rapidly gaining traction beyond Pinterest, influencing a growing number of Fortune 500 companies. In October, Airbnb’s CEO, Brian Chesky, revealed to Bloomberg that his company relies "a lot" on Alibaba’s Qwen to power its AI customer service agent, citing its effectiveness, speed, and affordability. Further testament to this trend can be observed on Hugging Face, a leading platform for downloading AI models, where Chinese developers are increasingly making their mark.

Jeff Boudier, who leads product development at Hugging Face, pointed to the cost factor as a primary driver for startups choosing Chinese models over their American counterparts. "If you look at the top trending models on Hugging Face – the ones that are most downloaded and liked by the community – typically, Chinese models from Chinese labs occupy many of the top 10 spots," he stated. "There are weeks where four out of five top training models on Hugging Face are from Chinese labs." This dominance is underscored by Qwen topping Meta’s Llama in September to become the most downloaded family of large language models on the platform.

Meta’s open-source Llama AI models, released in 2023, were once the go-to choice for developers building bespoke applications. However, the release of Llama 4 last year reportedly left developers underwhelmed, prompting Meta to collaborate with companies like Alibaba, Google, and OpenAI to train a new model set for release this spring. While Airbnb utilizes a variety of models, including US-based ones, it hosts them securely within its own infrastructure, ensuring data privacy.

Entering 2025, the prevailing sentiment was that despite substantial investments by US tech firms, Chinese companies were poised to take the lead in AI development. "That’s not the story anymore," Boudier remarked. "Now, the best model is an open-source model." A recent report from Stanford University corroborates this shift, finding that Chinese AI models "seem to have caught up or even pulled ahead" of their global counterparts in terms of both capability and adoption.

Sir Nick Clegg, former UK deputy prime minister and former head of global affairs at Meta, expressed his view that US firms were overly focused on the pursuit of AI that might one day surpass human intelligence. This ambition, which Meta CEO Mark Zuckerberg has committed billions of dollars towards achieving what he terms "superintelligence," is now being characterized by some experts as vague and ill-defined. This focus, they argue, has created an opening for China to dominate the open-source AI space.

"Here’s the irony," Sir Nick mused. In the competition between "the world’s great autocracy" and "the world’s greatest democracy" – China and America – China is "doing more to democratize the technology they’re competing over." The Stanford report also suggests that China’s success in open-source AI development may be partly attributed to strong government support.

Conversely, US companies like OpenAI are facing immense pressure to increase revenue and achieve profitability, a challenge that is now leading them to explore advertising as a revenue stream. While OpenAI released two open-source models last summer, its first in years, the company has primarily directed its resources toward proprietary models to drive financial returns. OpenAI CEO Sam Altman confirmed in October an aggressive investment in securing computing power and infrastructure deals with partners. "Revenue will grow super fast, but you should expect us to invest a ton in training, in the next model and the next and the next and the next," he stated. This strategic divergence, with China embracing open-source development and US firms prioritizing proprietary, profit-driven models, is fundamentally reshaping the global AI landscape and potentially signaling a quiet shift in leadership.