$3 trillion. That’s the estimated cost to build out the AI infrastructure required by 2030, according to a recent internal memo at Citigroup, a figure that’s fundamentally reshaping Wall Street’s approach to financing. The surge in demand for data centers, driven by the explosive growth of artificial intelligence, isn’t just a boom for tech companies – it’s forcing a dramatic rethink of how massive infrastructure projects are funded, and which financial institutions are equipped to lead the charge. Follow the money, and you’ll find banks like JPMorgan, Morgan Stanley, and even regional players like Citizens are building entirely new teams and strategies to capitalize on what JPMorgan’s global chair of investment banking, Fred Turpin, calls “the largest investment cycle in the history of capitalism.”
Just two years ago, a $100 million financing deal for a data center would have been considered a significant milestone. Today, as Adam Lewis, a managing director at Citizens, bluntly puts it, “If you can’t invest a billion dollars, we don’t even want to talk to you.” This isn’t simply inflation; it reflects a fundamental shift in the scale and complexity of these projects. Rising land and electricity costs, coupled with the specialized power demands of AI, have pushed data center financing beyond the scope of traditional commercial real estate loans and squarely into the realm of large-scale infrastructure finance. The year-over-year increase in deal size is exponential, leaving many traditional lenders sidelined.
The scramble to participate is driving a wave of internal restructuring at major banks. JPMorgan, Goldman Sachs, and Morgan Stanley have all formed integrated teams over the past two years, bringing together experts in lending, capital markets, and even energy and technology. This isn’t about simply throwing money at a hot sector; it’s about developing a deep understanding of the intricate mechanics of data center construction. John Greenwood, a partner at Goldman Sachs’ Capital Solutions Group, emphasizes the need to be “elbow to elbow with the bankers that cover sponsors” to ensure a seamless connection between project origination and capital distribution. This integrated approach is a direct response to the increasing complexity of deals, which now require assembling multiple sources of capital – bank loans, bonds, private credit, and institutional investors – into a single, cohesive structure.
Based on the original Business Insider report.
This shift is being fueled by a critical dynamic: the world’s largest tech companies, the so-called “hyperscalers,” are hitting the limits of their own balance sheets. While they can’t afford to fall behind in the AI race, the sheer cost of building and powering these facilities is becoming unsustainable to fund internally. This creates an opportunity for banks to act as intermediaries, bridging the gap between the demand for AI infrastructure and the availability of long-term capital from sources like sovereign wealth funds and pension funds. Morgan Stanley’s recent $2.6 billion financing for CoreWeave, using Nvidia chips as collateral, and the $27 billion bond deal for a Meta and Blue Owl joint venture, exemplify this new approach.
However, the buildout isn’t without its constraints. Bankers are now required to possess a level of technical expertise previously unheard of in traditional finance. Understanding electrical diagrams, land use permits, and power configurations is no longer optional – it’s essential for assessing project viability. Scott Wilcoxen, who leads digital infrastructure investment banking at JPMorgan, points out that most people assume cloud computing is “ephemeral,” but it’s fundamentally reliant on a “physical connection between individual users and the data sources.” This growing awareness of physical limitations – particularly around power, equipment, and labor – raises questions about the sustainability of the current growth trajectory. Goldman’s Greenwood even recounts a recent client meeting where the term “terrestrial” data centers was used, hinting at a future potentially extending to underwater or even space-based facilities.
What this means for your wallet: the massive investment in AI infrastructure will likely translate to higher costs for cloud computing services in the long run, as providers pass on their increased capital expenditures. But it also signals a broader economic shift, with AI-driven innovation poised to reshape industries and create new investment opportunities. The key question for investors and consumers alike is whether the projected returns from AI will justify the unprecedented level of capital being poured into its foundation – and whether the supply of essential resources like power and specialized hardware can keep pace with demand. Watch closely for signs of bottlenecks in these critical areas, as they will likely dictate the future pace of AI development and its ultimate impact on the economy.
