IREN Secures $3.65 Billion GPU Financing to Expand Microsoft-Backed AI Cloud Shift

IREN Limited ($IREN) said it has closed a $3.65 billion, investment-grade GPU financing facility designed to fund a major expansion of its AI cloud operations tied to an agreement with Microsoft ($MSFT), underscoring how former Bitcoin mining operators are increasingly repositioning themselves as power- and infrastructure-backed AI compute providers.

The company, which originated as a Bitcoin (BTC) mining data center operator, framed the financing as a key milestone in executing its AI cloud buildout. Shares of IREN ended the session at $65.33, up 2.82% on the day, as investors digested both the scale of the facility and what it signals about institutional appetite for 'asset-backed' AI infrastructure funding.

IREN described the transaction as the first publicly announced, investment-grade GPU financing in the U.S. private markets, receiving an A rating from Fitch and an A(low) rating from DBRS. The facility is structured as $2.1 billion in U.S. private notes priced at SOFR + 2.13% and a $1.55 billion delayed-draw term loan at SOFR + 2.25%, with interest-rate hedges in place. The company said the GPU assets and contracted cash flows underpin the deal, producing a blended cost of debt of about 6.00%.

By combining Microsoft’s upfront payment with the new facility, IREN said it has now secured roughly 96% of its planned $5.81 billion in GPU-related capital expenditures, with an all-in financing cost of 3.31%. For the AI infrastructure market—where high demand for advanced accelerators has collided with tight supply and expensive capital—investment-grade terms are notable, and point to the credit value of long-duration customer contracts and collateralizable hardware.

The financing uses a project-level approach that requires a minimum 1.05x debt service coverage ratio, and applies 'ring-fencing' to isolate project risks from the parent company. In practice, that means the GPU assets and Microsoft-linked revenue streams are pledged at the project level, a structure increasingly favored by lenders seeking clearer recourse in capital-intensive compute builds.

IREN said the improved capital structure supports a target of reaching 480 megawatts of AI cloud capacity by the end of 2026. The company is leaning on a portfolio of grid-connected land and power assets across North America, Europe, and the Asia-Pacific region, emphasizing access to renewable-heavy energy sources—an increasingly central constraint for large-scale AI training and inference clusters. In the current AI cycle, reliable power procurement and interconnection timelines have become as critical as silicon availability, pushing infrastructure operators with energy expertise into a more strategic role.

On the deployment side, IREN said it is working with BE Networks to simulate and validate network architecture for installing more than 50,000 Nvidia Blackwell Ultra GPUs. The company is using Nvidia’s DSX Air platform to test data center network design in a high-fidelity simulation environment before physical buildout, aiming to reduce implementation risk as it scales. The move highlights an industry shift toward pre-deployment validation as AI clusters grow larger and more complex, where networking bottlenecks can materially impair utilization and economics.

Market attention has also focused on the stock’s sharp run-up. IREN traded between an intraday high of $66.50 and low of $60.26, with volume of about 55.8 million shares. The stock is up roughly 70% to 75% year-to-date from early-year levels near $37.77, and sits below its 52-week high of $76.87 but far above its 52-week low of $8.315.

Wall Street expectations remain divided, reflecting both optimism around IREN’s transition and uncertainty around execution, pricing power, and the durability of AI compute demand. Data compiled by MarketBeat shows a consensus view of 'Moderate Buy' based on a mix of buy, hold, and sell ratings, with average price targets around $79 to $80. QuiverQuant’s aggregation of eight analysts pegs a median target at $82.50. Cantor Fitzgerald has issued a $99 target, Macquarie $90, and B. Riley Securities $88, while JPMorgan has remained more cautious with a $46 target—an unusually wide range that illustrates the valuation debate across the AI infrastructure cohort.

Strategically, IREN’s announcement reinforces the broader narrative of Bitcoin mining infrastructure being repurposed for AI. Miners built expertise in procuring power, operating at high utilization, and deploying power-dense data centers—capabilities that can translate into AI cloud operations when paired with enterprise-grade customers and long-term contracts. In IREN’s case, the company is positioning itself as a vertically integrated AI cloud provider built around large GPU clusters for both training and inference workloads.

Notably, the update focused on infrastructure financing, capacity expansion, and institutional partnerships rather than any crypto-native product strategy. There was no announcement around issuing a proprietary token or launching a token-based ecosystem; instead, the company’s messaging centered on contracting, hardware deployment, and lowering the cost of capital—signals aimed squarely at credit markets and long-duration enterprise compute demand.

Looking ahead, the key question for investors is whether IREN can translate its funding advantage and power footprint into consistent delivery of capacity on schedule, while keeping network architecture, cooling, and utilization at levels that justify a multi-billion-dollar GPU build. If execution holds, the company’s ability to combine renewable-linked energy assets, investment-grade project finance, and a mega-cap customer relationship could strengthen its standing in a global AI cloud market increasingly constrained by both electricity and advanced accelerators.