Firmus and Meta Partnered on AI Infrastructure
The agreement secures compute capacity across Southeast Asia to accelerate Meta's AI model development.
Updated on Sept. 28, 2026 in Data Centers
Live Poll
Do you trust that large-scale infrastructure partnerships between major tech companies benefit the general public?
Firmus and Meta have entered into strategic AI infrastructure agreements to provide the social media giant with dedicated compute capacity at Firmus sites throughout Southeast Asia. Meta will serve as the anchor customer for the Melbourne AI factory, which utilizes Nvidia Blackwell Ultra chip systems.
Why it matters
This partnership addresses Meta's increasing requirement for large-scale infrastructure to support its intensive AI research and model development. The move highlights the critical race to secure high-performance compute capacity in the Asia-Pacific region.
The Melbourne facility leverages Nvidia Blackwell Ultra chip systems to support high-performance computing requirements. The infrastructure is specifically engineered to maximize compute performance per watt of energy.
The players
Firmus
An infrastructure developer specializing in building and operating energy-efficient AI factories across the Asia-Pacific region.
Meta
A global technology company focused on social networking and intensive research into artificial intelligence and machine learning models.
Nvidia
The dominant designer of high-performance graphics processing units that power the underlying compute architecture for modern AI research.
The details
Firmus constructs specialized AI factory sites across the Asia-Pacific region designed for high-density compute loads. These facilities prioritize electrical efficiency, optimizing the ratio of computational output relative to energy consumption. By using Nvidia Blackwell Ultra systems—a series of advanced graphics processing units designed for large-scale machine learning—the infrastructure provides the high memory bandwidth required for training large language models.
Timeline
September 22, 2026: Firmus and Meta officially announced the infrastructure agreements.
The Tech Race
This deployment of Nvidia Blackwell Ultra systems marks a key milestone in the regional competition to host the most efficient AI training clusters. It follows a broader trend where hyperscalers seek out specialized infrastructure developers to meet their specific power and throughput demands.
Users of Meta's AI-driven services will likely experience faster model iteration times as this regional compute capacity comes online. The deployment effectively creates a dedicated backbone for future AI products within the Asia-Pacific region.
The takeaway
The partnership establishes Firmus as a key supplier for Meta's expanding computational needs in the Asia-Pacific region. Watch for future disclosures on site capacity metrics to gauge the scaling speed of these AI factories.
Further reading
For more on the current expansion of global AI compute capacity, visit the Data Centers section.
Live Poll
Do you trust that large-scale infrastructure partnerships between major tech companies benefit the general public?
