ByteDance employs a workaround to the U.S. prohibition on straight purchases of Nvidia's advanced AI chips by leasing them from cloud providers based in the United States. Despite the U.S. government disallowing direct sales of these sophisticated AI chips to entities in China, their use remains permitted within American territory.
Navigating Regulatory Gaps
Several Chinese technology firms have exploited this regulatory loophole. According to The Information, ByteDance, the company that owns TikTok, reportedly leases Nvidia's high-end AI chips from Oracle to fulfill its AI computational requirements. Oracle runs TikTok in the US as a part of a licensing partnership with ByteDance. Similarly, state-run China Telecom has looked into similar arrangements with various cloud providers to gain access to these chips. Additionally, Alibaba and Tencent have discussed acquiring these Nvidia chips for U.S.-operated data centers.
American export controls aim to curb China's access to sophisticated technologies such as advanced AI. However, these restrictions do not cover scenarios where Chinese companies use these technologies within the U.S. This oversight allows companies like ByteDance to utilize cutting-edge AI resources without breaching export laws. The rules have meant Nvidia has been forced to sell downgraded chips in China.
Industry Impact and Possible Regulatory Revisions
Analysts believe that the way Chinese companies access AI chips through U.S. cloud services might prompt a review of current export regulations. The situation underscores the complexities involved in enforcing technology restrictions amidst global economic interdependencies. Nvidia has been engaging with different stakeholders to address these regulatory hurdles while continuing to cater to a global clientele.
Facing these developments, Chinese regulators have advised local tech firms to diminish dependence on Nvidia chips. Companies like Huawei are at the forefront of creating domestic alternatives. Meanwhile, Nvidia's acquisition of startups focusing on AI chip cost-efficiency represents a strategic move to stay competitive in the AI hardware industry.
