IREN's first 50MW AI cloud deployment is live for Microsoft, backed by NVIDIA's Exemplar Cloud certification.
IREN's first 50MW AI cloud deployment is live for Microsoft, backed by NVIDIA's Exemplar Cloud certification.

IREN's first 50MW AI cloud deployment is live for Microsoft, backed by NVIDIA's Exemplar Cloud certification.
IREN's first 50MW AI cloud deployment is now running for Microsoft, with NVIDIA certifying the GB300 NVL72 cluster as an Exemplar Cloud — locking in the first tranche of a $9.7bn, five-year contract.
"Delivering Horizon 1 demonstrates the strength of our vertically integrated model and our ability to execute complex AI infrastructure projects at speed and scale," Daniel Roberts, co-founder and co-CEO of IREN, said.
Horizon 1 is the first of four 50MW direct-to-chip liquid cooled deployments scheduled for Microsoft at IREN's Childress, Texas campus in 2026. NVIDIA's Exemplar Cloud designation followed testing of the GB300 NVL72 deployment, confirming the cluster can support demanding AI workloads with performance, reliability and scale. IREN targets 480MW gross AI Cloud capacity in 2026 and 1.2GW in 2027.
The delivery comes as IREN raised its 2026 annualized run-rate revenue target to more than $4bn in July, after signing $2.8bn in new customer contracts. The company completed its acquisition of Mirantis, a cloud software provider, on Aug. 4, adding orchestration software to its stack.
Why the delivery matters
The Microsoft contract, announced in November 2025, is the largest single commitment in IREN's history and anchors its pivot from bitcoin mining to AI cloud services. The company's vertically integrated model — spanning data center design, engineering and construction — gives it end-to-end control over deployments, compressing delivery timelines that hyperscaler-backed rivals often stretch over years.
Direct-to-chip liquid cooling, which routes coolant through cold plates in direct contact with processors, is central to the GB300 NVL72 architecture. The approach lets operators pack more compute per rack while cutting the energy overhead of air cooling, a factor that matters as power becomes the binding constraint on AI capacity. NVIDIA's Exemplar Cloud program tests partner clusters for performance, reliability and scale; passing it tells prospective customers that IREN's infrastructure meets the bar for production AI workloads.
The certification also strengthens IREN's hand in a crowded market. CoreWeave, Crusoe and other GPU-cloud specialists are racing to secure hyperscaler contracts, and NVIDIA's endorsement gives IREN a credential that smaller rivals lack. Microsoft's acceptance of Horizon 1 converts a signed agreement into recognized revenue, with Horizons 2-4 scheduled for delivery later this year.
What the milestone means for investors
IREN's shares have tracked the company's AI pivot closely, with the stock reacting to each contract milestone. The delivery de-risks a portion of the $9.7bn backlog, though the remaining three Horizons and the 2027 expansion to 1.2GW still carry execution risk. NVIDIA's certification adds a quality signal that could help IREN win additional customers beyond Microsoft, supporting the raised $4bn ARR target. The company's more than 3,000-person site team at Childress will now shift focus to delivering Horizons 2-4, which IREN expects to complete before year-end.
This article is for informational purposes only and does not constitute investment advice.