TikTok’s Parent Company Advances AI Infrastructure in Southeast Asia
ByteDance, the company behind TikTok, is making significant strides in building computing infrastructure in Southeast Asia. This move, utilizing high-end Nvidia processors, seems to be part of its broader strategy to gain a competitive edge in global AI development. It’s an interesting tactic, perhaps, to navigate around U.S. export restrictions on AI technologies.
Reports indicate that ByteDance has partnered with Aorani Cloud, a local cloud computing provider, to roll out roughly 500 Nvidia Blackwell systems in Malaysia. These systems will be equipped with approximately 36,000 B200 AI chips. There’s a strong indication that this setup aims to bolster the company’s AI initiatives outside China while also catering to a growing international client base.
The deal involves Aolani purchasing servers from Aivres, which specializes in Nvidia-powered server assembly. If everything goes according to plan, the hardware expenses might exceed $2.5 billion—a substantial investment. A spokesperson from Aolani confirmed that, as of now, the company operates about $100 million worth of hardware.
About a quarter of ByteDance’s revenue comes from markets beyond China, positioning it to compete with major U.S. firms like Google and OpenAI in various AI applications aimed at everyday users. The company has launched over ten AI apps tailored for the Chinese audience, along with corresponding versions for international users. Some of these include a chatbot named Dola, a video creation tool called Dreamina, and an academic assistance app called Gauth.
ByteDance’s AI video generation platform, Seedance, has garnered considerable attention for its capability to transform written scripts into engaging short film sequences. According to a January study by venture capital firm Andreessen Horowitz, ByteDance operates five of the top 50 consumer AI applications globally, in terms of monthly active users.
The company supports a robust team of AI researchers and engineers located in Singapore, San Jose, and Seattle. These teams collaborate with their counterparts in China for fundamental AI research and contribute to developing AI models and products, many targeting non-Chinese markets.
This agreement underscores how Chinese tech firms are finding ways to bypass U.S. export controls that have hindered Nvidia from selling advanced AI chips directly to China for more than three years. Executives within China have expressed concerns that limited access to high-performance computing is stifling their AI progress.
In light of these export restrictions, Chinese companies are increasingly turning to foreign markets, where U.S. controls are less stringent. This rise has spawned a new market of intermediaries who help establish data centers equipped with Nvidia chips, leasing this computing power to Chinese tech enterprises.
A spokesperson from Aorani emphasized the company’s commitment to adhering to all relevant export control laws, clarifying that clients do not own or have claims over the chips in Aorani’s data centers.
Nvidia also commented on the export controls, stating they are intended to facilitate the establishment and operation of cloud infrastructure outside of regulated territories like China. To ensure compliance, Nvidia’s team scrutinizes all cloud partners before any chip sales, direct or indirect.
In a somewhat related issue, Winton Hall, author of “Code Red,” pointed out that China has been concealing its data-gathering AI tools within popular applications like TikTok. In a recent interview, he suggested that users of these apps might be inadvertently conceding their privacy and security to the Chinese government.
Hall noted that Chinese officials often boast about TikTok’s ability to influence young Western minds with its algorithms while simultaneously collecting vast amounts of data for intelligence purposes. This, he argues, is a strategic play against the United States, powered by AI.
Senator Marsha Blackburn recently described Hall’s work as a vital must-read, especially for those concerned about the implications of AI in line with conservative values. He pointed towards the importance of leveraging AI’s potential without compromising safety and privacy. Journalist Michael Shellenberger also endorsed Hall’s book, highlighting its critical role in fostering discussions on combating Big Tech’s authoritarian tendencies before it’s too late.
