# IREN Shifts From Bitcoin Mining to AI Data Centers With Microsoft, NVIDIA Deals

By Token Post

Canonical URL: https://www.tokenpost.com/news/business/22037
Published: 2026-07-13T00:26:46.000Z
Updated: 2026-07-13T00:26:46.000Z
Section: Business

> IREN is pivoting from Bitcoin mining to AI cloud infrastructure through major deals with Microsoft and NVIDIA to build large-scale data center capacity.

IREN Limited ($IREN) is accelerating its pivot from Bitcoin mining to AI cloud infrastructure, aiming to reposition itself as a large-scale data center and GPU compute provider just as demand for ‘AI compute’ tightens globally. The strategic shift comes amid a sharp pullback in the stock, underscoring how quickly market expectations have reset for former mining names trying to become AI infrastructure plays.

Shares of IREN were recently changing hands around $41.14, roughly 43% below the company’s record high of $76.87. The move follows an extended rally off cycle lows, after which the stock entered a consolidation phase that has also weighed on a broader group of crypto-linked infrastructure equities.

From a technical perspective, traders have been watching near-term resistance around the 4-hour chart’s 200-period exponential moving average near $49.98 and the 50-period exponential moving average near $51.68. On the downside, a key support zone is clustered in the $27.57 to $29.99 range—often described as a ‘golden ratio’ Fibonacci retracement area—viewed by many market participants as a critical level for the longer-term trend. In European trading, IREN recently closed at €36.01, down about 1.5% on the day and roughly 19% over the past month, while still up about 147% year-on-year.

The core of IREN’s transformation is being shaped by a pair of headline partnerships with major U.S. technology firms. The company disclosed a $9.7 billion agreement with Microsoft ($MSFT), and said it secured a $3.65 billion GPU financing facility to support the build-out. According to the company, the structure covers roughly 96% of GPU capital expenditures at an average interest rate of about 6%, easing the near-term balance sheet burden typically associated with hyperscale AI infrastructure rollouts.

IREN has also signed a five-year partnership valued at $3.4 billion with NVIDIA ($NVDA). Under the arrangement, NVIDIA is expected to run AI cloud workloads at IREN’s Childress, Texas site using air-cooled Blackwell systems—NVIDIA’s next-generation architecture designed for large-scale model training and inference. The agreement also includes an option for NVIDIA to purchase up to 30 million shares at $70 per share over the next five years, implying a potential $2.1 billion equity investment if exercised.

To scale execution, IREN has strengthened its leadership bench with hires from major cloud and data-center operators. The company appointed Kambiz Aghili, previously at Oracle ($ORCL), as chief product officer, and named Michael Nudelman—whose background includes Google (Alphabet, $GOOGL) and CyrusOne—as chief development officer. The appointments are intended to support product development for AI cloud services, large-scale data-center construction, and power infrastructure expansion.

IREN operates sites in Australia and Canada and says it controls a grid-connected power portfolio totaling about 5 gigawatts. Management has framed its end-state ambition as a fully ‘vertically integrated’ AI cloud and data-center supplier, spanning power procurement, site development, and compute delivery. The company also noted a partnership with BE Networks to simulate deployment scenarios for NVIDIA Blackwell GPU infrastructure, a step aimed at shortening time-to-deployment for next-generation hardware.

Financially, IREN’s internal roadmap targets approximately $8.7 billion in revenue and about $500.48 million in profit by 2029, implying exceptionally rapid growth from current levels. Some analysts model an even more aggressive trajectory, projecting 2029 revenue of $14.9 billion and profit of $1.4 billion, reflecting expectations that AI infrastructure revenue can scale quickly once capacity comes online and utilization rates normalize.

Freedom Capital Markets recently upgraded IREN to ‘strong buy’ from ‘hold’ and set a $58 price target, implying roughly 49% upside from recent levels. The firm cited the AI data-center transition and the NVIDIA agreement as central to its reassessment. Analysts following the story expect revenue to rise from an estimated $717 million this year to about $3.1 billion by fiscal 2027, with gains driven predominantly by AI infrastructure rather than the legacy Bitcoin mining business.

Still, the pivot is unfolding against a softer backdrop for the sector. Arlington Capital Management Inc. disclosed a $3.45 million position in IREN, signaling continued institutional interest, but an index tracking AI-infrastructure-linked Bitcoin miners has fallen about 16% over the past month. Market participants have also noted instances of insider selling across the group, adding to near-term sentiment pressure.

For IREN, the headline contracts, fresh executive hires, and a clearly articulated capacity build-out plan have strengthened the company’s medium-term fundamental narrative. At the same time, analysts and investors are watching familiar risks in capital-intensive infrastructure models—customer concentration, leverage tied to GPU financing, and execution timelines—as the company attempts to convert large commitments into sustainable, recurring ‘AI cloud’ revenue.

