NVIDIA secures a $2.1 billion warrant in IREN as part of a 5GW AI data center agreement.

      NVIDIA plans to invest up to $2.1 billion in IREN as part of an agreement that combines the chipmaker’s reference architecture with the data-center operator’s 5-gigawatt infrastructure pipeline. The partnership was announced by both companies on Thursday.

      The structure is atypical. Instead of a straightforward equity round, IREN has granted Nvidia a five-year warrant for up to 30 million shares at an exercise price of $70. Following the announcement, the stock closed regular trading at $56.85, then experienced roughly a 9% increase in extended trading.

      If Nvidia fully exercises the options, the total investment would amount to $2.1 billion at the strike price, with the cash outflow occurring at the chipmaker's discretion rather than upon closing the deal.

      Initially, operational focus will be on the Sweetwater campus in Texas, which is planned to have 2GW of capacity. Last year, IREN established a $9.7 billion cloud agreement with Microsoft for capacity at the same location; the collaboration with Nvidia follows a similar strategy, positioning the chipmaker as both a partner and shareholder.

      What Nvidia is acquiring

      This agreement places IREN in a category that has become a familiar part of Nvidia's portfolio over the past 18 months. The chipmaker has invested in CoreWeave, Nebius, and various other so-called "neoclouds," which are firms that purchase Nvidia GPUs at scale and lease them back to hyperscalers and frontier-model builders.

      Nebius's acquisition of Eigen AI to maximize GPU token usage was among the most notable moves in this sector. The underlying premise is clear: AI demand is outpacing hyperscaler capacity at a rate quicker than what the hyperscalers can accommodate, allowing Nvidia to benefit doubly if neoclouds use its chips to fill the gap.

      The figures supporting this idea have increased each quarter. The four largest tech companies in the US have projected over $700 billion in capital expenditure for 2026, much of which is allocated for AI infrastructure.

      Even with this pace, providers of AI models and emerging enterprise users continue to face waitlists for capacity. IREN's 5GW represents a significant part of the solution, and the broader Texas grid expansion sets the groundwork for a multi-year commitment.

      IREN began as Iris Energy, a Bitcoin mining company with grid-connected infrastructure in Texas and British Columbia. In 2024, it shifted focus to AI computing, dropping the Iris Energy name in favor of its current IREN branding and cloud-centric positioning.

      The Microsoft partnership in 2025 validated this shift at scale, while the Nvidia announcement solidifies the supply side.

      The transition from cryptocurrency to AI cloud computing is becoming increasingly common, with companies like Hut 8, Applied Digital, and Core Scientific redirecting portions of their capacity.

      What distinguishes IREN at this point is its portfolio of grid interconnections in areas with inexpensive, ample energy—a constraint that has become more pressing than capital for new AI ventures. The Sweetwater site serves as a demonstration point. Texas regulators have been quicker to approve large interconnections compared to other states, and the campus is located near renewable energy generation that IREN can secure through long-term contracts.

      A 2GW campus is comparable in scale to the largest hyperscaler developments in the region.

      The neocloud model

      Referring to IREN as a neocloud summarizes a specific business model: constructing dedicated capacity, signing long-term contracts with one or two main clients (in this case, Microsoft), and financing GPU purchases through a combination of vendor financing, equipment leasing, and equity.

      Meta’s $21 billion agreement with CoreWeave stands as the largest example of this type of anchor-tenant arrangement, with CoreWeave’s multi-year deal with Anthropic being another case. The model is now recognized well enough that public investors evaluate these companies based on contracted capacity rather than traditional cloud revenue multiples.

      NVIDIA's role in these structures has become more direct over time, evolving from merely selling GPUs to taking warrants or equity stakes in buyers. This approach serves to share in operational gains and ensure that the capacity dedicated to its chips remains priced and supported in a preferred manner.

      The IREN agreement extends this approach into a public-company collaboration rather than private funding.

      The calculations that must be considered

      While 5GW is a significant figure, the primary constraints are timing rather than capital. Issues such as substation interconnections, transformer lead times, GPU shipping schedules, and renewable Power Purchase Agreements all have multi-year cycles that cannot be expedited.

      The Sweetwater campus, which is closest to operational readiness, will receive most of the initial Nvidia-IREN co-investment.

      Beyond this, the partnership is more of a framework than a concrete build plan; future initiatives will depend on demand and the regulatory and supply landscape.

      This pattern is reminiscent of Nscale’s expansion in Portugal alongside Microsoft, where the