The global supply of computing power is struggling to keep pace with the growing demand driven by artificial intelligence (AI) applications, according to Daniel Roberts, co-chief executive and co-founder of Iren, a prominent AI infrastructure company. Roberts described the current AI data center expansion as fundamentally different from previous technology investment cycles, highlighting the unprecedented scale and dynamics of the sector.
Iren, originally a cryptocurrency mining firm, has transformed into a key player in the AI cloud computing market, backed by Nvidia, Microsoft, and other major customers. The company plans to invest up to $30 billion in AI infrastructure over the next several years to meet increasing demand for processing capacity. Roberts noted that each unit of new supply tends to generate several multiples of additional demand, driven by the expanding use of AI agents and faster processing requirements.
The rush to expand AI computing capacity involves numerous emerging developers alongside established technology companies, all competing to bring new data centers online. Goldman Sachs projects that U.S. data center capacity will double by the end of 2027 compared to 2024 levels and triple by 2030, reaching about 125 gigawatts—enough energy to power more than 100 cities the size of San Francisco. Despite this ambitious growth, Roberts expressed skepticism that supply will ever fully meet demand, citing physical and social constraints as limiting factors.
Data center construction faces mounting challenges due to community resistance across the United States. Local opposition often centers on concerns about the impact on regional energy and water resources, acting as a brake on unchecked expansion. Roberts emphasized that these social and environmental factors impose tangible limits on how much capacity can be added.
Iren distinguishes itself by owning the entire operational stack—land, buildings, computing equipment, and software—an approach different from many rivals who rely on long-term leases to serve hyperscale tenants. While this vertical integration allows Iren to capture more value amid rising compute prices, it has required substantial share issuance, with the company's total shares outstanding growing significantly since its initial public offering.
Over the past year, Iren has mobilized approximately $19 billion through a mix of convertible notes, chip-backed loans, customer prepayments, and equity sales. A significant portion of this capital supports purchasing Nvidia GPUs, the specialized processors central to AI workloads, and constructing new data centers. Nvidia, in turn, plays multiple roles as a key supplier, customer, and investor, having entered a five-year, $3.4 billion capacity lease agreement with Iren and secured rights to acquire up to $2.1 billion in stock.
Roberts dismissed concerns about the close financial ties between Iren and Nvidia as strategic collaboration aimed at accelerating growth across the AI ecosystem. He noted that the companies maintain daily communication on operational and financial matters to adapt support as the market evolves.
As of June, Iren held $7.6 billion in cash alongside $7.6 billion in debt, with additional borrowing capacity and unmortgaged assets poised to fuel its planned infrastructure investment through 2027. The company’s rapid rise in market value—from about $60 million in late 2022 to roughly $17 billion today—reflects the volatile yet lucrative nature of building the computational backbone for AI’s expanding role across industries.