Core42 expands US footprint as UAE pursues AI ambitions

• Tripling size of Buffalo data centre
• Adding Nvidia infrastructure
• Customer requirements changing

A UAE artificial intelligence infrastructure company is tripling the size of a major US data centre as the Gulf country races to secure the computing power driving the global AI boom.

Core42, a provider of sovereign cloud and advanced computing services, is expanding its Lake Mariner facility in Buffalo, New York, from 18 megawatts to 60MW.

The additional capacity will support thousands of chips used to train and operate large language models and other AI applications.

The upgrade will add Nvidia infrastructure alongside the site’s existing AMD-powered systems, giving customers access to two of the dominant chip platforms in the rapidly growing sector.

Core42 said it operates 10 AI sites worldwide, representing 25MW of deployed computing capacity. The company expects that figure to exceed 100MW in 2026, as it accelerates its expansion across the US and Europe.

Sherif Tawfik, chief business officer at Core42, said international growth was closely linked to the UAE’s ambitions in artificial intelligence.

“Overseas facilities are not separate from our UAE AI ambitions – they are a key part of enabling them,” he told AGBI.

“As the UAE positions itself as a global AI hub, the underlying infrastructure has to be global and scalable to serve customers across jurisdictions.”

Core42 has emerged as one of the UAE’s most prominent AI infrastructure companies, supplying cloud, data and high-performance computing services for governments and enterprises. Its parent, G42, is one of the UAE’s largest technology groups.

The business has been at the centre of G42’s expansion and recently secured $550 million in trade finance facilities from HSBC, including one to the tune of $310 million announced in May. The funding is intended to accelerate its global growth as demand for AI services surges.

The Buffalo site already hosts “Maximus”, a computing system powered by US semiconductor company AMD, which ranks among the world’s 20 most powerful systems in the TOP500 rankings.

Lake Mariner forms part of a network spanning the US, Europe and the Middle East. Core42 already operates facilities in Dallas, Sunnyvale, Stockton and Minneapolis, including the Condor Galaxy supercomputers developed with US chipmaker Cerebras.

In Europe, the company has headquarters in Dublin and expanded operations into Italy and France.

Countries and technology groups are competing for access to the processors, data centres and energy needed to develop advanced AI systems.

Access to large-scale computing resources is viewed as one of the most important battlegrounds in AI, alongside talent, capital and semiconductor supply chains.

Further reading:

• Anthropic IPO filing lifts Gulf appetite for US AI
• UAE’s Core42 raises $550m to scale AI deployment in Europe and US
• Talent squeeze and ‘death by pilot’ – UAE’s big AI challenges

The world’s largest technology companies are spending tens of billions of dollars building AI data centres, with access to Nvidia’s highly sought-after processors emerging as one of the industry’s key competitive advantages.

UAE-backed groups including MGX, Mubadala and ADQ have invested heavily in AI services, data centres, semiconductors and digital infrastructure as the country seeks to establish itself as a global industry leader.

Earlier this year, G42 signed an agreement to develop hyperscale data centres and AI infrastructure in Vietnam and announced plans to deploy an AI system in India with Cerebras, highlighting the growing international footprint of UAE-backed technology projects.

Tawfik said customer requirements were changing rapidly as businesses began deploying AI at scale.

“As organisations move beyond experimentation and begin deploying AI across enterprise and national-scale operations, they require access to high-performance, scalable compute capacity,” he said.

Customers were increasingly focused on resilience, governance, performance and cost efficiency as AI moved from experimentation to production deployment, he added.

“Disciplined scaling will be one of the defining factors for AI infrastructure providers over the next few years.”