.png?w=1024)
Southeast Asia is becoming a new front in the race to build the infrastructure behind frontier AI.
OpenAI has signed a multiyear agreement with Australian AI infrastructure company Firmus for dedicated computing capacity at two planned data centers in Malaysia. The deal makes OpenAI an anchor customer, with the new facilities slated to use Nvidia Vera Rubin systems. Firmus separately operates a broader AI infrastructure portfolio spanning Malaysia, Singapore, Indonesia, and Australia.
For Malaysia, the agreement adds another heavyweight name to a data-center market already expanding rapidly. For OpenAI, it offers something equally valuable: additional computing infrastructure to support growing AI demand across Asia.
OpenAI locks in Malaysian AI capacity
Under the new agreement, OpenAI will contract capacity from two Firmus AI Factory sites planned for Malaysia. Financial terms were not disclosed.
The partnership pushes Firmus’ total contracted capacity across its customers above 900 megawatts. The company currently has seven AI factory sites across Australia, Singapore, Indonesia, and Malaysia, although only its Australian and Singapore locations are operational today.
Five additional sites are under development and are expected to become ready for service over the next 24 months, according to Firmus.
The Malaysian facilities will use Nvidia’s DSX AI Factory platform and Vera Rubin NVL72 systems, according to Firmus. Nvidia says its Vera Rubin platform is designed for large-scale AI training and inference and can deliver up to 10 times more agent throughput at scale than its previous-generation Grace Blackwell platform. Nvidia said Vera Rubin was ramping into full production earlier this year.
The 10x performance figure is according to Nvidia.
The deal extends a broader push by AI companies to secure enormous amounts of computing infrastructure as demand for training and inference continues to grow.
What eWeek found: Southeast Asia’s AI race is becoming an infrastructure race
OpenAI’s Malaysia deal fits into a much larger regional shift. Across Southeast Asia, governments and technology companies are pouring money into the physical infrastructure needed to support advanced AI.
Here’s what stands out:
- Malaysia is attracting major AI infrastructure investment. The Malaysian Investment Development Authority said data-center and cloud-computing projects accounted for RM95.8 billion in approved investment during the first half of 2026, close to 44% of the country's total approved investment during that period. MIDA has positioned Malaysia as a growing Southeast Asian hub for digital infrastructure.
- OpenAI is joining an already accelerating buildout. In August, eWeek reported on a $150 million Malaysian AI infrastructure agreement spanning AI compute, cybersecurity, networking, and data-center systems.
- Singapore is expanding more selectively. The city-state recently provisionally allocated another 200 MW of data-center capacity to four operators while maintaining energy-efficiency and sustainability requirements.
- The regional AI hardware economy is widening. In the Philippines, electronics industry leaders have projected that exports could reach roughly $53 billion to $54 billion in 2026, with growing AI and data-center demand helping drive demand for semiconductor and electronics components. eWeek recently examined that trend.
Taken together, the pattern is becoming harder to miss. Southeast Asia is no longer simply a market where U.S. AI companies sell models and services. Parts of the region are becoming increasingly important locations for the infrastructure and components that support those systems.
Why Malaysia matters to OpenAI
Compute has become one of the most important constraints facing frontier AI companies.
Larger models, reasoning systems, and increasingly autonomous AI agents can require enormous amounts of processing capacity, turning access to GPUs, electricity, cooling, networking, and data-center space into strategic assets.
Firmus gives OpenAI access to another source of dedicated regional computing capacity while expanding the company’s infrastructure footprint in Asia-Pacific.
Malaysia has attracted data-center investment through a combination of strategic location, regional connectivity, a pro-business investment environment, and its proximity to Singapore. That growth also comes with resource constraints. Malaysia’s Data Centre Task Force clears projects only when they have secured power and water and can demonstrate green compliance, according to MIDA.
Those considerations are likely to become more important as higher-density AI hardware requires increasingly large amounts of power and cooling.
Southeast Asia is moving closer to the center of AI
For enterprise technology leaders, the geography of AI infrastructure matters because it can affect where computing capacity is available, how quickly providers can expand services, and how resilient regional infrastructure becomes.
The OpenAI-Firmus deal does not mean Malaysia is about to replace the United States as the center of frontier AI development. Nor does the agreement establish new OpenAI data-residency policies for customers in the region.
What it does show is that the physical infrastructure supporting frontier AI is becoming more geographically distributed.
That could prove especially important in Southeast Asia, where governments are simultaneously pursuing AI investment, sovereign computing capacity, local technology ecosystems, and tighter oversight of energy-intensive infrastructure.
The next phase of the AI race may still be measured in models and benchmarks. Increasingly, though, it is also being measured in megawatts, GPUs, and the countries willing to host them.
Also read: See how Cisco, Nvidia, and Supermicro are building rack-scale AI factories as demand for high-density compute, networking, and cooling continues to grow.