NVIDIA and IREN Announce 5GW AI Infrastructure Partnership

NVIDIA and IREN Limited announced a strategic partnership to accelerate deployment of next-generation AI infrastructure, with an intended scale of up to 5 gigawatts of NVIDIA DSX-aligned capacity across IREN's global data center pipeline, according to NVIDIA's press release and IREN statements. The announcement includes a five-year right for NVIDIA to purchase up to 30 million ordinary shares of IREN at an exercise price of $70 per share, creating a potential $2.1 billion investment pathway, as reported by NVIDIA and Bloomberg. The companies identified IREN's 2-gigawatt Sweetwater campus in Texas as a planned flagship site for NVIDIA's DSX AI factory deployments, per the joint announcement. Separately, IREN issued a GlobeNewswire release describing a five-year, approximately $3.4 billion managed GPU cloud services contract with NVIDIA to deploy air-cooled Blackwell systems within roughly 60MW at IREN's Childress, Texas campus.

Editorial analysis - technical context: NVIDIA's DSX architecture is presented as a reference design that integrates compute, networking, software, power, and operations into AI factory deployments, per NVIDIA's materials. Industry reporting and the companies' statements emphasize integration across the full stack rather than single-component scale outs, a pattern that aims to reduce friction when moving from bare metal GPUs to managed AI cloud services. Data Center Knowledge and StorageReview note the Sweetwater energization and IREN's recent cloud and operations acquisitions, which indicate a focus on combining land, renewable-rich power, data center development, and cloud orchestration into a vertically integrated offering.

Partnerships tying major accelerator vendors to data center developers at multi-gigawatt scale have become a central mechanism for increasing accessible AI compute capacity. Observers quoted in Bloomberg and Yahoo Finance framed the deal as part of an industry-wide push to lock in supply, campus-level power, and integrated operations as demand for large training workloads grows. The inclusion of an equity purchase right and a five-year managed-services contract is consistent with recent vendor-developer arrangements that blend capital exposure, long-term demand signals, and operational partnerships. For practitioners, this trend affects capacity planning, procurement lead times, and vendor selection when projecting availability of large-scale GPU clusters.

For practitioners: Expect enterprise and research teams that require hyperscale training cycles to monitor vendor-composed AI factory offerings rather than only hyperscaler bare-metal options. Industry reporting highlights that IREN's acquisitions of Mirantis and Nostrum Group and the energization of initial Sweetwater phases add Kubernetes orchestration and regional power capacity to the deployed stack, which matters for teams evaluating managed GPU availability, tenancy models, and integration overhead.

Observers should track the following indicators to gauge real-world impact and timeline.

• •Deployment progress at IREN's Sweetwater campus, including published energization milestones and MW staged for service, as reported by IREN or independent data center trackers.
• •Regulatory clearances or filings related to NVIDIA's five-year purchase right and any subsequent exercises, as reported by NVIDIA, IREN, or securities filings.
• •Commercial terms and SLAs for the IREN-managed GPU cloud services contract, especially availability of air-cooled Blackwell systems and orchestration stacks, per IREN releases and customer announcements.
• •Broader market response from other infrastructure providers and third-party cloud GPU operators, including pricing and capacity announcements cited in trade press.

Editorial analysis: The partnership exemplifies a convergence of accelerator vendors, campus-scale power planning, and cloud orchestration that the industry has pursued to shorten lead times for hyperscale AI capacity. Companies assembling land, power, cooling, GPUs, and software operations at the campus level are likely to shape where large training clusters appear and how practitioners access them. This announcement is a clear data point in that ongoing shift but does not by itself guarantee timelines or contracted customer access; readers should rely on subsequent operational milestones and filings for confirmation.