IREN (IREN), an information middle operator centered on AI cloud infrastructure, introduced that it has agreed to buy greater than 50,000 specialised processing chips from NVDA, increasing its capability by roughly 50%.
The B300 GPUs (graphics processing models) deliver the Sydney-based firm’s whole AI computing fleet to round 150,000 GPUs. GPUs are specialised chips that carry out massively parallel computations, enabling quick coaching and manipulation of synthetic intelligence fashions.
The corporate has additionally filed for a possible market share sale of as much as $6 billion as a part of a broader capital administration technique. Shares fell 5% in pre-market buying and selling Thursday on potential dilution.
Further {hardware} can be phased in all through the corporate’s air-cooled knowledge facilities in Mackenzie, British Columbia, and Childress, Texas, via the second half of 2026. As soon as totally deployed, the expanded fleet is predicted to assist greater than $3.7 billion in annual AI cloud income, positioning IREN as one of many largest AI cloud infrastructure suppliers globally.
IREN mentioned it has secured roughly $9.3 billion in financing over the previous eight months via buyer upfront funds, convertible notes, GPU leases and financing preparations, and expects roughly $3.5 billion in further capital expenditures for brand spanking new GPU deployments within the second half of 2026.
