Core42 Expands U.S. AI Push with AMD and Nvidia Infrastructure in New York

Core42 has expanded its artificial intelligence infrastructure at the Lake Mariner site in New York, lifting the facility’s total capacity from 18MW to 60MW as the Abu Dhabi-based company scales its U.S. footprint.

The 42MW expansion adds new AMD and NVIDIA infrastructure to the site, strengthening Core42’s ability to support high-performance AI workloads across training, fine-tuning and inference. The company, part of the UAE’s G42 group, described the New York deployment as a key North American hub within its global AI infrastructure network.

The New York buildout comes as demand for data center capacity, graphics processing units and sovereign AI infrastructure continues to intensify across enterprise and government markets. AI companies and cloud providers are under pressure to secure power, cooling, chips and long-term hosting capacity as large models become more expensive to train and deploy.

Core42’s U.S. Footprint Expands

Lake Mariner is part of Core42’s broader U.S. infrastructure buildout, which the company says includes deployments in New York, Dallas, Sunnyvale, Stockton and Minneapolis. The network also includes Condor Galaxy supercomputers developed with Cerebras, giving Core42 a multi-site presence across the American AI compute market.

The New York expansion is notable because it increases capacity at a site already positioned around high-performance workloads. Core42 said the additional AMD and NVIDIA infrastructure supports a heterogeneous architecture, allowing workloads to run across multiple accelerator platforms rather than relying on a single chip provider.

That approach reflects a broader shift in AI infrastructure strategy. Large enterprises are increasingly looking for flexibility across GPUs and specialized accelerators, particularly as demand for NVIDIA systems remains high and alternative platforms from AMD, Cerebras and others become more relevant to cost, availability and workload optimization.

Lake Mariner Becomes Core42’s AI Cloud Hub

Core42’s Lake Mariner deployment is also tied to the company’s AI Cloud strategy. In October 2025, Core42 launched a self-service AI Cloud platform with NVIDIA accelerated computing, designed to give customers access to compute for training, fine-tuning and real-time inference. The Lake Mariner expansion gives that broader platform more U.S.-based capacity.

The company has framed AI Cloud as an access layer for distributed infrastructure across jurisdictions. In practical terms, that means customers can provision compute across different regions while operating under a consistent model for security, governance and deployment.

Core42’s messaging around “sovereign” infrastructure is central to that strategy. Governments and regulated industries are adopting AI at a faster pace, creating demand for cloud and compute systems that address data residency, jurisdictional control and compliance requirements. The New York expansion adds U.S. capacity to that international model, while Core42 continues to build out deployments in Europe and the Middle East.

TeraWulf Agreement Backs the New York Site

The Lake Mariner site has been connected to Core42’s U.S. growth since late 2024, when TeraWulf announced long-term data center lease agreements with Core42. Under that agreement, TeraWulf said it would deliver more than 70MW of turnkey data center infrastructure at Lake Mariner in Upstate New York to host Core42’s deployment.

TeraWulf said the infrastructure would be released for production in phases between the first and third quarters of 2025 and customized for Core42’s GPU clusters. The company also said the site would use Dell integrated rack systems and direct liquid-cooled GPU servers.

That earlier agreement provides important context for the latest expansion. Lake Mariner was not a one-off deployment; it formed part of a long-term U.S. infrastructure push by G42 and Core42, with room for additional scaling.

Maximus Ranking Shows Lake Mariner’s Scale

Core42’s Lake Mariner infrastructure has already drawn industry attention through its Maximus supercomputer deployment. The company said the site’s AMD Instinct MI300-based Maximus cluster secured a Top 20 ranking on the global TOP500 supercomputing list.

The TOP500 listing identifies Maximus in Buffalo, United States, and shows the MAXIMUS-384 system ranked No. 20 on the November 2025 list. The system is listed with Dell PowerEdge XE9680 infrastructure, Intel Xeon Platinum processors and AMD Instinct MI300X accelerators.

That ranking gives the expansion a technical benchmark beyond the headline power figure. While megawatts indicate the scale of available power capacity, supercomputing rankings offer a separate view into system performance. Together, they show why Lake Mariner is being positioned as more than a conventional data center site.

AI Compute Becomes a Strategic Asset

Core42’s expansion highlights how AI infrastructure has become strategic real estate for companies, governments and investors. Compute capacity is no longer just a technical resource; it is increasingly tied to national AI ambitions, cloud sovereignty, chip supply, energy availability and geopolitical technology partnerships.

The U.S. buildout gives Core42 access to one of the world’s deepest AI markets while supporting customers that require American-based infrastructure. At the same time, it strengthens G42’s position as a global AI infrastructure operator with capacity distributed across multiple regions.

Talal M. Al Kaissi, chief executive officer of Core42, said: “We are scaling our U.S. infrastructure in line with long-term deployment programs. Increasing our U.S. capacity at Lake Mariner strengthens our ability to serve hyperscale, AI-native and large enterprise workloads, and further extends the build out of our AI infrastructure globally.”