Article Summary by TokenPost.ai

🔎 Market Interpretation

Repositioning narrative: IREN is accelerating a shift from Bitcoin mining to AI cloud/data-center infrastructure, attempting to capture tightening global demand for GPU compute.

Reset expectations reflected in price: Shares around $41.14 remain roughly 43% below the $76.87 peak, showing that markets are still discounting execution and funding risk despite major AI headlines.

Key technical levels: Near-term resistance is watched around the 4H 200 EMA (~$49.98) and 50 EMA (~$51.68). A major support zone sits at $27.57–$29.99 (commonly labeled a “golden ratio” Fibonacci retracement area), viewed as pivotal for the longer-term trend.

Cross-market performance snapshot: European close at €36.01 (about -1.5% on day, -19% over a month), but still about +147% year-on-year—highlighting elevated volatility typical of “pivot” stories.

Sector headwinds persist: An index tied to AI-infrastructure-linked Bitcoin miners fell about 16% over the past month, and reports of insider selling have pressured sentiment broadly, even as institutions selectively add exposure.

💡 Strategic Points

Microsoft-scale demand signal: IREN disclosed a $9.7B agreement with Microsoft, positioning itself as a potential hyperscale AI compute supplier if capacity comes online as planned.

GPU financing reduces upfront strain (but adds leverage risk): A $3.65B GPU financing facility is said to cover ~96% of GPU capex at ~6% average interest, potentially easing near-term balance sheet pressure while increasing sensitivity to utilization and contract durability.

NVIDIA partnership anchors workload credibility: A 5-year, $3.4B partnership has NVIDIA running AI cloud workloads at IREN’s Childress, Texas site using air-cooled Blackwell systems (next-gen architecture for training/inference).

Embedded equity option could become a major catalyst: NVIDIA holds an option to buy up to 30M shares at $70 over five years—implying up to $2.1B potential equity investment if exercised (supportive if it validates execution; dilutive if framed as financing substitute).

Operational capability build: Executive hires from major operators—Kambiz Aghili (ex-Oracle) as CPO and Michael Nudelman (ex-Google/Alphabet, CyrusOne) as CDO—are meant to strengthen productization, delivery timelines, and large-scale build execution.

Vertical integration thesis: IREN cites a grid-connected power portfolio of ~5 GW across sites in Australia and Canada, targeting control over power procurement, site development, and compute delivery—key to competing on cost and reliability.

Deployment acceleration efforts: Partnership with BE Networks to simulate NVIDIA Blackwell deployment scenarios aims to shorten time-to-deployment—critical in a market where uptime and speed-to-capacity are monetization drivers.

Ambitious 2029 financial targets: Company targets ~$8.7B revenue and ~$500.48M profit by 2029; some analysts forecast even higher (~$14.9B revenue, $1.4B profit), reflecting a “capacity + utilization” scaling assumption.

Street view turning more constructive: Freedom Capital Markets upgraded to Strong Buy with a $58 target (~49% upside), expecting growth from ~$717M this year to ~$3.1B by FY2027, driven primarily by AI infrastructure.

Key risks to monitor: Customer concentration (large counterparties), leverage and refinancing tied to GPU financing, construction/energization timelines, and the ability to translate headline contracts into recurring AI cloud revenue and stable utilization.

📘 Glossary

AI compute: Specialized computing capacity (often GPUs) used for training and running AI models.

GPU (Graphics Processing Unit): High-parallelism processors central to modern AI training and inference workloads.

Hyperscale: Extremely large-scale data center deployments typically associated with major cloud providers.

Capex (Capital Expenditures): Upfront spending to build/expand long-lived assets such as data centers and GPU clusters.

Utilization rate: The percentage of installed compute capacity that is actively used and generating revenue.

Customer concentration: Dependence on a small number of customers for a large share of revenue; increases contract/renewal risk.

EMA (Exponential Moving Average): A technical indicator that weights recent prices more heavily to gauge trend and potential support/resistance.

Fibonacci retracement / “golden ratio” zone: Technical analysis levels derived from Fibonacci ratios; “golden ratio” commonly refers to ~61.8% retracement often watched as support in pullbacks.

Blackwell (NVIDIA architecture): NVIDIA’s next-generation GPU/platform designed for large-scale AI training and inference.

Vertically integrated: Operating across multiple parts of the value chain (power, land/site, build, and compute delivery) to control costs and execution.

Insider selling: Sales of shares by company insiders; not always negative but can pressure sentiment in high-duration growth stories.