Core42 expands New York AI cluster to 60MW

By AI, Created 10:11 AM UTC, June 02, 2026, /AGP/ – Core42 expanded its Lake Mariner site in Buffalo, New York, from 18MW to 60MW, adding 42MW of U.S. AI compute capacity. The move deepens the G42 company’s U.S. footprint and strengthens its ability to serve hyperscale, enterprise and AI-native workloads.

Why it matters: - Core42 is adding more AI compute capacity in the U.S. at a time when demand for training and inference infrastructure remains high. - The Buffalo expansion reinforces a heterogeneous setup that can run workloads across AMD and NVIDIA platforms. - The buildout also extends Core42’s distributed AI footprint across the U.S., Europe and the Middle East.

What happened: - Core42 expanded its AI cluster at the Lake Mariner site in New York by 42MW. - Total site capacity rose from 18MW to 60MW of high-performance AI production infrastructure. - The expansion increases Core42’s U.S. AI infrastructure at a site in Buffalo. - Talal M. Al Kaissi, Core42’s chief executive officer, said the company is scaling U.S. infrastructure in line with long-term deployment programs.

The details: - The site’s existing AMD foundation now includes additional AMD and NVIDIA capacity. - Core42 said the mixed-accelerator design supports workload optimization across multiple platforms. - The AMD Instinct MI300-based Maximus cluster at Lake Mariner ranked No. 20 on the TOP500 supercomputing list. - Core42 said that ranking validates its ability to deliver high-performance AI infrastructure at global scale. - Core42’s U.S. footprint also includes deployments in Dallas; Sunnyvale and Stockton, California; and Minneapolis. - Those sites include the Condor Galaxy supercomputers, developed with Cerebras. - Core42 said the combined U.S. sites support workload-optimized deployment for frontier training and high-speed inference at production scale. - Core42’s AI Cloud platform, introduced in October 2025, serves as the access layer to the distributed infrastructure. - The platform lets customers provision compute across jurisdictions under a consistent operating model. - The platform supports training, fine-tuning and real-time inference. - Core42 said the platform uses heterogeneous infrastructure with multiple options to help improve price-performance.

Between the lines: - The Lake Mariner expansion signals that Core42 is moving beyond isolated deployments and into a broader multi-site infrastructure strategy. - Adding NVIDIA alongside AMD suggests Core42 wants flexibility to match workloads with the right accelerator rather than rely on a single stack. - The company is also tying the U.S. expansion to international growth, which can matter for customers that want geographically distributed compute options. - Core42 said 2025 included the launch of its European headquarters in Dublin and expanded AI compute deployments across Italy and France.

What’s next: - Core42 said it has 10 operational sites globally. - The company also said additional deployments are planned for 2026. - Core42 said the ongoing buildout is meant to support AI infrastructure at national and enterprise scale.

The bottom line: - Core42 is turning Lake Mariner into a larger U.S. AI hub while broadening a global network built to serve heavyweight AI workloads.