Key Takeaways
- • $1 trillion in global AI data center spending projected for 2026 (Goldman Sachs estimate), with JPMorgan forecasting $697 billion in the US and Bank of America projecting a path toward $1.2 trillion by 2027, but capital availability is not the primary constraint.
- • AI hyperscalers issued $159 billion in bonds globally in 2026, a 9x increase from $17 billion in 2024, demonstrating Wall Street's aggressive capital deployment into data center infrastructure financing despite structural headwinds.
- • Power supply represents the critical bottleneck, with Bloomberg New Energy Finance estimating a 19-gigawatt shortfall for AI data centers by 2035 at current growth rates, alongside persistent chip shortages and skilled labor constraints.
- • Regulatory backlash is intensifying, including a one-year moratorium on data center development in New York and power hookup audits in Texas, creating operational delays that capital alone cannot overcome.
- • Shaky accounting practices and speculative growth projections raise sustainability concerns, with analysts warning that the financial models underlying this trillion-dollar build-out may not account for infrastructure delivery timelines, regulatory friction, and power grid limitations.
Summary
Wall Street is executing a massive capital deployment into AI data center infrastructure, but the market faces a fundamental paradox: money is no longer the limiting factor. Goldman Sachs estimates global AI hyperscaler spending will reach $1 trillion in 2026, with JPMorgan projecting $697 billion in US investment alone, and Bank of America identifying a path toward $1.2 trillion by 2027. This capital influx is being channeled through the debt markets, where AI hyperscalers have issued $159 billion in bonds globally during 2026—a nine-fold increase from $17 billion in 2024—demonstrating the financial industry's confidence in the AI infrastructure narrative.
However, the actual bottlenecks constraining this expansion are structural rather than financial. Power supply has emerged as the paramount constraint, with Bloomberg New Energy Finance projecting a 19-gigawatt shortfall in available power capacity for AI data centers by 2035 if growth continues at current trajectories. Beyond energy, the build-out faces compounding friction: persistent chip shortages despite new manufacturing capacity investments, a severe shortage of skilled construction labor to meet project timelines, and an escalating regulatory environment that includes a one-year moratorium on new data center development in New York and power hookup audits in Texas.
Industry analysts acknowledge these structural headwinds directly. George Gianarikas, an analyst at Canaccord Genuity covering power generation companies, emphasized that infrastructure requirements extend far beyond capital: "Not only do we need the equipment, not only do we need the permits, but we need the people. And what's happening in conjunction with all that is the fact that people are rallying against data centers." Public opposition, reflected in protests and regulatory pauses across multiple jurisdictions, compounds the timeline and cost uncertainties embedded in hyperscaler capex projections.
Financial concerns about the sustainability of this build-out are intensifying among institutional observers. Multiple sources note that the accounting methodologies underlying these trillion-dollar spending forecasts may obscure critical infrastructure delivery risks and speculative growth assumptions. The convergence of unlimited capital availability with constrained physical infrastructure, skilled labor, and regulatory approval creates a scenario where financial resources alone cannot guarantee the timely completion of AI data center capacity needed to sustain revenue projections embedded in current equity and debt valuations.
However, the actual bottlenecks constraining this expansion are structural rather than financial. Power supply has emerged as the paramount constraint, with Bloomberg New Energy Finance projecting a 19-gigawatt shortfall in available power capacity for AI data centers by 2035 if growth continues at current trajectories. Beyond energy, the build-out faces compounding friction: persistent chip shortages despite new manufacturing capacity investments, a severe shortage of skilled construction labor to meet project timelines, and an escalating regulatory environment that includes a one-year moratorium on new data center development in New York and power hookup audits in Texas.
Industry analysts acknowledge these structural headwinds directly. George Gianarikas, an analyst at Canaccord Genuity covering power generation companies, emphasized that infrastructure requirements extend far beyond capital: "Not only do we need the equipment, not only do we need the permits, but we need the people. And what's happening in conjunction with all that is the fact that people are rallying against data centers." Public opposition, reflected in protests and regulatory pauses across multiple jurisdictions, compounds the timeline and cost uncertainties embedded in hyperscaler capex projections.
Financial concerns about the sustainability of this build-out are intensifying among institutional observers. Multiple sources note that the accounting methodologies underlying these trillion-dollar spending forecasts may obscure critical infrastructure delivery risks and speculative growth assumptions. The convergence of unlimited capital availability with constrained physical infrastructure, skilled labor, and regulatory approval creates a scenario where financial resources alone cannot guarantee the timely completion of AI data center capacity needed to sustain revenue projections embedded in current equity and debt valuations.
Go deeper than the headline
Research every firm in this story, track the ones that matter and get the daily briefing — free for 7 days.