G42's Core42 Triples US AI Compute Capacity to 60MW as Abu Dhabi Scales Its Global Infrastructure Footprint

Core42, the sovereign cloud and AI infrastructure company owned by Abu Dhabi's G42 Group, has announced a 42MW expansion of its AI compute facility at Lake Mariner in Buffalo, New York, bringing total site capacity from 18MW to 60MW of high-performance AI production infrastructure. The move significantly extends G42's footprint in the United States and reinforces the group's strategy of building globally distributed, heterogeneous AI infrastructure at production scale.

The expansion adds both NVIDIA and AMD capacity to the existing AMD-based deployment at the site. The existing cluster, known as Maximus and built on AMD Instinct MI300 accelerators, achieved a Top-20 ranking on the global TOP500 supercomputing list, validating Core42's ability to deliver performance at the scale required by frontier training and inference workloads. Adding NVIDIA infrastructure alongside the existing AMD foundation creates a heterogeneous architecture that enables workload optimisation across multiple accelerator platforms, a design approach that gives operators greater flexibility when supporting diverse enterprise and hyperscale requirements.

A growing US and global footprint

Lake Mariner is Core42's primary North American AI infrastructure hub, but it is far from the company's only US presence. The group also operates deployments in Dallas, Texas, Sunnyvale and Stockton in California, and Minneapolis, Minnesota, the latter including the Condor Galaxy supercomputers developed in collaboration with Cerebras. Together, these sites form a distributed AI architecture across the United States designed to support both frontier model training and high-speed inference at production scale.

Internationally, Core42 established its European headquarters in Dublin in 2025 and has expanded AI compute deployments across Italy and France. The company's AI Cloud platform, launched in October 2025, serves as the unified access layer across this global infrastructure, allowing customers to provision compute across jurisdictions under a consistent operating model. With 10 operational sites globally and additional deployments planned through 2026, Core42 is positioning itself as a genuine alternative to hyperscale cloud providers for organisations with sovereign AI or performance requirements that standard public cloud infrastructure cannot meet.

What this means for the GCC AI agenda

The Lake Mariner expansion carries significance beyond the headline capacity number. It reflects the broader ambition of Abu Dhabi and the wider Gulf to position sovereign AI infrastructure not just as a domestic capability, but as a globally competitive offering. G42 is increasingly operating as an international AI infrastructure player rather than a regional technology group, competing for hyperscale and enterprise AI contracts in the United States, Europe and beyond.

For Saudi Arabia and other GCC states that are building out their own national AI infrastructure programmes, the Core42 model provides a reference point. The combination of heterogeneous hardware, globally distributed capacity and a unified platform layer is the architecture that serious AI infrastructure at national scale requires. Core42's ability to achieve a TOP500 ranking from an Abu Dhabi-headquartered company signals that Gulf-origin AI infrastructure is now operating at the performance frontier, not merely catching up to it.

Talal M. Al Kaissi, Chief Executive Officer of Core42, said the expansion was part of long-term deployment programmes aligned to serving hyperscale, AI-native and large enterprise workloads, and that it further extended the global build-out of the company's AI infrastructure.