Unacknowledged portion of Silicon Valley’s AI is dominantly dependent on Chinese foundational research and engineering talent, birthing a co-dependency on Chinese AI models that could shatter the US tech sector, according to Bloomberg.
When Nvidia’s CEO Jensen Huang told the Financial Times that “China is going to win the AI race,” he triggered new chilling fears down Silicon Valley’s spine, as US developers, investors, and leading platforms increasingly adopt cheaper.
Huang then went to add that China was simply “nanoseconds behind,” highlighting the level of velocity in which the AI industry is moving and shifting toward.
For now, the US remains the dominant powerhouse in cutting edge chips and foundational breakthroughs, but the availability Chinese open-source AI, powered by low-cost models, massive state backing, and opensource strategies are quietly gaining global developer loyalty and even winning over high profile US firms.
Chinese AI Adoption
The shift has been visible for months, but only recently gained undeniable momentum. Venture capitalist Chamath Palihapitiya revealed on the All-In podcast that one of his companies moved significant workloads to Kimi K2 model from Beijing based Moonshot AI because it is “frankly just a ton cheaper than OpenAI and Anthropic.”
Airbnb CEO Brian Chesky followed with an equally telling admission. He said the company hadn’t integrated with ChatGPT because its tools weren’t “quite ready,” adding that Airbnb’s agent uses more than a dozen models and is “relying a lot” on Alibaba’s Qwen.
“It’s very good. It’s also fast and cheap,” he said, perfectly describing cheaper AI models, considering Chesky’s close ties to Sam Altman.
Even former OpenAI Chief Technology Officer (CTO) Mira Murati’s new startup, Thinking Machines Lab, gave credit to Alibaba Qwen models for inspiring its latest research.
Evidence of this movement also appears in unexpected places of Chinese AI models. Coding platform Cursor, valued around $10 billion, released an update whose inner monologue strangely switched to Mandarin fueling speculation it was built on DeepSeek.
In parallel, US unicorn, Cognition AI, appears to have based its new agent on a model from Zhipu AI cognition (Z.ai), which the Beijing firm tacitly acknowledged on X, saying it “highlights the positive impact and value of open-source contributions to the ecosystem.”
Hugging Face data, listed by the ATOM Project, confirmed the trends that in early 2024, Meta’s Llama saw 10.6 million downloads versus Alibaba Qwen’s 500,000. Last month alone, Chinese Qwen hit 385.3 million cumulative downloads, surpassing Llama’s 346.2 million, with Qwen-derived models accounting for over 40% of new uploads.
US developers still rely on higher AI chips Huang emphasized that Chinese AI venture capital “Nvidia builds the GPU, but we also build the CPU, the networking, the switches” but China’s low cost, open-source tools are pulling global talent toward Beijing’s ecosystem.
Silicon Valley and Chinese AI
Huang’s remarks came during multiple visits to Taiwan, where he stressed Chinese AI export controls and Nvidia’s reliance on manufacturing partner TSMC. Demand for Nvidia’s Blackwell chips remains “very strong,” with supply pressures rising across memory and component makers.
Still, political restrictions create this sort of tensions, “Currently, we are not planning to ship anything to China,” Huang said, mentioning US export controls and a lack of China AI development interest in the weaker H20 chip. “China does not want Nvidia in the country.”
Washington’s strategy of containing Beijing through an aggressive chip embargo is backfiring – to say the least.
At the same time, US policymakers are creating their own Chinese AI models competition strategy. The Trump administration’s AI Action Plan promises to do “whatever it takes to lead the world in artificial intelligence,” pushing for faster data center construction and accessible power supplies areas where China has surged ahead.
According to Communications Today, China is perceived as the one and only true US competitor in the global AI race. The more the West attempts to wall off Chinese technology, it is inadvertently catalyzing the creation of a parallel, possible more dominant AI ecosystem.
Eventually, if things remain as they are today, the West will no longer control or even benefit from this AI ecosystem.
Beijing’s expansion is backed by huge public investment, AI focused curriculum in schools, and aggressive state support. Its AI companies DeepSeek, Baidu, Alibaba, Moonshot, Zhipu and others now rival US systems in reasoning, coding, and multimodal capabilities, with Chinese AI models like DeepSeek AI R1 built at a fraction of US costs.
Yet US leaders insist the gap remains. OpenAI, Google and xAI are still top global leaderboards such as Chatbot Arena and Humanity’s Last Exam.
“It is very hard to say how far ahead we are,” Sam Altman told Congress. “But I would say not a huge amount of time.”
While Washington maintains key advantages in chips and limit research labs, China’s open-source rush, cheaper model ecosystem, and enthusiastic adoption by US developers have improved the competitive balance.
If the US wants to stay on top, Huang warned, it must “run fast” because China already is and everyone knows how fast Asians are.
Inside Telecom provides you with an extensive list of content covering all aspects of the tech industry. Keep an eye on our Intelligent Tech sections to stay informed and up-to-date with our daily articles.