The $5 Trillion Reality Check
While equity markets obsess over annual recurring revenue and large language model capabilities, the credit markets have turned their attention to a much heavier reality: concrete, copper, and cooling systems. The artificial intelligence boom has transitioned from a software revolution to a massive industrial construction project, creating a historic demand for capital that public markets are straining to absorb.
New data reveals that the cost to build the global infrastructure for data centers and AI energy supplies will exceed $5 trillion by 2030. While tech giants generate significant cash flow, the scale of this build-out goes beyond organic balance sheet capacity. Consequently, the market is witnessing a flood of debt issuance. Since September, hyperscalers like Amazon, Alphabet, and Oracle have issued nearly $90 billion in investment-grade bonds, cluttering portfolios and widening spreads as Wall Street struggles to digest the supply.
The Great Risk Offload
A structural shift is occurring in how this infrastructure is financed. To protect their own balance sheets, major technology firms are increasingly offloading the risk of physical construction to smaller intermediaries, developers, and special-purpose vehicles. As reported by the nytimes.com, companies like Microsoft and Meta are utilizing leasing structures to secure computing power without booking the full weight of the associated debt directly.
This has pushed the financing burden into the private markets. According to recent estimates by UBS and Morgan Stanley, private credit is expected to supply over half of the $1.5 trillion needed for data center build-outs through 2028. This represents a pivotal moment for credit investors: the opportunity has moved from funding the technology companies themselves to funding the essential assets—real estate and power—that enable them.
Volatility in the Pipe
The sheer volume of issuance is creating pockets of volatility. Oracle’s credit default swaps (CDS)—insurance against default—recently spiked to levels unseen since 2009, driven by concerns over its aggressive spending plans and the rising cost of capital. Furthermore, the Bank of England has issued warnings regarding the financial stability risks posed by this leverage if valuations in the AI sector correct.
Market fears of a “data center glut” are emerging, with some lenders reportedly financing up to 150% of construction costs based on future valuation uplifts. This exuberance is a classic signal of a market overheating at the fringes, creating a bifurcation between high-quality infrastructure projects and speculative developments.
What This Means for Borrowers
For mid-market heavy industry, construction, and infrastructure service providers, this macro environment presents both a challenge and an opportunity. The “crowding out” effect in public bond markets means that capital for physical assets is becoming more expensive and selective. Banks, wary of regulatory scrutiny and exposure to a single sector, are tightening standards for new construction loans.
The Bond Capital View: We view this not as a software bubble, but as a hardware super-cycle. While public equity markets chase the “magic” of AI, we recognize that the physical build-out—steel, power grids, and HVAC systems—requires patient, flexible debt financing. We are less interested in the speculative valuation of algorithms and deeply focused on the tangible assets required to run them. For borrowers in the industrial supply chain, this is the time to secure distinct, disciplined partners who understand that while technology changes, the fundamental value of essential infrastructure remains constant.
