When most people think about the artificial intelligence revolution, they picture a stock market story: Nvidia's extraordinary run, the Magnificent Seven's trillion-dollar valuations, the flood of venture capital flowing into AI startups. What they rarely picture is a bond story. But while stocks have captured most of the headlines, the AI buildout has simultaneously been reshaping the global debt markets in ways that most ordinary investors have no idea are happening, even though those changes are flowing directly into their retirement accounts. According to a Morgan Stanley forecast reported by Reuters on June 10, 2026, global AI-related debt issuance is on track to nearly double to roughly $570 billion this year, more than twice the prior year's total. That number makes AI-linked debt not just a large segment of the credit markets but, according to M&G Investments data cited across the industry, the single largest sector in the investment-grade bond market, having surpassed even United States banks as the biggest constituent of the JPMorgan US Liquid Index as of October 2025. For the first time in history, the largest investment category in a bond benchmark that anchors trillions of dollars in passive funds is a single transformative technology.

Why Profitable Tech Giants Are Borrowing at Historic Scale

Understanding why this is happening requires understanding a fundamental shift in how the technology industry finances itself. For most of the past decade, the largest technology companies were self-funding machines. Amazon, Alphabet, Meta and Microsoft generated enormous free cash flows that covered most of their investment needs, leaving them among the most financially conservative large borrowers in the investment-grade market. That model has broken down under the weight of AI infrastructure spending.

UBS estimates that hyperscaler capital expenditures in 2026 are on pace to consume nearly 100 percent of operating cash flows, compared with a ten-year average of approximately 40 percent. Amazon, Alphabet, Meta, Microsoft and Oracle are collectively expected to spend somewhere between $700 billion and $805 billion in capital outlays this year alone, depending on the forecast source, with that figure projected to cross $1 trillion in 2027. When spending outpaces earnings at this scale, even the most profitable companies in human history turn to debt markets to fill the gap. Bank of America data shows that Amazon, Alphabet, Meta, Microsoft and Oracle collectively issued $121 billion in US corporate bonds in 2025, compared with an average of just $28 billion per year between 2020 and 2024, a fourfold jump in annual corporate bond issuance from these five companies within a single year. By May 31, 2026, nearly $236 billion in AI-linked debt had been issued globally in 2026 alone, already four times the volume raised over the same period in 2025, putting the Morgan Stanley projection of nearly $570 billion for the full year well on track.

The Apollo and Blackstone Deal That Showed What AI Debt Looks Like in Practice

No single transaction illustrates the new mechanics of AI infrastructure financing more clearly than the deal that Bloomberg first reported on May 28, 2026. Apollo Global Management and Blackstone were working to bring additional investors into a roughly $36 billion debt financing package structured around Anthropic's AI infrastructure expansion, one of the largest private credit transactions ever assembled. The deal was structured through a special-purpose vehicle that borrowed money, received an equity investment, purchased Google's custom tensor processing units, and leased those chips back to Anthropic for deployment at data centers in New York, Texas, Louisiana and Indiana. Broadcom, which helps Google develop the TPU chips, backstopped the largest portions of the transaction through a residual value support agreement, providing the credit quality needed to attract institutional investors to what would otherwise be a concentrated bet on a single AI company's ability to make its lease payments.

The deal ultimately finalized at approximately $35 billion, reported by Bloomberg on June 5, pricing across three tranches: roughly $6 billion in A1 notes, $25 billion in A2 notes, and $4.5 billion of B notes, with Apollo and Blackstone distributing pieces of the debt to outside investors rather than holding the entire position on their own. The significance of this transaction extends well beyond its size. It established AI compute, the chips and TPUs powering artificial intelligence operations, as a genuinely infrastructure-grade asset class in the eyes of private credit markets. Analysts covering the deal have described it as a potential template for future AI infrastructure financings, allowing companies across the industry to access computing power without immediately drawing down equity or issuing public bonds, keeping compute financing structurally separate from a company's core balance sheet. It also gave Google a mechanism to lock in massive, predictable demand for its TPU platform, deepening its strategic relationship with one of the most important frontier AI developers at a scale of commercial commitment that a simple licensing arrangement never could have achieved.

What the Bank for International Settlements Said on June 28

Into this environment, on June 28, 2026, the Bank for International Settlements, the institution often described as the central bank for central banks, released its annual economic report, and the warnings it contained about AI-linked debt deserve careful attention from anyone who holds a bond fund, a target-date retirement account, or any other fixed-income investment product tied to major investment-grade indices.

The BIS named an AI spending bust as one of three major pressure points currently threatening global financial stability, alongside persistent inflation and mounting fiscal stress across sovereign governments. Its concern was not that AI is fraudulent or that the underlying technology lacks genuine value. The report explicitly acknowledged that AI has so far provided important momentum to global growth and could significantly boost productivity over the coming decade. The concern was structural. The five largest hyperscalers are spending more than $1 trillion on AI-related capital expenditure across 2025 and 2026 combined, a sum already outpacing their earnings and free cash flow, forcing some to issue debt to cover the gap. The BIS warned directly that intense competition may have driven investment beyond levels justified by realistic returns, stating in precise terms that disappointment in returns could trigger a sudden pullback in financing and turn the capital expenditure boom into a protracted investment bust with potential knock-on effects on financial conditions.

