IREN (IREN), an information middle operator targeted on AI cloud infrastructure, mentioned it agreed to purchase greater than 50,000 specialised processing chips from Nvidia (NVDA), increasing its capability by about 50%.
The B300 GPUs, or graphic processing models, will take the Sydney-based firm’s complete AI compute fleet to about 150,000 GPUs. A GPU is a specialised chip for performing giant numbers of parallel computations, enabling the coaching and operation of synthetic intelligence fashions at excessive pace.
The corporate additionally filed for a possible at-the-market share sale of as much as $6 billion as a part of its broader capital administration technique. The shares dropped 5% in pre-market buying and selling on Thursday as a result of potential dilution.
The extra {hardware} is anticipated to be deployed in phases by way of the second half of 2026 throughout the corporate’s air-cooled knowledge facilities in Mackenzie, British Columbia, and Childress, Texas. As soon as absolutely deployed, the expanded fleet is projected to help greater than $3.7 billion in annualized AI cloud income, positioning IREN among the many bigger AI cloud infrastructure suppliers globally.
IREN mentioned it has secured about $9.3 billion in funding over the previous eight months by way of buyer prepayments, convertible notes, GPU leasing and financing preparations, with roughly $3.5 billion in further capital expenditures anticipated for the brand new GPU deployments within the second half of 2026.