Blue Owl Capital has committed $2.4 billion to help IREN accelerate its next phase of AI infrastructure deployment, putting equipment financing at the center of a large-scale compute expansion. The financing gives IREN capital specifically tied to NVIDIA accelerated computing infrastructure, including Blackwell Ultra GPUs, destined for its Mackenzie data center campus in British Columbia, Canada. Blue Owl announced the transaction on August 28, positioning the deal as a major asset-backed financing structure for AI compute rather than a conventional data center funding round. The deal arrives as AI infrastructure operators seek financing models that can match the speed at which GPUs move from purchase orders into revenue-generating compute capacity.

Blue Owl Structures $2.4 Billion Compute Financing

The $2.4 billion facility splits evenly between a $1.2 billion senior secured term loan and $1.2 billion of senior secured notes. IREN will use the proceeds to purchase air-cooled NVIDIA Accelerated Computing Infrastructure, with NVIDIA Blackwell Ultra GPUs forming a central part of the deployment at Mackenzie. The financing structure matters because Blue Owl will release capital through tranches during a defined draw period rather than putting the entire amount into the project at once. That approach connects funding with hardware delivery and commissioning, giving IREN a financing mechanism that tracks the physical rollout of its compute fleet.

IREN is building more than a GPU cluster at Mackenzie, as the company continues to position itself as a vertically integrated AI Cloud platform covering data centers, compute and software for AI training and inference. Its broader development strategy includes a global data center pipeline exceeding 5GW, giving the financing a much larger context than a single equipment purchase. The Mackenzie deployment represents another step in converting that development pipeline into operational AI capacity that customers can access. The transaction therefore links financial capital, GPU procurement and infrastructure deployment within one financing framework.

IREN Targets Growing AI Training Demand

Anthony Lewis, IREN’s chief financial officer, framed the financing around the company’s need to keep pace with customer demand for compute capacity. “Customer demand for AI compute is accelerating, and we are scaling rapidly to meet it. This financing supports the continued build-out of our AI Cloud infrastructure at Mackenzie, expanding our capacity to serve a growing customer base across both AI training and inference workloads,” His statement points to a business model that depends on bringing additional compute online quickly enough to capture demand from customers running increasingly intensive workloads. The financing gives IREN another mechanism to expand capacity without treating GPU procurement as a one-off capital expenditure decision.

The Mackenzie project sits inside IREN’s broader push to build infrastructure around both training and inference. Training workloads can demand large pools of accelerated computing for model development, whereas inference creates a different utilization profile as deployed models serve production applications. That distinction makes the hardware deployment commercially important because IREN can potentially support multiple workload categories from the same underlying NVIDIA infrastructure. The company can therefore pursue a broader customer mix as additional capacity comes online.

NVIDIA Positions GPUs As Investable Infrastructure

NVIDIA’s view of the transaction extends beyond IREN and Blue Owl, focusing on whether GPU infrastructure can become a repeatable financing asset class. Nico Caprez, NVIDIA’s vice president for global AI infrastructure growth, argued that CUDA creates flexibility across workloads and customers, which can support longer-duration infrastructure investment. “Compute Unified Device Architecture (CUDA) makes NVIDIA AI factories fungible across any customers and workloads—and better over time. That makes NVIDIA AI infrastructure an investable infrastructure asset class for long-term capital. Blue Owl’s $2.4 billion asset-backed GPU financing for IREN is another great proof point—and a repeatable model for financing AI infrastructure at scale,” said Nico Caprez, Vice President, Global AI Infrastructure Growth at NVIDIA. The argument effectively treats the GPU stack as an infrastructure asset whose utility can extend across different customers rather than as hardware tied to a single application.

That framing could matter as AI infrastructure moves deeper into capital markets. Traditional data center financing has typically centered on buildings, power systems, land and long-lived physical infrastructure, but accelerated computing introduces a different class of equipment with a much faster technology cycle. Blue Owl’s structure places GPUs directly inside a secured financing framework, creating a potential bridge between private credit and the rapidly expanding AI compute economy. Meanwhile, NVIDIA’s emphasis on CUDA interoperability strengthens the case that the equipment can retain operational relevance across different AI workloads.

Blue Owl Brings Data Center Experience

Blue Owl is approaching the transaction through its digital infrastructure investment experience, rather than treating the financing as a conventional corporate debt deal. Kurt Tenenbaum, senior managing director at Blue Owl, emphasized the operational complexity that comes with financing AI equipment at this scale. “This is equipment finance at AI scale, and it demands the same discipline,” “Our investment track record, combined with operating experience across more than 100 data centers worldwide, lets us tailor this facility to how AI Infrastructure is actually delivered and deployed. That’s what makes us a partner of choice for the AI compute ecosystem, while providing our investors with strong risk-adjusted returns.”

The scale of the facility illustrates how GPU procurement is becoming inseparable from infrastructure finance. IREN needs capital to acquire compute hardware, but the lender needs a structure that reflects delivery schedules, deployment conditions and the underlying value of the equipment. The tranche-based arrangement addresses that timing by allowing funding to move alongside hardware delivery and commissioning rather than relying solely on a traditional construction draw model. As a result, the transaction creates a financing template that could become relevant to other AI infrastructure developers pursuing large GPU deployments.

AI Equipment Financing Moves Into The Mainstream

The IREN transaction highlights a broader change in how the AI infrastructure market can finance growth. GPU fleets increasingly represent a substantial portion of the capital required to bring AI capacity online, making equipment financing a strategic tool rather than a secondary funding option. For infrastructure operators, the ability to finance hardware in stages can reduce the gap between procurement commitments and deployed compute capacity. For capital providers, the underlying equipment offers a tangible asset base within an industry experiencing sustained demand for accelerated computing.

IREN’s more than 5GW development pipeline gives the transaction additional significance because the company’s financing needs can extend well beyond Mackenzie. The immediate facility focuses on NVIDIA infrastructure for the British Columbia campus, but the underlying model could support future deployments if lenders become comfortable underwriting GPU-backed infrastructure at scale. Blue Owl’s participation signals that private capital is increasingly willing to engage directly with the hardware layer of the AI economy. The result is a closer relationship between GPU manufacturers, compute operators and financial institutions as AI capacity becomes an infrastructure market of its own.