Mr. Ratanaphon Wongnapachant, CEO of SIAM.AI Cloud, celebrated the company’s achievement for being one of the first countries in South East Asia to receive a world-class infrastructure in technology such as the NVIDIA GB200 NVL72.
This acquisition reflects Thailand’s status in the world’s top level in creating AI infrastructure. After revealing this supercomputer module, the company hopes to receive more demand from domestic and foreign users.
NVIDIA GB200 NVL72 will empower business companies, research institutions, and government agencies through its AI capability and advanced process, accelerating innovation in healthcare, finance, manufacturing, smart cities, and other industries.
This supercomputer module costs about THB 100 million. Two of this new GPU is in the possession of Siam AI with an aim to import additional 46 before May this year. Its server equipment, however, costs approximately THB150 million, and there are also network and installation processes, which cost around THB200 million.
As for the NVIDIA DGX Blackwell B200, SIAM.AI Cloud had already purchased eight, with 1.4TB of GPU memory space. Each costs THB11 million but can train AI three times faster than its predecessor model and 15 times faster than the current one used to train SiamGPT. SIAM.AI Cloud plans to reveal these supercomputer modules by 2Q25.
As SIAM.AI Cloud becomes Thailand’s first company to be selected as NVIDIA Cloud Partner (NCP), the new AI cloud infrastructures will be constructed at Chon Buri and Pathum Thani, using its THB3.25 billion investment. Previously, the company received investment support from The Board of Investment Office (BOI) through tax exemption. This project is one of many that the government supports and can begin attending to its customers, which mostly are large hyper-scale organizations, immediately.
Furthermore, there will be an acquisition of GPU clusters worth more than THB7 billion. The company forecasted the investment in AI infrastructure would exceed THB70 billion this year, excluding the investment in Siam GPT and Siam Robotic.
