IREN
Iris Energy
Summary
What they do:
IREN (formerly Iris Energy) is a vertically integrated AI infrastructure operator running a dual model — leasing data center capacity to Microsoft under a $9.7B contract while simultaneously operating its own GPU cloud with 150,000 GPUs, all powered from its 810MW+ operational campus at Childress, Texas.
Why they matter:
IREN is the most advanced operator in L24 by nearly every metric — largest power portfolio (4.5GW+), highest-value single contract (Microsoft $9.7B), fastest revenue growth (+355% YoY in Q1 FY26), and dual hosting+cloud exposure that captures value across both infrastructure and compute layers.
Recent performance:
Q1 FY2026 (Sep 2025) revenue $240.3M (+355% YoY), net income $384.6M, adjusted EBITDA $91.7M; stock ~$48, market cap ~$16B. Targeting $3.4B ARR by end of CY2026.
Our Verdict
The most operationally advanced miner-to-AI pivot with a $9.7B Microsoft anchor, 150,000 GPUs generating cloud revenue, and a 4.5GW power pipeline — IREN is executing at a pace that justifies its premium valuation, with the dual hosting+GPU-cloud model creating the widest moat in L24.
Structural trends
Structural
75
/ 100
Moat
6/10
Dual Model + Microsoft Anchor + NVIDIA Partner
AI Exp.AI Exposure
Pure Play~90% AI
Play Type
EmergingAI Growth
~355% YoY
Rel. Value
86
COMPELLINGPriceLIVE
$47.37
+9.98%
Live via Yahoo Finance · refreshes every 5 min
Market Cap
$15.7B
P/E Ratio
32.9
P/S Ratio
20.8x
52W High
$76.87
52W Low
$5.24
52W Chg
804.0%
Beta
4.31
IREN emerged from Iris Energy, an Australian-founded Bitcoin mining company that operated sites in British Columbia and Texas using renewable energy. The company rebranded to IREN in November 2024, reflecting a strategic pivot from mining to AI infrastructure that has been the most successful transformation in the L24 layer.
The Childress, Texas campus is the crown jewel — 810MW operational with 2,100MW under construction, representing one of the largest single-site AI data center developments in the world. The campus hosts both Microsoft's dedicated infrastructure (under the $9.7B, 5-year contract announced November 2025) and IREN's own GPU cloud fleet. The Microsoft deal includes phased deployments of Horizon 1-4 data centers (200MW critical IT load, liquid-cooled), with Microsoft committing to a 20% prepayment. When fully online, the Microsoft contract alone is expected to deliver approximately $1.9B in annual recurring revenue at 85% EBITDA margins.
What distinguishes IREN from every other L24 operator is the dual model. While CORZ, APLD, and WULF are pure infrastructure lessors (they build facilities and lease them to tenants), IREN also operates its own GPU cloud — 150,000 GPUs as of March 2026, serving AI training and inference customers directly. This dual approach means IREN captures both the infrastructure margin (hosting) and the compute margin (cloud), creating higher total revenue per MW than any peer. The company secured $3.6B in GPU financing at below 6% interest to fund the cloud expansion — a deal that reflects NVIDIA's confidence in IREN as a preferred partner.
The total power portfolio exceeds 4.5GW across sites in Texas, British Columbia, and planned expansions. The company is targeting $3.4B in annualized run-rate revenue by end of CY2026 — an extraordinary target for a company that generated $52.8M in revenue just a year earlier. Even accounting for the typical gap between "ARR target" and actual recognized revenue, IREN's trajectory is the fastest in the layer.
Supply Chain Dependencies
Upstream Suppliers
Downstream Customers
The Catch
IREN's biggest vulnerability is the gap between aspiration and execution. The $3.4B ARR target by end CY2026 requires roughly tripling from current run-rate within 12-15 months — achievable only if both Microsoft deployment accelerates and GPU cloud customer acquisition hits targets simultaneously. If either leg underperforms, the stock's ~16x P/S multiple compresses toward infrastructure-peer levels (5-10x), implying 30-50% downside. The Microsoft contract, while transformative, expires in 2029-2030, and Microsoft is simultaneously building its own data centers — renewal is not guaranteed. The GPU cloud business, while differentiated by owned infrastructure, has no proprietary software moat and competes with well-funded operators (CoreWeave, Lambda) and the hyperscalers themselves.
If They Win
If IREN delivers the $3.4B ARR target, renews and expands the Microsoft contract, and builds the GPU cloud into a top-3 independent AI compute platform, the company becomes the category-defining "neocloud" — a vertically integrated infrastructure + compute platform that generates $5B+ in annual revenue with 40%+ EBITDA margins. In this scenario, IREN's 4.5GW power portfolio supports a decade of expansion, the dual model creates economics that pure lessors and pure cloud providers cannot match, and the stock trades at $100+ as the market re-rates IREN from an infrastructure company to a cloud platform.
Others in Operate the Data Center
Not financial advice. All scores generated via AI algorithms using public data.