ByteDance bets on Malaysia to access cutting-edge Nvidia chips
ByteDance, the company behind TikTok and one of China’s biggest tech giants, has found an alternative and highly strategic route to get its hands on the most powerful processors on the market. According to information published by the Wall Street Journal, the company is building up computing power with top-tier Nvidia chips outside of China, in partnership with Aolani Cloud, a cloud infrastructure provider based in Southeast Asia. The goal is to deploy approximately 500 Nvidia Blackwell systems in Malaysia, totaling around 36,000 B200 chips — an investment that, if completed, could exceed $2.5 billion. This move comes at a sensitive time in trade relations between China and the United States, as Washington maintains strict restrictions on exporting advanced semiconductor technology to Chinese companies, making the choice of Malaysia as an operational base a calculated step loaded with geopolitical significance.
The current landscape is pretty clear: Chinese companies that rely on cutting-edge hardware to train and run AI models are facing increasingly steep barriers to acquiring chips directly. U.S. sanctions, which have been expanding since 2022, specifically target high-performance processors used in data centers focused on artificial intelligence. Nvidia, which dominates this market by a wide margin, has already had to create limited versions of its products to serve customers in China, but even those alternative versions face growing regulatory scrutiny. Faced with this blockade, ByteDance went with an approach that isn’t exactly new but is gaining scale and visibility through this partnership: setting up infrastructure outside mainland China, in a country that isn’t in the crosshairs of U.S. restrictions and that simultaneously offers logistical and regulatory advantages for large-scale operations.
The role of Aolani Cloud and Aivres in this equation
Aolani Cloud isn’t a name as well-known in the West as AWS, Google Cloud, or Azure, but it has been gaining relevance precisely by filling a strategic space in Southeast Asia’s cloud ecosystem. The company offers cloud infrastructure services focused on high-performance computing, and Malaysia has been one of its main operational hubs. An Aolani spokesperson told the Wall Street Journal that the company currently operates with roughly $100 million in hardware, which gives you an idea of the massive leap this deal with ByteDance would represent for the business.
Another important link in this chain is Aivres, the company responsible for assembling the servers using Nvidia chips. According to the WSJ report, it is through Aivres that Aolani Cloud acquires the servers that will be used by ByteDance. This structure of intermediaries might seem complex at first glance, but it makes perfect sense from both a regulatory and operational standpoint. The chip purchases happen through entities based outside mainland China, and the servers are assembled and deployed on Malaysian soil, creating a flow that, at least on paper, complies with current export rules.
ByteDance’s choice of this partner is anything but random — Aolani Cloud functions as a bridge that allows the Chinese giant to access state-of-the-art Nvidia hardware without directly running into U.S.-imposed restrictions. In practice, since the operation runs through entities based in Malaysian territory, the transaction follows a different regulatory path than what would apply if ByteDance tried to acquire the same equipment directly in China.
Receive the best innovation content in your email.
All the news, tips, trends, and resources you're looking for, delivered to your inbox.
By subscribing to the newsletter, you agree to receive communications from Método Viral. We are committed to always protecting and respecting your privacy.
Malaysia as a global infrastructure hub for artificial intelligence
It’s worth pointing out that Malaysia has been positioning itself more and more aggressively as a tech hub in Southeast Asia. The country offers tax incentives, relatively cheap energy, and a prime geographic location to serve markets across both Asia and Oceania. Major tech companies, including Microsoft and Google, have already announced billions of dollars in data center investments in the country over the past two years. ByteDance’s entry with a project of this magnitude, brokered through Aolani Cloud, reinforces Malaysia as a top destination for AI infrastructure and puts the country at the center of a global race for computing capacity.
For Aolani Cloud, the deal represents a huge jump in scale and visibility, positioning the company as a viable partner for global-level AI infrastructure projects. We’re not talking about a small contract here — if the reported numbers are accurate, this single project could multiply the value of hardware Aolani currently operates by more than twenty times.
Another important point is that Aolani Cloud doesn’t just provide the physical space for the servers — it also offers technical support and infrastructure management, which allows ByteDance to focus its resources on developing AI models without having to worry about all the operational complexity of maintaining a large-scale data center in a country where it didn’t previously have a significant presence. This division of responsibilities is common in projects like these, but the scale — 36,000 B200 chips — puts this deal on an entirely different level from what we typically see in traditional cloud computing partnerships.
