There’s a class of very speculative Bitcoin (BTCUSD) mining companies that have converted their expertise in mining Bitcoin into AI computing operations. These companies have been making a killing, with IREN (IREN) being one of the most successful.
IREN is now a high-performance computing (or HPC) company. It buys the hardware and has its own data centers, from which it sells compute. Computing capacity is scarce, and AI companies are buying from essentially whoever they can get compute from. Thus, the business model has been working surprisingly well.
More News from Barchart
IREN’s Multi-Billion-Dollar Deals on the Rise
Last year in November, IREN announced a massive $9.7 billion deal with Microsoft (MSFT). IREN will give Microsoft access to Nvidia (NVDA) GB300 GPUs for over five years. It then entered into an agreement with Dell (DELL) to buy the hardware for $5.8 billion. The GPUs would be deployed in phases through 2026 at IREN’s 750MW Childress, Texas, campus across four liquid-cooled data centers called Horizon 1 through 4.
IREN ended up missing revenue estimates in Q2 FY2026, but this was mostly due to Bitcoin-related revenue declining as the company is focusing almost entirely on AI now. It has a cash position of $2.6 billion in Q3 FY2026 and reported $9.2 billion secured from funding sources the quarter before. It used a portion of this cash to buy cloud infrastructure software provider Mirantis in a deal worth $625 million.
The deals kept on rising, with IREN announcing a $3.4 billion AI cloud contract with Nvidia.
The New Deal with Dell
IREN signed a $1.6 billion purchase agreement with Dell for air-cooled Blackwell systems to be deployed at its data center in Childress. Once all of this comes online, the company says its annualized recurring revenue will climb from $3.7 billion to $4.4 billion.
One thing that’s interesting here is that IREN is not building data centers from the ground up. It is often retrofitting buildings and using leftover equipment and employees from its Bitcoin mining operations to drive down costs. This makes it arguably a better bet than many other “neocloud” companies that don’t have existing facilities or a workforce familiar with GPUs.
