AMD, NeutraDC partner to advance AI across Southeast Asia

NeutraDC Singapore has signed a memorandum of understanding with AMD to explore the use of AMD Instinct accelerators and collaborate on AI integration projects in Southeast Asia.

NeutraDC, part of Telkom Indonesia and operator of data centers in Indonesia and Singapore, said the agreement includes technical consulting, joint research, and training programs focused on AI.

The partnership was initiated in October 2025 and aims to test proof-of-concept deployments and develop AI-driven services for enterprise customers.

Executives from both companies said the collaboration will expand access to high-performance AI infrastructure in the region.

NeutraDC operates 29 data centers and provides cloud-connect and data center services to businesses in Southeast Asia.

No financial details or specific project timelines were disclosed.

🔗 Source: NeutraDC

🧠 Food for thought

Implications, context, and why it matters.

NeutraDC’s AI plans rest on high-density power and liquid cooling at scale

The MoU hinges on NeutraDC hosting AMD Instinct MI300-class (AMD’s data center AI accelerator family) clusters that draw high power and need liquid cooling. T5 pegs AI GPU spaces near 700 watts per square foot 1, while NeutraDC has detailed such capability only for its under-construction Batam, Indonesia site 2.
Batam markets high-density racks and liquid cooling (circulating fluid to remove heat) 2, with 18 MW installed capacity and operations slated for H1 2026 3. Absent verified rack power density or confirmed liquid cooling at Singapore or other active Indonesian sites, where AI racks can need 30 kW or more per rack versus 5–10 kW 1, any rollout likely trails a long “technical consulting” phase.

AMD resellers and system integrators in Southeast Asia see rising MI300 AI demand

Enterprises want AMD Instinct MI300 infrastructure in Southeast Asia, based on this partnership. That creates near-term openings for authorized AMD distributors and OEMs such as Supermicro, a server maker with MI300 systems 4.
System integrators (IT services firms that build and deploy systems) can pitch preconfigured options like Supermicro’s 8U 8-GPU MI300X or 4U air-cooled MI300A systems 4 for AI training that leans on high on-GPU memory bandwidth. AMD-authorized distributors in Asia Pacific 5 can run proof-of-concept (PoC) programs using AMD remote access to MI300X accelerators 6.