NVIDIA IREN AI infrastructure partnership targets up to 5GW
NVIDIA and IREN announced a strategic plan to deploy up to 5 gigawatts of AI infrastructure, with IREN positioned to host NVIDIA-powered AI factories at global scale. Central to the NVIDIA IREN AI infrastructure partnership is NVIDIA’s DSX AI factory architecture and full-stack systems, paired with IREN’s land, power, development, and operations capabilities. The companies frame the effort as a path to standardize and accelerate large-scale GPU data centers for enterprise training and inference workloads [1][2].
Why the NVIDIA IREN AI infrastructure partnership matters
The headline figure is scale: up to 5 GW of capacity across sites designed around DSX. The flagship is IREN’s Sweetwater campus in Texas, planned as a 2 GW DSX AI factory that will serve as a reference for subsequent deployments. The goal is to expand access to high-performance compute for AI-native startups, enterprises, and public sector users by speeding buildout and creating consistent blueprints for data center design and operations [1][2].
What DSX brings to large-scale deployments
NVIDIA describes DSX as an AI factory architecture that brings together reference designs, systems, and software into a standardized model for building and running compute at scale. In practice, this aims to streamline how providers stand up and operate infrastructure for training and inference, while aligning facilities to the performance and reliability characteristics of NVIDIA’s current and future platforms [2]. For decision-makers, DSX signals a maturing pattern for deploying and managing large AI estates, reducing integration friction and accelerating time to capacity [2].
IREN’s role: vertically integrated AI cloud and operations
IREN operates as a vertically integrated AI cloud provider and data center developer, with a track record running large-scale GPU fleets, predominantly NVIDIA systems, to support frontier and enterprise AI workloads. The company has already deployed tens of thousands of NVIDIA GPUs, including next-generation Blackwell architecture systems, and focuses on delivering capacity at high density for training and inference [1][3].
On the infrastructure side, IREN’s sites are backed by grid-connected power in renewable-rich regions across North America, Europe, and Asia-Pacific. That footprint aligns AI compute expansion to substantial, often lower-carbon electricity and supports the power availability needed for rapid scaling [1].
Flagship site: Sweetwater, Texas
IREN’s Sweetwater AI campus in Texas is planned as a 2 GW DSX AI factory and will function as the model site for subsequent builds. The campus is intended to showcase the DSX reference approach, from systems integration to operations, providing a template for follow-on deployments in other regions [1][2]. For enterprises evaluating capacity, Sweetwater signals what a standardized, high-density site could offer in performance, scalability, and procurement consistency [1][2].
Business implications for enterprise compute access
For buyers, the partnership indicates more standardized on-ramps to NVIDIA-powered capacity via IREN’s AI cloud and hosted infrastructure. Enterprises, AI-native startups, and public sector users could benefit from predictable architectures, access to large-scale GPU clusters, and a consistent operating model across sites. Sourcing decisions will still hinge on workload profiles, latency needs, SLAs, data governance, and budget fit, but standardized DSX designs may simplify evaluations and speed time to deployment [1][2]. For practical frameworks on capacity planning and vendor selection, explore AI tools and playbooks.
Power, sustainability, and site selection
The plan emphasizes grid-connected power in renewable-rich regions, which is relevant for both feasibility and sustainability reporting. Access to substantial, often lower-carbon electricity can support corporate emissions goals while addressing the power intensity of training and inference at scale. Site selection will continue to balance grid constraints, interconnection timelines, and local regulatory factors that shape build speed and operating costs [1].
Risks, timelines, and the broader partner ecosystem
Both companies characterize the scale, timing, and impact of these AI factories as forward-looking. Key risks include uncertainty in demand for AI compute, infrastructure buildout challenges, regulatory and power constraints, and the performance and availability of NVIDIA’s current and future platforms [1][2].
The effort fits NVIDIA’s broader strategy of enabling an ecosystem of AI infrastructure partners, with NVIDIA providing reference architectures, systems, software, and AI factory designs, and partners supplying and operating the physical infrastructure. Recent announcements highlight similar collaborations focused on scaling full-stack AI cloud capacity [4][5]. For official details, see NVIDIA’s newsroom announcement [2] or the NVIDIA announcement (external) [2].
What to watch next
In the near term, track Sweetwater’s execution as the 2 GW reference site and how quickly additional DSX-based campuses come online. For enterprises, the NVIDIA IREN AI infrastructure partnership could mean faster access to standardized capacity and clearer procurement paths as large-scale GPU data centers proliferate. For investors and operators, watch power contracting in renewable-rich regions and the pace of interconnections that govern how quickly 5 GW of planned capacity translates into live compute [1][2].
Sources
[1] NVIDIA and IREN Announce Strategic Partnership to Accelerate Deployment of up to 5 Gigawatts of AI Infrastructure
https://www.globenewswire.com/news-release/2026/05/07/3290674/0/en/nvidia-and-iren-announce-strategic-partnership-to-accelerate-deployment-of-up-to-5-gigawatts-of-ai-infrastructure.html
[2] NVIDIA and IREN Announce Strategic Partnership to Accelerate Deployment of up to 5 Gigawatts of AI Infrastructure
https://nvidianews.nvidia.com/news/nvidia-and-iren-announce-strategic-partnership-to-accelerate-deployment-of-up-to-5-gigawatts-of-ai-infrastructure
[3] IREN AI Cloud TM Strengthens Leadership with NVIDIA Blackwell Deployment
https://iren.com/resources/blog/iren-ai-cloudstrengthens-leadership-with-nvidiablackwell-deployment
[4] NVIDIA and Partners Build America’s AI Infrastructure and Create …
https://nvidianews.nvidia.com/news/nvidia-partners-ai-infrastructure-america
[5] NVIDIA and Nebius Partner to Scale Full-Stack AI Cloud
http://nvidianews.nvidia.com/news/nvidia-and-nebius-partner-to-scale-full-stack-ai-cloud
