China’s AI Firms Are Addicted to Nvidia Chips. Now They Are Facing Withdrawal.
The move by the U.S. to cut off Chinese access to high-end computing power has dealt a blow to Beijing’s AI ambitions.
Chinese tech firms are struggling as they face growing restrictions on accessing Nvidia's powerful AI chips, according to recent reports. The U.S. Commerce Department has closed loopholes that allowed Chinese companies to bypass export restrictions by accessing Nvidia chips remotely through data centers in Southeast Asia. This move has particularly impacted companies like Megaspeed, which built data centers in the region and shipped billions of dollars worth of Nvidia chips to China.
The new rules have already begun to impact Chinese AI firms, with bankers reassessing plans to finance Nvidia chip purchases for ByteDance and an Aolani spokesman stating the company is working to remain compliant. China's Ministry of Commerce has criticized the new rules as an abuse of export controls, stating that they will significantly limit China's access to high-end American computing power.
China has long relied on Nvidia hardware to train its flagship AI models, but the new restrictions may now make it difficult for Chinese labs to access these chips remotely. Chinese AI companies are now exploring alternative methods to access data centers in the U.S. and Europe through local entities, but this may prove challenging due to risky arrangements and limited computing power.
Despite China's push to promote domestic chip production, the country remains at least 10 years behind the U.S. in making the best AI chips. Many of China's top AI labs still rely on Nvidia hardware for their flagship models, especially during intensive early-stage training. The new regulations have left Chinese AI companies searching for ways to maintain their computing power, with Huawei Technologies being the primary candidate to fill the void.
Huawei has been using their networking expertise from the telecommunications business to create large domestic computing clusters using AI chips. However, these clusters are not yet powerful enough to meet the demand for AI training. Additionally, U.S. export controls on chip-making equipment pose further challenges for China in producing high-quality AI chips in large quantities.
Written by urgent.news from Hindustan Times - World News's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.