Iris Energy has struck a deal with AI startup Poolside to give the latter cloud access to Nvidia GPUs.
The data center provider has agreed to a cloud service agreement with Poolside for 248 of Nvidia’s top-of-the-range H100 GPUs, which are used for training AI models. The initial contract is three months, with a potential three month extension, Iris Energy said.
Daniel Roberts, co-Founder and co-CEO of Iris Energy, said: “We are pleased to partner with a company the caliber of Poolside. We look forward to further growing our GPU cloud services business and servicing the growing market demand we are seeing for these services.”
Iris Energy was founded in 2019 build and operate data centers and electrical infrastructure needed to mine Bitcoin and other cryptocurrencies. But declining interest in crypto assets, along with several high-profile scandals, have caused problems for many businesses in the sector, and in 2022 Iris revealed two of its subsidiaries had defaulted on debts relating to the purchase of cryptomining equipment.
Last year, the company pivoted to become a provider of AI infrastructure in an attempt to cash in on growing interest in generative AI systems.
According to its website, it currently operates four data centers - three in the Canadian province of British Colombia, and one in Texas - all of which are powered by renewable energy. The company says it currently has 180MW capacity across the four sites, with 80MW additional space under construction.
Poolside closed a $126m seed funding round last year and is building what it claims will be “the world’s most capable AI for software development.” It says it selected Iris after a rigorous testing process, and the contract got underway earlier this month.
Jason Warner, CEO and co-founder of Poolside, said: “We’re excited to select Iris Energy as a cloud services partner. Our business, like many others, has a substantial growing demand for GPU compute and we have been impressed with the professionalism and quality shown by Iris Energy during our testing and selection process.”