What ByteDance plans to do with all that computing power
According to the Wall Street Journal report, ByteDance plans to use this computing capacity for AI research and development outside of China and to meet the growing global demand for artificial intelligence from its clients. This is a relevant detail because it shows the company isn’t just thinking about TikTok or its consumer-facing products. ByteDance wants to position itself as an AI platform capable of offering computing services and advanced models to the international market.
Training large-scale language models, known as LLMs, requires an insane amount of computing power. Every new model that comes along — whether from Google, OpenAI, or Meta — demands clusters with tens of thousands of GPUs working simultaneously for weeks or even months. Having access to 36,000 B200 chips built on the Blackwell architecture puts ByteDance in a position to compete on equal footing with these companies, at least when it comes to infrastructure. It’s a clear signal that the company is serious about its ambition to become a global AI powerhouse, and not just a social media company with smart algorithms.
Nvidia Blackwell B200 chips and the race for AI supremacy
Nvidia’s Blackwell B200 chips represent the most advanced AI processors currently available on the market. The Blackwell architecture was specifically designed to handle the extreme processing demands that large language models and other generative AI applications require. We’re talking about chips capable of performing trillions of operations per second, with significantly better energy efficiency than the previous generation.
For ByteDance, having access to 36,000 of these units means being able to train proprietary AI models with speed and quality comparable to — or even surpassing — what companies like OpenAI and Google are doing. It’s a heavy bet, but one that makes sense when you consider the scale of the company’s ambitions in the artificial intelligence space.
The context of U.S. sanctions and the H200 chip situation
Nvidia, for its part, finds itself in a tricky position. The company is by far the largest supplier of AI hardware in the world, and demand for its products is so high that customers often wait months to receive their orders. At the same time, the company has to navigate a regulatory minefield, since the U.S. government could interpret certain sales to partners of Chinese companies as a violation of the spirit of the sanctions, even if technically the transaction falls within the rules.
One detail that adds even more layers to this story: in February, Reuters reported that the United States was willing to allow ByteDance to purchase Nvidia H200 chips, but Nvidia itself reportedly didn’t agree to the proposed conditions for the use of those processors. In other words, even when there’s a green light from the U.S. government, negotiations involving Chinese companies and advanced hardware remain extremely complicated. This piece of information reinforces why the Malaysia route, with Blackwell B200 chips and intermediaries like Aolani Cloud and Aivres, becomes such an attractive alternative for ByteDance.
The deal between ByteDance and Aolani Cloud will certainly attract Washington’s attention, and it wouldn’t be surprising if new regulations emerge in the coming months to try to close loopholes like this one. Nvidia has publicly stated that it complies with all current export laws, but the political pressure around advanced chips and China is only going to increase. As of the publication of the Wall Street Journal report, neither Nvidia, ByteDance, nor Aolani Cloud had responded to requests for comment made by Reuters.
What this means for the future of global AI
What becomes clear from this move is that the race for AI computing capacity is reshaping alliances, trade routes, and even global geopolitics. ByteDance is showing it won’t just sit around waiting for restrictions to ease — quite the opposite, it’s seeking creative and legally viable paths to stay competitive. The partnership with Aolani Cloud in Malaysia is just the most recent and most visible example of a trend that’s likely to intensify in the coming years: Chinese tech companies building AI infrastructure in third-party countries, creating a new global map of data centers that reflects not only technical demands but also political realities.
This phenomenon is already being called by some analysts AI geography — the geography of artificial intelligence. Countries like Malaysia, Indonesia, the United Arab Emirates, and even Mexico are starting to emerge as strategic destinations for data centers that, under normal circumstances, would be built on American or Chinese soil. The reason is simple: these locations offer a regulatory middle ground, competitive operational costs, and in many cases, governments willing to roll out the red carpet for billions of dollars in tech investment.
For anyone following the industry, one thing is certain: Nvidia chips remain the most valuable currency of the artificial intelligence revolution — and everyone is willing to go far, literally, to get them. ByteDance, with this move in Malaysia, is making it clear that it intends to be a leading player in this race, no matter what barriers come its way 🚀
