Nvidia, the world’s largest microchip company, will invest up to $2.1 billion in Iren, an Australian company specializing in data center management, as part of a broader collaboration to build artificial intelligence infrastructure with a capacity of 5 gigawatts.
WHAT NVIDIA AND IREN WILL DO
The strategic partnership, announced Thursday, “aims to accelerate the large-scale construction of factories for artificial intelligence by combining Nvidia’s Dsx architecture with Iren’s expertise in energy, land, data centers, GPU [graphics processing unit, ed.] installation, and infrastructure management. Nvidia’s GPUs – processors that perform the calculations necessary for training and running AI systems, in short – are in high demand due to their performance but come with very high prices: up to $40,000 each.
INTEREST IN NEOCLOUDS
Iren is considered a neocloud, that is, a specialized provider of cloud computing services to large tech companies engaged in AI development. In this regard, last year Iren signed a $9.7 billion agreement with Microsoft on cloud computing.
This is not the first time Nvidia has invested in a neocloud: in January, for example, it announced a $2 billion funding to the US-based CoreWeave, which will use the resources to build data centers with a computing capacity of over 5 GW by 2030. CoreWeave’s main client is also Microsoft.
A NEW “CIRCULAR” DEAL?
Both Iren and CoreWeave are heavy users of Nvidia processors. For this reason, its investments in the two companies are defined as “circular” financing: Nvidia financially supports its customers with the goal of fueling demand for its own products.
To those who fear that circular deals could lead to the bursting of the alleged “artificial intelligence bubble,” Nvidia CEO Jensen Huang responded that these investments – albeit large – represent only a small part of spending on technological infrastructure. Indeed, this year American Big Tech companies are expected to invest over $700 billion in artificial intelligence.
