- IREN is planning $25 billion to $30 billion in fiscal 2027 capital spending while billionaire investor Ken Griffin has disclosed a 4.7% economic stake in the company.
IREN Limited (NASDAQ: IREN) is facing a major financing test as it prepares for one of the largest AI infrastructure expansions among publicly traded data-center operators. The company plans to spend between $25 billion and $30 billion during fiscal 2027, according to its latest financial update.
At the same time, a Sept. 30 SEC filing showed Kenneth Griffin may be deemed to beneficially own 18.69 million IREN shares, equivalent to 4.7% of the company based on the reported share count. Citadel Securities LLC separately reported 14.15 million shares, or 3.6%, while Citadel Securities Group and Citadel Securities GP each reported 17.86 million shares, or 4.5%.
The overlapping Citadel disclosures should not be added together as separate economic positions. They are part of the same reporting group connected to Griffin. The development comes as investors weigh whether IREN can finance its aggressive expansion without placing additional pressure on its balance sheet or existing shareholders.
IREN's $30 Billion Expansion Creates a Major Financing Question
IREN's fiscal 2027 spending target is far larger than the company's current annual revenue base. The company generated $707 million of revenue in fiscal 2026, including $128.8 million from AI Cloud Services and $578.2 million from Bitcoin mining. Adjusted EBITDA was $245.7 million, while the company reported a $702.6 million net loss.
The financial results reflect IREN's transition from Bitcoin mining toward AI computing infrastructure. AI Cloud Services revenue increased nearly eightfold from $16.4 million in fiscal 2025 to $128.8 million in fiscal 2026. IREN says it already has $4 billion of contracted annualized revenue for 2026 capacity, with approximately $1 billion of operating ARR. Its 2026 capacity is largely sold out, while the company is in discussions with customers for a significant portion of its 2027 capacity.
That demand is central to the expansion case. IREN is developing a data-center pipeline exceeding 5 gigawatts and expects to deliver approximately 0.8 gigawatts of IT capacity during 2027. The company has also secured substantial financing tied to specific customer contracts. Its $3.65 billion investment-grade GPU financing for the Microsoft contract, combined with customer prepayments, covers approximately 96% of the related GPU capital expenditure.
IREN also secured another $2.8 billion of GPU financing for non-investment-grade customer deployments, including a $2.4 billion facility led by Blue Owl and Pacific Investment Management Company for the Mackenzie expansion.
For investors tracking the broader AI infrastructure spending cycle, the important question is how much of IREN's future construction can be funded through customer commitments and asset-level financing rather than new equity.
The company has already demonstrated one financing model through its Microsoft agreement, but a $25 billion-to-$30 billion annual capital program requires financing on a much broader scale.
Ken Griffin's IREN Stake Comes as AI Demand Expands
Griffin's disclosed position gives IREN another high-profile institutional investor, but the filing does not explain his investment thesis or guarantee future performance. The Sept. 30 Schedule 13G states that the reported securities were acquired in the ordinary course and not for the purpose of changing or influencing control of IREN.
IREN's latest operating results provide the fundamental backdrop for the investment. The company delivered the first of four 50-megawatt liquid-cooled deployments at its Childress, Texas, campus to Microsoft during fiscal 2026. Horizon 2 was under commissioning, while Horizons 3 and 4 were in late-stage construction, with delivery targeted for the fourth quarter.
IREN also has a separate relationship with NVIDIA. An SEC filing shows the company agreed in May to provide dedicated GPU services to NVIDIA under a five-year agreement valued at approximately $3.4 billion. NVIDIA also received investment rights to purchase up to 30 million IREN ordinary shares at $70 per share, subject to vesting conditions tied to deliveries of up to 600,000 NVIDIA GPUs.
That arrangement creates another link between IREN's expansion and continued demand for AI computing capacity. Investors have nevertheless seen how quickly the market can reassess the stock. IREN closed at $41.28 on Oct. 6, up 1.98%, before falling during trading on Oct. 7 as the broader market weakened.
The stock remains well below its 52-week high of about $76.87, according to current market data. IREN's financing strategy is therefore likely to remain one of the most important issues for shareholders. The company received an SEC no-action response on Sept. 23, allowing it to file a definitive proxy statement without first filing a preliminary proxy for specified shareholder proposals under Australian law.
For investors comparing AI stocks, IREN offers more direct exposure to the physical infrastructure required for AI workloads, but its capital requirements are also much higher relative to its current revenue. The company is moving rapidly from Bitcoin mining toward AI Cloud Services, with NVIDIA GPUs, Microsoft contracts, and large data-center projects forming the core of the strategy.
For shareholders, the central issue is whether contracted AI demand can scale quickly enough to support the enormous investment required. IREN's long-term investment case increasingly depends on turning power capacity, GPUs, and data centers into recurring AI revenue while maintaining access to affordable financing.
The latest filings do not resolve that question. They do show that major institutional investors are participating while IREN prepares for a capital-intensive expansion that could reshape the company's earnings profile over the next several years.
The next major test will be financing and executing the 2027 build-out without allowing construction costs, borrowing expenses, or potential equity issuance to outpace the revenue generated by the new AI capacity.