Google, Meta, and other AI 'hyperscalers' are embarking on a $1 trillion debt spree after years of generating massive cash flows. This is why the shift toward borrowing by major tech companies is significant.

Historic Patterns of Capital Investment Booms

Throughout the last two centuries, major waves of capital investment have often concluded with bankruptcies, industry consolidation, and significant losses—yet those who endure frequently emerge stronger. For example, the late 1990s saw enormous investments in fiber-optic infrastructure, with billions spent to lay cables across continents and oceans. This period ended with the collapse of companies such as WorldCom and Global Crossing. Similarly, the U.S. shale boom led oil and gas firms to accumulate $350 billion in debt, only to see hundreds of bankruptcies after oil prices plummeted in 2014 and 2015. Going further back, the early 20th century’s rapid electrification resulted in about half of the 3,000 small utilities and power companies either vanishing or being acquired during a decade of intense consolidation. Despite these upheavals, the survivors benefited from inherited infrastructure, lower costs, and greater market power.

The Rise of AI and Massive Capital Outlays

Today, artificial intelligence is at the center of a new investment surge. Five tech giants—Alphabet, Amazon, Meta, Microsoft, and Oracle—are leading the charge, pouring unprecedented sums into building the vast data centers and cloud infrastructure that AI demands. According to a recent Moody’s report, these companies have collectively committed $969 billion, with $662 billion earmarked for future data center leases. While much of this spending is funded through operational cash flow, the scale has prompted a shift toward issuing bonds to bridge the gap between capital expenditures and free cash flow.

In 2025 alone, these five firms raised approximately $121 billion in new bond debt, a dramatic increase from $40 billion in 2020. Wall Street forecasts suggest that AI-related bond issuance could total between $100 billion and $300 billion this year. Over the next several years, total investment in data centers may reach as much as $3 trillion.

These developments are reshaping the financial landscape for internet companies, many of which previously operated with little or no debt. The influx of bond financing introduces new stakeholders, obligations, and risks, fundamentally altering how these businesses are managed and valued. Unlike equity investors, bondholders prioritize fair compensation for risk, including the dangers of overbuilding and excess supply.

Expert Perspectives on Investment Cycles

Mohit Mittal, chief investment officer of core strategies at Pimco, notes that every major capital spending cycle in history eventually faces the risk of overinvestment. He cautions that the next couple of years could see excessive investment, potentially leading to corrections or slower growth.

This debt-driven expansion is also changing the financial profiles of tech companies. Mittal points out that companies with asset-light models typically enjoy higher equity valuations, while those with more tangible assets tend to see lower multiples.

Long-Term Investment Strategies

At the close of 2025, Alphabet, Oracle, and Meta issued over $70 billion in bonds within a few weeks. The broader investment-grade bond market saw nearly $500 billion flow into taxable bond funds last year, with foreign investors purchasing about $304 billion in U.S. corporate bonds, according to Breckinridge Capital.

Kevin SigRist, chief investment officer for North Carolina’s $143 billion pension system, highlights that the yields on bonds from these tech giants hover around 5%, making them attractive, especially given the companies’ strong financial positions. However, he notes that the sector’s bond spreads are tight, and there is little additional yield for longer maturities. Despite this, these bonds remain appealing to traditional buyers like pension funds and life insurers.

These investors are well-matched to the hyperscalers, which generally boast top-tier credit ratings and issue bonds with maturities stretching 30, 40, or even 100 years. In February, Alphabet issued a rare 100-year bond, a first for a tech company in decades. More details about the buyers are expected as disclosures are released throughout the year, but SigRist anticipates strong demand for these offerings, similar to what was seen in 2025.

Anders Persson, chief investment officer at Nuveen, observes that investors’ willingness to purchase ultra-long-term debt signals confidence in the risk-reward balance of these opportunities.

For instance, Alphabet’s long-term debt rose from $10.9 billion at the end of 2024 to $46.5 billion by the end of 2025, while its cash reserves reached $126.8 billion. With a market capitalization of about $3.6 trillion, total obligations represent just over 3% of the company’s value, even when including future lease commitments.

How This Cycle Differs from the Past

Persson, who analyzed tech companies during the dot-com bubble, notes a stark contrast between then and now. During the dot-com era, many issuers had little or no revenue or free cash flow, making the bubble’s collapse almost inevitable. Today’s tech leaders—Alphabet, Microsoft, Amazon, and Meta—have robust balance sheets and strong cash flows, making them far more resilient. Even significant missteps in capital allocation are unlikely to threaten their solvency.

“This time is different,” Persson remarks, acknowledging the cliché but emphasizing the more cautious approach being taken today.

Among the five hyperscalers, Oracle stands out with a Baa2 credit rating, just above junk status. Lower ratings mean higher default risk and require issuers to offer greater yields to attract investors. Oracle has already taken on more debt than its peers, with over $248 billion in future data center lease commitments and $124 billion in borrowings. In 2025, Oracle issued $25.8 billion in notes maturing as late as 2065 and recently announced plans to raise an additional $45–50 billion this year through both debt and equity. According to Bloomberg, Oracle is also planning significant layoffs to manage its cash needs for data center expansion.

The Momentum and Risks of Massive Investment

Large-scale investment cycles often create a self-reinforcing momentum, as companies fear missing out more than they fear making a wrong bet. The danger is that the industry may build more capacity than the market can immediately absorb, a pattern seen repeatedly in history.

Alphabet’s Ruth Porat (left) tours a newly opened data center in England in 2025.

Moody’s has cautioned investors to consider both on-balance sheet debt and economic debt from future lease commitments when assessing risk. For example, Alphabet and Meta, despite their strong credit ratings, had to pay premiums over their existing debt to complete recent deals, reflecting both their ambitious plans and the market’s caution about the volume of debt expected in the coming years, as noted by Janus Henderson analysts.

For Persson, the key question is not whether the risks are too high, but whether the bonds are priced to fairly compensate investors for the full spectrum of risks associated with nearly $1 trillion in commitments.

Across the five hyperscalers, on-balance sheet debt totals about $420 billion, with even larger obligations tied up in leases, many of which have yet to begin. According to GAAP, only leases that are “reasonably certain” to be renewed appear as liabilities, so much of this information is buried in financial statement footnotes. Nonetheless, bond fund managers are factoring these commitments into their analyses.

Without considering leases, these companies’ leverage remains low, and as of the end of 2025, they held more cash than debt. Including lease obligations, leverage is still modest but rising, signaling a potential area of concern for the future.

“We’re factoring these commitments into our assessments, especially given their size and growth potential,” Persson explains. “Ultimately, these are obligations that must be honored, so we treat them as debt when evaluating credit quality.”

The risks are real: companies face both economic and balance-sheet debt, a shift from asset-light to asset-heavy models, and the possibility that revenues may not keep pace with spending. The stock market’s sensitivity to AI news means that each quarter brings uncertainty about how shares will react.

Despite these risks, the leading firms are financially robust enough to weather potential missteps—unlike the casualties of previous investment booms. However, the true outcome will only become clear in hindsight.

As Mittal from Pimco puts it, “You only find out after the fact. If you start to see it ahead of time, others do too, and investment naturally slows.”

“Every company’s experience will be unique,” Mittal adds. “There will be both winners and losers as this unfolds.”

This article was originally published on Fortune.com.