A Capital Supercycle Meets a Credit Moment

The rise of generative AI has kicked off one of the most capital-intensive buildouts in modern history. At the heart of this boom are data centres—massive, power-hungry, hyperscale facilities purpose-built to host AI models, GPUs, and the compute infrastructure driving the next era of digital productivity. 

Estimates suggest this wave of investment could total $400 billion to $3 trillion globally over the next decade. Banks, private credit, infrastructure debt, institutional capital, and Wall Street syndicates are all rushing in to fund it. 

For credit investors, the opportunity is significant. But this is not a “rising tide lifts all boats” environment. Many projects will succeed. Others will stall or fail. Selectivity is paramount. 

State of the Market: Scale, Speed, and Sponsors

Big Tech—Microsoft, Amazon, Google, Meta, OpenAI—is leading the charge, pouring billions into new AI-specific data centre capacity. Proposed networks of hyperscale facilities from projects such as Hyperion, Prometheus, and Stargate are targeting hundreds of billions in long-term investment. 

Construction is underway across North America, Europe, and Asia, with governments, utilities, and regulators scrambling to keep up with the pace. 

Capital is coming from multiple sources:

• Big Tech

• Banks

• Infrastructure funds (Brookfield)

• Private equity and growth capital (KKR, Silver Lake) 

• Institutional debt (PIMCO, Apollo, Blackstone) 

• Syndicates, increasingly willing to structure bespoke financings  

Geopolitics is adding urgency. U.S. – China tech decoupling, European data sovereignty policies, and U.S. industrial incentives are accelerating capital flows and shaping jurisdictional risk. 

At the same time, Big Tech is increasingly outsourcing capacity, leasing from third-party operators and developers rather than owning everything outright. This trend is creating new avenues for credit capital to fund standalone projects and platforms. 

Capital Requirements and Funding Trends

The scale of capital flowing into AI-focused data center infrastructure is staggering. Estimates suggest the global buildout could require upwards of $3 trillion in investment by 2035. This surge is being financed through a blend of corporate debt, project-level financing, private credit and institutional equity. 

Corporate bond issuance remains a key channel, especially for hyperscalers and large operators with investment-grade ratings. However, much of the growth is being fueled by more bespoke capital solutions: project finance vehicles, infrastructure bonds, and private credit structures that allow for non-recourse financing and flexible repayment terms. Large institutional allocators and direct lenders—including names such as Apollo, PIMCO, and Blackstone—are actively deploying capital at scale. 

Recent transactions reflect the diversity of structures and participants. eStruxture secured $1.35 billion CAD in structured financing, split between $740 million CAD in asset-backed securities and a $600 million CAD revolving credit facility, to fund its AI-ready data centre expansion in Canada. Meta issued $10.5 billion USD in its largest ever high-grade bond offering for general corporate purposes including AI infrastructure spend, and is now pursuing a $29 billion USD private financing package—$26 billion USD in debt and $3 billion USD in equity provided by PIMCO and Blue Owl—to support its U.S. AI data centre buildout. Silver Lake announced a $400 million USD joint venture to secure and develop powered land for data centre projects, addressing critical power constraints. 

Increasingly, we are seeing HoldCo/OpCo separations, sustainability-linked debt, and bridge-to-permanent financing models. The use of performance-based triggers, ESG-linked pricing adjustments, and hybrid securities reflects the growing complexity of these deals. Meanwhile, spreads are beginning to tighten for high-quality issuers, particularly those with long-term leases from creditworthy tenants and operations in power-secure, strategically located markets.

As demand accelerates, capital structures are evolving—and in many cases, still being tested. The market is actively negotiating how risk is shared between sponsors, tenants, and lenders, with pricing mechanisms and terms continuing to shift. 

Key Considerations for Credit Investors

Data centre projects sit at the intersection of real estate, energy, and high-performance computing, and thus present a unique and complex risk profile. 

Capital structure is a primary concern. Many operators are pursuing aggressive growth, often funded with high levels of leverage. It’s critical to analyze debt service coverage ratios, refinancing risks, and the quality of covenant protections. Special attention should be paid to whether or not deals are non-recourse at the project level or carry corporate-level support, as this significantly alters downside scenarios. 

Tenant concentration is another key issue. While long-term, take-or-pay contracts with hyperscalers like Microsoft or Google can provide predictable cash flows, they also create significant exposure to a handful of counterparties. The quality of those contracts, including early termination rights, renewal options, and alignment with debt maturities, can dramatically affect the risk-return profile. 

Execution risk is nontrivial. AI-focused data centres require specialized construction, power provisioning, and cooling infrastructure. Permitting delays, community opposition, power interconnection challenges, or even contractor underperformance can derail projects and create liquidity issues. Many of these projects require not just technical expertise but political and logistical coordination as well as local community outreach that extends beyond traditional development. 

Power is both an asset and a liability. These facilities are massively energy-intensive, and while some operators have secured long-term supply agreements or are building on-site generation, others are exposed to volatile spot markets and regulatory risk. Investors need to assess  the extent energy costs are (or are not) passed through to tenants, and whether margin compression is a realistic threat. 

Where Deals Break

Despite the strong macro story, not every opportunity will deliver attractive or even positive credit outcomes. Several underwriting red flags are already emerging in the current deal flow, particularly as capital continues to chase deployment at speed. One of the more common warning signs is a lack of contractual backing. Just because hyperscalers are expanding globally does not mean they will lease a given facility at a given time. Deals built on speculative demand or verbal tenant interest should be viewed with caution. 

Another frequent issue is the misalignment between lease durations and debt maturities. In some structures, tenants have three-to-five-year leases while the financing is structured on a ten-year term. If tenants do not renew, this creates significant refinancing and cash flow risk. Similarly, projects without fully secured power interconnections, or that lack binding utility agreements, can face delays or cost blowouts that impair debt service. 

Weak sponsors are another concern. Inexperienced developers entering the market without a proven track record in high-density data centre construction often underestimate timelines and overpromise performance. Their financing partners bear the consequences when schedules slip, or capital runs short. Deals with minimal equity commitment or sponsor over-distribution via weak covenants are also vulnerable. 

Finally, a rising concern is technological rigidity. Facilities designed for current-generation hardware but without adaptability for future cooling or density requirements risk becoming stranded assets. Credit investors should be wary of projects that underinvest in flexibility or dismiss the likelihood of significant future retrofit capex. 

An Emerging Credit Landscape Still Taking Shape

The AI data centre buildout is reshaping the infrastructure investment landscape at a remarkable pace. Credit markets are responding just as quickly and deploying capital into new structures, geographies, and sponsors as capital races to meet demand. The scale of investment is unprecedented, but so is the complexity and variability of each opportunity. 

There is no question this is a market worth watching. The convergence of long-term leases, real assets, and strong thematic tailwinds creates a compelling credit backdrop. But the risks are real: underwriting discipline varies, execution challenges, and not all capital is being deployed with equal precision. 

As the sector continues to mature, the opportunity set for credit investors will likely grow more attractive…but also more selective. In a fast-moving market, patience and precision could be the difference between a resilient, cash-generating credit and a problematic one.