CoreWeave Details Expansion Financing, Power Constraints and NVIDIA Growth Plans at Conference

Core Viewpoints - CoreWeave's business model relies on longer-dated "take-or-pay" contracts, providing visibility into future cash flows while incurring costs during hypergrowth, with stabilized margins in the mid-20s expected over the contract life [1][7] - The company has experienced a significant revenue increase from $200 million to $5 billion in a few years, with expectations to double again this year and in 2027, supported by a backlog of approximately $66 billion to $67 billion [2] Financing and Capital Expenditure - CoreWeave finances its rapid expansion through longer-dated "take-or-pay" contracts and asset-level delayed draw term loans (DDTLs), having financed about 90% of contract-level CapEx with DDTLs, with a total of $30 billion of CapEx tied to backlog [7][12] - The company anticipates substantial capital needs, with guidance including $30 billion of CapEx, primarily sourced from DDTLs, customer prepayments, and opportunistic top-level financings [12] Power Strategy and Infrastructure - CoreWeave prioritizes near-term, grid-connected power, with about 3.1 gigawatts of contracted power expected to be largely online by the end of 2027, and plans to scale self-build joint ventures to manage physical constraints [6][15] - The company emphasizes the importance of time to power and the source of power in site selection, typically contracting power 12 to 24 months before it comes online [13][14] Partnership and Product Expansion - CoreWeave is expanding its partnership with NVIDIA, targeting an additional 5 gigawatts of AI cloud capacity by 2030, and is focused on growing higher-margin add-on services beyond GPUs, which have reached a $100 million run rate [5][17] - The company is adapting to shifts in GPU technology and cooling methods, with a focus on liquid-cooled data centers and the adoption of newer GPU generations like Blackwell [18] Market Position and Execution - CoreWeave's approach to borrowing through DDTLs is based on specific contracted customer revenues, with a focus on contract quality and customer creditworthiness to drive lender confidence [8][10] - The company is focused on executing large compute builds while deploying newer technology, suggesting that the market may underestimate the sophistication required to deliver AI cloud at scale [19][20]