Abu Dhabi-based technology group G42 has significantly expanded its digital infrastructure presence in the United States through its subsidiary, Core42. The company has signed a lease as the primary tenant for a 20-megawatt data center conversion in downtown Minneapolis. This move underscores a broader strategic commitment to building high-performance AI infrastructure across America, driven by a unique bilateral agreement between the US and the UAE.
A Strategic Lease in Minneapolis
The new facility is located at 1001 Third Avenue South, a six-story building that has been transformed from office space into a modern data center. The conversion was executed by Virginia-based developer Legacy Investing, which spent more than $70 million on the project.
Key details of the Minneapolis expansion include:
- Capacity: Core42 has secured 20 megawatts (MW) of power capacity, bringing the total facility capacity to 21 MW.
- Ownership Transition: In January 2026, the property was acquired by a joint venture between Cloud Capital and Bahrain-based asset manager Arcapita for $235 million.
- Future Growth: The new owners plan to expand the site’s capacity further, from 21 MW to 31 MW.
- Tenant Profile: Arcapita described the tenant as a “leading provider of sovereign AI and cloud inferencing solutions,” referring to Core42’s role in providing secure, government-grade AI capabilities.
This lease is not merely a commercial real estate transaction; it is a physical manifestation of the US-UAE AI Acceleration Partnership. Under this government-to-government agreement, the UAE commits to investing one dollar in US AI infrastructure for every dollar it invests in Middle East data centers built using American technology. This 1:1 investment ratio ensures that Core42’s global expansion directly supports the US economy and technological ecosystem.
Building a Nationwide AI Network
Core42’s Minneapolis lease is part of a rapidly growing network of AI supercomputing resources across the United States. The company is positioning itself to compete with major hyperscalers by combining leased infrastructure with proprietary high-performance computing clusters.
The Condor Galaxy Network
Core42 has partnered with Sunnyvale-based AI chip maker Cerebras Systems to build the Condor Galaxy network. This distributed supercomputing system delivers 20 exaFLOPs of AI compute power. Facilities powering this network are located in:
– California
– Texas
– Minnesota (including the new Minneapolis site)
The Maximus-01 Supercomputer
In November 2025, Core42 commissioned Maximus-01, a landmark addition to its US infrastructure. Located at TeraWulf’s Lake Mariner data center campus in Buffalo, New York, the system is ranked as the 20th most powerful supercomputer in the world according to the Top500 list.
Technical specifications of Maximus-01:
– Hardware: Powered by more than 9,000 AMD Instinct MI300X GPUs.
– Performance: Delivers 114.50 petaflops of performance, optimized for large-scale AI training and inference.
– Networking: Utilizes Broadcom’s high-bandwidth networking chips and Arista Networks’ Ethernet switches, highlighting a deep reliance on American technology partnerships.
– Lease Terms: Core42 signed a ten-year lease with TeraWulf in 2024 for over 70 MW of infrastructure, with an option to expand by an additional 135 MW. The facility features direct liquid-cooled Dell PowerEdge GPU servers.
Why This Matters: Sovereign AI and Global Competition
The expansion of Core42’s US footprint raises important questions about the future of global AI infrastructure. By securing massive amounts of compute power in the US, G42 is not just leasing space; it is establishing itself as a sovereign AI provider. This means it offers secure, independent AI capabilities to governments and enterprises that may be hesitant to rely solely on major public cloud providers.
The use of American technology partners—such as AMD, Broadcom, Arista, and Dell—in these deployments aligns with the geopolitical goals of the US-UAE partnership. It ensures that while the capital and some operational expertise come from the Middle East, the technological backbone remains firmly rooted in the US industrial base.
Furthermore, Core42’s presence is not limited to hardware. The company opened an engineering office in San Francisco in 2024 and currently employs more than 30 people across its US operations, with continued hiring. This combination of physical infrastructure, high-performance supercomputers, and local engineering talent signals a long-term commitment to competing in the American market.
Conclusion: Core42’s expansion into Minneapolis, alongside its powerful supercomputing initiatives in New York and other states, demonstrates a strategic shift toward becoming a global leader in sovereign AI infrastructure. By leveraging the US-UAE AI Acceleration Partnership, the company is bridging geopolitical interests with technological ambition, creating a robust, American-built foundation for its global AI services.