The report's more specific concern involved where the money is actually coming from. An increasing share of AI investment does not flow through regulated banks, where capital requirements, stress testing and supervisory oversight provide some protection against systemic amplification. It moves through hedge funds, private credit vehicles and other non-bank financial intermediaries that operate with considerably less regulatory scrutiny. Private credit funds originated over $40 billion in loans to AI-related companies in 2025, compared with approximately $3 billion in 2010, a more than tenfold increase in fifteen years. Zhang Tao, the BIS's chief representative for Asia and the Pacific, was explicit about the risk this creates: if the market has any sort of correction, the interconnectedness of the financial system and the interplay of vulnerabilities could mean the speed of a correction could be much faster than previous banking crisis episodes. Non-bank financial institutions now hold 53 percent of total sovereign debt in advanced economies, up from 44 percent in 2021, meaning the same leveraged hedge funds financing AI infrastructure are also the largest holders of government bonds, creating what the BIS calls a new sovereign-financial stability nexus in which stress in any one part of this interconnected system can travel rapidly to the others.

What Ordinary Bond Investors Need to Understand

Here is the practical consequence that most ordinary investors have not yet absorbed: if you own a target-date retirement fund, a bond index fund, or any other broadly diversified fixed-income product, you almost certainly now hold AI infrastructure debt without having chosen to do so, and without necessarily knowing how much of it you own.

This happens through simple index mechanics. Major bond indexes, including the Bloomberg US Aggregate and the JPMorgan US Liquid Index, are weighted by market value. Every eligible new AI bond that a hyperscaler issues increases that company's share of the index, and every passive fund tracking those benchmarks must buy proportionally more. Target-date funds, which held approximately $4.8 trillion in assets at the end of 2025, own those bond index funds. As Morgan Stanley itself has observed, the sheer volume of new AI bond supply has become the dominant force moving bond prices in the investment-grade market, even when the broader economic backdrop holds up, meaning that supply-and-demand dynamics specific to AI infrastructure spending now materially influence the interest rates and bond prices faced by everyone from pension funds to individual retirement savers.

The specific risk this creates is not that AI bonds are necessarily bad investments in absolute terms, since the companies issuing them are, as the BIS acknowledged, among the most profitable businesses in human history, with strong investment-grade credit ratings. The risk is concentration. S&P Global Ratings has already warned that Amazon's leverage will increase substantially and that the company will likely post negative free operating cash flow for at least two years as it supports its data center buildout, a credit concern for a company that is simultaneously one of the largest investment-grade bond issuers in the world. When a single sector grows from a small portion of an index to its dominant segment within a short period, a sector-specific shock begins to behave like a market-wide shock for anyone holding the benchmark. That is precisely what the BIS's annual report was describing when it flagged AI as a systemic risk rather than simply an investment-specific one.

The Historical Precedent Worth Understanding

The BIS's own report drew an explicit historical parallel that investors should take seriously. It mentioned the canal mania of the 1830s, the British railway bubble of the 1840s, and the dot-com crash of 2000, each of which began with a genuine technological breakthrough that attracted more capital than commercial returns could ultimately justify, and each of which ended in economic disruption. Morgan Stanley has separately noted that AI hyperscaler capital expenditure is on track to exceed even the scale of dot-com era telecommunications capex in both magnitude and duration, the spending cycle that directly preceded the most severe technology sector crash since the Great Depression.

None of this proves that AI infrastructure debt is mispriced today, or that the companies issuing it will fail to generate returns sufficient to cover their obligations. The structural difference from 2001, and it is a meaningful one, is that the companies doing the borrowing today are enormously profitable, well-capitalized and cash-generative by historical standards, not loss-making startups with no revenue hoping to become profitable before their runway ran out. But belief in the underlying technology is not the same thing as belief in every financing structure wrapped around it, and the BIS's June 28 report represents the most authoritative voice in global finance stating plainly that the financing structures now supporting AI infrastructure carry systemic risks that regulators and investors do not yet fully understand.

The Bottom Line

The growth of AI-linked debt from a small corner of the bond market into its largest single sector, combined with the extraordinary scale and structural novelty of transactions like the Apollo-Blackstone-Anthropic TPU deal, represents one of the most significant transformations in the global credit markets in a generation. For investors, the most important single implication is that AI infrastructure has moved from a technology story into a financial plumbing story, one that now connects directly to the bond funds inside retirement accounts, the sovereign debt markets where governments raise money, and the privately financed chip-lease deals that determine which AI companies can access enough computing power to compete. The BIS's June 28 warning does not mean this architecture is about to collapse. It means the architecture has grown large and interconnected enough that central bankers felt compelled to formally name it as a systemic risk in their most important annual publication. For ordinary investors, that is the signal worth paying attention to.

*This article is for informational purposes only and does not constitute financial or investment advice. Always conduct your own research or consult a licensed financial advisor before making investment decisions. Data referenced is sourced from Morgan Stanley via Reuters, Bloomberg, Yahoo Finance, the Bank for International Settlements, M&G Investments, UBS, Bank of America, S&P Global Ratings, and Fortune as of June 28, 2026.*