When you buy through affiliate links in our content, we may earn a commission at no extra cost to you. Learn 
China has achieved a significant breakthrough in AI, successfully training a generative AI model across multiple data centers and GPU architectures.
This feat is especially notable given the current restrictions imposed by U.S. sanctions that limit China’s access to advanced chips.
Overcoming GPU Shortages Through Innovation
One of the primary hurdles for China in its AI journey has been the limited access to top-tier GPUs (Graphics Processing Units) due to U.S. sanctions. These restrictions prevent Chinese companies from acquiring cutting-edge chips like Nvidia’s A100, which is crucial for training sophisticated AI models.
In response, Chinese researchers have developed ways to combine available GPUs from various brands to create a unified training environment.
China firms' AI breakthrough can meld GPUs from different brands into one training cluster — Baidu says new tech fuses thousands of GPUs together to help sidestep shortages https://t.co/gMFcIJ8z74 pic.twitter.com/UTgG875w0k
— Anj Bryant (@anjbryant) May 18, 2024
Training a generative AI (GAI) model across different data centers, as China has now done, is a technical challenge.
Traditionally, training models across GPUs within a single data center is difficult enough, let alone involving multiple data centers. Patrick Moorhead, Chief Analyst at Moor Insights & Strategy, shared on X that China was the first country to accomplish this milestone.
The U.S. had imposed these restrictions as part of a broader strategy to curb China’s AI ambitions, fearing that advanced AI technologies could be used for military purposes. To comply with these sanctions, Nvidia developed the H20 chip, a weaker version of its more advanced models.
However, reports suggest that even these H20 chips could soon face further restrictions, adding more pressure on China’s AI industry.
However, Chinese tech companies like Huawei have maintained their commitment to advancing AI development. Huawei has publicly stated that it will continue to push forward with AI research, and this latest achievement demonstrates the country’s determination to remain a global player in the AI field.
China has long set ambitious goals for becoming a leader in AI by 2030. With the government investing heavily in AI research, there has been rapid progress in areas such as machine learning, facial recognition, and natural language processing.
However, the reliance on imported technology has remained a vulnerability, especially as tensions with the U.S. escalated.
By developing new methods for training GAI models with mixed GPU architectures across multiple data centers, China has shown that it can adapt to the constraints imposed by sanctions.
