In a covert operation that highlights the lengths to which Chinese AI companies are going to evade U.S. export controls, four Chinese tech workers recently flew from Beijing to Kuala Lumpur, each carrying 15 hard drives packed with 80 terabytes of data intended for training large-scale AI models.
Collectively, the drives held approximately 4.8 petabytes of information — enough to fuel the development of multiple advanced large language models (LLMs).
The operation, meticulously planned over months to avoid scrutiny from customs authorities, underscores the challenges of transferring massive datasets online without attracting unwanted attention.
Transmitting such volumes of data through conventional internet channels would have been time-consuming and risked detection by regulatory bodies monitoring cross-border data flows.
Upon arrival in Malaysia, the suitcases full of hard drives were delivered to a data center where 300 Nvidia AI servers had been leased for processing and model training.
This move reflects a growing trend among Chinese firms seeking to circumvent U.S. restrictions on the export of advanced AI chips and technologies critical to building cutting-edge AI systems.
Also read:
- What Fuels 'Dad TV'? A Peek into Fathers' Viewing Habits
- Amazon Doubles Ad Load on Prime Video, Breaking Trust with Users
- Zuckerberg Assembles Team to Build Superintelligence
The incident illustrates the difficulty Washington faces in enforcing its bans on exporting high-performance AI hardware to China. Despite tightened regulations, Chinese companies are finding creative workarounds, from leveraging third-party countries like Malaysia to physically transporting data in suitcases.
These efforts reveal the high stakes in the global AI race, where access to data and computing power is paramount.
As the U.S. continues to tighten its grip on AI-related exports, such operations may become more common, raising questions about the effectiveness of current restrictions and the lengths to which companies will go to stay competitive in the rapidly evolving AI landscape.
