ByteDance Nvidia Blackwell chips have become a new flashpoint in the global AI infrastructure race. The latest reports indicate ByteDance is working with Southeast Asia-based Aolani Cloud to deploy about 500 Nvidia Blackwell systems in Malaysia, adding up to roughly 36,000 B200 chips for AI workloads outside China.
The move reflects a broader model already associated with Chinese AI groups using leased compute in Southeast Asian data centers rather than placing the most advanced training clusters inside China. That approach has been linked to overseas facilities in Singapore and Malaysia, where operators can host high-end Nvidia hardware while customers train models remotely.
The wider policy backdrop helps explain why this matters. U.S. officials tightened China-focused AI and advanced computing controls in March 2025 and later discussions around location verification and other safeguards signaled growing concern about offshore deployment and rerouting risks. At the same time, separate reporting has described an alleged chain in which 32 GB200 server racks containing about 2,300 Blackwell chips moved through an Indonesian telecoms provider before reaching a Chinese AI customer.
For ByteDance, the Malaysia buildout suggests the next phase of AI competition may depend less on domestic chip access alone and more on who can secure compliant overseas capacity at scale.
