The beginning of April 2026 paints a very clear picture of a new investment cycle in data centers across the Americas: artificial intelligence is no longer just driving targeted expansions but is increasingly behind entire campuses measured in hundreds of megawatts, long-term energy agreements, and projects blending computing, power generation, industrial land, and geopolitics. The United States continues to lead with the largest-scale announcements, but Mexico, Brazil, Peru, Argentina, and Canada are beginning to emerge more prominently in the conversation.
The list of projects announced or under development this month shows two speeds. On one side are traditional or colocation data centers, like KIO’s MEX8 in Mexico City or Gtd Peru’s campus in Lurín. On the other are facilities designed specifically for AI, with figures that a few years ago would have seemed disproportionate: 600 MW in West Virginia, 600 MW in Georgia, 2 GW in Texas, 10 GW in Ohio, or even Latin American projects focused on renewable energy and major tech clients.
This scale shift is no coincidence. Generative AI and new agent-based models have altered the sector’s economics. It’s no longer just about hosting servers for cloud, e-commerce, or streaming. Major operators now seek secured power, advanced cooling, proximity to fiber networks, the ability to install large-scale GPUs, and agreements that allow growth for years without needing a new interconnection each time market demand increases capacity.
The U.S. leads the mega projects in AI
The most eye-catching announcement this month was the PORTS Technology Campus in Piketon, Ohio, driven by the U.S. Department of Energy along with SoftBank Group and AEP Ohio. The project proposes up to 10 GW of AI-focused development on the former federal facility in Portsmouth, supported by new power generation. According to the Department of Energy, SB Energy, a SoftBank subsidiary, plans to build 10 GW of new generation capacity, including 9.2 GW of natural gas, to power the campus and connect capacity to the local grid.
The scale of this project places it in a different league. A 10 GW campus isn’t just a data center expansion; it’s an industrial policy piece aimed at maintaining U.S. leadership in AI, creating jobs, reusing federal land, and strengthening national tech capabilities against China. It also raises an uncomfortable conversation: much of the energy comes from natural gas, bringing to the surface ongoing tensions around AI leadership, electric costs, emissions, and local acceptance.
Texas continues to emerge as a key hub for new infrastructure. Poolside and CoreWeave are progressing with Project Horizon near Fort Stockton, a 2 GW AI campus combining energy and compute capacity for large-scale training loads. Also notable is the Fluor/TeraWulf campus in Kentucky, with up to 480 MW, focused on HPC and AI with direct access to power grids.
Additional projects include Penzance Management’s campus in Falling Waters, West Virginia, with 600 MW and an estimated investment of $4 billion; Project Ruby in Columbus, Georgia, led by Atlas Development, with 600 MW and over $5 billion; and hyperscale developments by Carlyle and CyrusOne in Fort Bliss, Texas, and Dugway Proving Ground, Utah, under military lease models.
There are also capacity commitments already in motion. NTT DATA, Inc. announced an additional 115 MW across data centers in Gainesville, Chicago, and Sacramento, while Aligned Data Centers secured $2.58 billion for expansion in Dallas, Phoenix, and northern Virginia. The market’s message is clear: while traditional hubs remain important, growth is shifting toward territories with land, energy, and capacity to absorb massive loads.
Latin America gains influence but with different models
In Latin America, Brazil and Mexico are solidifying as the most active markets. KIO has begun building MEX8 in Mexico City, a 4 MW data center in Santa Fe, with a $70 million investment and scheduled to open by early 2027. Though small compared to U.S. AI campuses, it is significant due to proximity to corporate clients, low latency, and strengthening of Mexico’s digital infrastructure.
Brazil is taking a different path. OMNIA Data Centers, in partnership with Casa dos Ventos Energias Renováveis, is developing an AI megacenter at the Port of Pecém, Ceará, linked to ByteDance, with approximately 200 MW powered by renewable energy. Ascenty continues developing multiple campuses in São Paulo and Campinas, while ODATA, now part of Aligned Data Centers, advances hyperscale phases in São Paulo and Rio de Janeiro for large cloud providers.
Brazil also features several projects announced after March that are worth noting. I Squared Capital acquired a majority stake in Elea Data Centers in late April to accelerate expansion. Elea operates nine campuses with over 300 MW capacity and more than 1 GW in development, including Rio AI City— a project with 1.5 GW of certified renewable energy in its first phase, scalable up to 3.2 GW.
Scala Data Centers is also moving forward with Scala AI City, a large-scale campus focused on AI. Recent reports indicate an initial investment of around 3 billion reais, with capacity possibly reaching 4.75 GW. If successful, Brazil could become one of the major AI infrastructure hubs outside the U.S., supported by renewable energy, ample land, and regional cloud demand.
Peru features a smaller scale but shows growth signs. Gtd Peru is expanding its Lurín data center in Lima, with an initial 2 MW and a $13 million investment, aiming for 20 MW. Argentina presents an even more uncertain yet ambitious project: Sur Energy plans an AI data center in Patagonia with up to 500 MW and a potential $25 billion investment, possibly in Vaca Muerta or Neuquén, location still unconfirmed.
Energy, land, and regulation: the new bottlenecks
The common denominator across these announcements is energy. The discussion around data centers now isn’t limited to square meters, certifications, or connectivity but primarily revolves around electricity: available megawatts, connection timelines, who pays for the grid, generation sources backing the projects, and how costs are shared among operators, utilities, and consumers.
Canada exemplifies this trend. Bell Canada announced a 300 MW campus in Sherwood, near Regina, Saskatchewan, with a $1.7 billion CAD investment focused on AI. It is part of Bell AI Fabric, with CoreWeave and Cerebras already lined up as clients. Also in Canada, Jet.AI and Consensus Core are developing two hyperscale campuses in Manitoba, including a site near Winnipeg with a mix of grid connection and self-generation.
Other recent notable agreements include TransAlta, CPP Investments, and Brookfield exploring a project in Parkland County, Alberta, with an initial PPA of 230 MW, scalable up to 1 GW. Also worth mentioning is the expansion of Bell and Buzz HPC in Merritt, British Columbia, with an initial 6.5 MW dedicated to GPU clusters within Bell AI Fabric ecosystem.
In the U.S., one of the most significant post-March developments is Hut 8’s Texas deal. In May, it signed a 15-year lease worth $9.8 billion for its Beacon Point campus in Nueces County. The first phase offers 352 MW of capacity for an undisclosed client and is part of a planned 1 GW campus, with electricity connection expected by 2027. Such contracts show the market’s move toward securing capacity years in advance.
An important recent approval is the Joliet Technology Center in Illinois, a 795-acre campus developed by PowerHouse Data Centers and Hillwood, with up to 1.8 GW at full capacity and an investment estimated around $20 billion. Also notable is the $16 billion financing for the Related Digital campus in Saline Township, Michigan, designed for Oracle and linked to the Stargate ecosystem, with more than 1 GW capacity.
Regulatory and social pressures will intensify. Local communities are starting to ask what they receive in exchange for large power consumption, water use, noise, landscape impact, and infrastructure strain. In some U.S. states, moratoriums, new electric tariffs for large loads, and debates about whether data centers should bear the full cost of power generation and grid infrastructure are already underway.
The conclusion from March, reinforced by April and May’s news, is that America is entering an accelerated phase of AI infrastructure development. Not all projects will finish on time; some will be scaled down, others depend on pending energy agreements, and some will face local opposition. But the market’s direction seems clear: competitive advantage will no longer just depend on GPUs but on securing energy, land, permits, connectivity, and a supply chain capable of building at industrial scale.
Key projects and announcements in the Americas
| Project / Operator | Country | Capacity or Investment | Status or Focus |
|---|---|---|---|
| PORTS Technology Campus, SoftBank / AEP / DOE | USA | Up to 10 GW | AI campus in Piketon, Ohio, with new power generation |
| Project Horizon, Poolside / CoreWeave | USA | 2 GW | AI campus near Fort Stockton, Texas |
| Joliet Technology Center, PowerHouse / Hillwood | USA | Up to 1.8 GW | Approved campus near Chicago |
| Beacon Point, Hut 8 | USA | Initial 352 MW, total 1 GW | Lease contract of $9.8B in Texas |
| Penzance, Falling Waters | USA | 600 MW, $4B | AI campus in West Virginia |
| Project Ruby, Columbus | USA | 600 MW, over $5B | Hyperscale campus in Georgia |
| Fluor / TeraWulf, Kentucky | USA | Up to 480 MW | HPC and AI with direct grid access |
| Bell AI Fabric, Saskatchewan | Canada | 300 MW, $1.7B CAD | AI campus near Regina |
| TransAlta / CPP / Brookfield, Alberta | Canada | Initial 230 MW, up to 1 GW | Potential project with PPA |
| KIO MEX8 | Mexico | 4 MW, $70M | New data center in Mexico City |
| OMNIA / Casa dos Ventos / ByteDance | Brazil | Approx. 200 MW | AI megacenter with renewables at Pecém |
| Elea Rio AI City | Brazil | 1.5 GW initial, up to 3.2 GW | AI campus with renewable energy |
| Scala AI City | Brazil | Up to 4.75 GW | AI development campus in progress |
| Gtd Perú, Lurín | Peru | Initial 2 MW, up to 20 MW | Data center in Lima |
| Sur Energy, Patagonia | Argentina | Up to 500 MW, potential $25B | AI project with unconfirmed location |
Frequently Asked Questions
Which country leads new data center projects in the Americas?
The United States dominates capacity announcements, especially in AI, with projects reaching or exceeding 1 GW across campuses and hundreds of megawatts.
Why is Brazil appearing more frequently on the data center map?
Due to its market size, renewable energy availability, regional cloud demand, and large-scale projects like Rio AI City or Scala AI City.
What is the main bottleneck for these projects?
Energy. Developers need available power, grid connection, associated generation, and long-term agreements to ensure stable operation.
Will all these projects be built?
Not necessarily. Many depend on permits, funding, end clients, electrical connection, and local acceptance. The list reflects market ambition, not guaranteed execution.