How AI Debt Is Pushing Up Bond Yields—and Why Pimco Sees Opportunity

The AI buildout is no longer only an equity-market story. It is also reshaping credit markets as technology companies borrow at unprecedented speed to finance data centers, computing capacity and related infrastructure.

Pimco’s Marc Seidner described the result as “too much, too fast” issuance and said it is causing “indigestion” in fixed-income markets. He said it was “very possible” that the borrowing wave helped push the 10-year U.S. Treasury yield to about 4.75% earlier in August—but that is a contribution, not a complete explanation.

The evidence: Big Tech has become a major bond issuer

The clearest sign of the shift is the scale of new supply. Amazon, Alphabet, Meta and Oracle issued about $194 billion of bonds through July 7, 2026, compared with roughly $108 billion during all of 2025, according to a Reuters analysis of LSEG data. 17

Amazon’s July offering illustrates both the size of the financing need and the market’s continuing appetite for highly rated technology debt. The company sought $25 billion across eight tranches, with maturities ranging from 2029 to 2066. Reported peak demand reached about $62 billion. 18

Other estimates produce different totals because they use different issuer groups, dates and definitions of AI-related debt. Morgan Stanley forecast nearly $570 billion of global AI-related debt issuance in 2026, up from about $236 billion through May 31. Goldman Sachs separately estimated nearly $500 billion of AI-related debt issuance so far in 2026, with hyperscalers accounting for about 40% of that total.

The figures are therefore not interchangeable. They do, however, point in the same direction: AI infrastructure has become a substantial source of bond-market supply.

Why more corporate debt can lift yields

The market-clearing mechanism is a supply-and-demand story.

Long-dated corporate bonds require investors to absorb more duration, or sensitivity to changes in interest rates. When hyperscalers issue large amounts of long-maturity debt at the same time that governments are selling substantial quantities of Treasuries, investors have to decide how much of that additional rate exposure they want.

If demand does not rise as quickly as supply, issuers generally need to offer higher yields or wider spreads. Bond prices and yields move in opposite directions, so weaker demand can translate into lower prices and higher yields across parts of the market. New corporate deals may also need a concession—extra yield compared with existing bonds—to attract buyers.

Reuters reported that the AI issuance wave was testing investor demand, while its market analysis said the increase in corporate supply was one factor affecting longer-term Treasury yields. Pimco’s own credit research has also highlighted record hyperscaler issuance in dollar markets and the growing weight of these companies in investment-grade indexes.

That does not mean every dollar borrowed by a technology company directly raises the Treasury yield. Corporate bonds and government bonds are different instruments. But they draw on overlapping pools of capital, and investors often compare their relative yields, credit quality and duration.

What the evidence does—and does not—prove

Several estimates circulating in the market put AI borrowing at roughly $182 billion in the first half of 2026, with a possible total near $400 billion for the year. Other forecasts are higher or lower. The differences reflect whether analysts include Microsoft, data-center operators, private credit, leases, project finance, foreign-currency borrowing and debt issued by suppliers or related entities.

The evidence supports three conclusions:

• AI-related borrowing has increased sharply.
• The new supply is concentrated in long-duration, investment-grade credit that competes with other fixed-income assets.
• Investors are becoming more attentive to pricing, leverage and the pace of issuance.

It does not establish that AI debt alone added 10 to 20 basis points to long-term U.S. yields. Reuters’ commentary citing Goldman Sachs put the broader effect closer to roughly 5 basis points, describing the impact as marginal relative to other forces.

The same caution applies to the claim that AI borrowing pushed the 10-year Treasury yield to 4.75%. Pimco said that was possible, but its wording does not establish a precise causal contribution. Inflation expectations, Federal Reserve policy, Treasury issuance, fiscal concerns, term premia and global demand for safe assets can all move long-term yields at the same time. Reuters likewise said those cyclical and fiscal forces matter more than AI financing for the overall direction of rates.

The South Korean spillover: stronger chips, weaker bonds

AI’s financial effects also reach economies that supply the infrastructure. South Korea’s first-quarter growth was supported by a 5.1% increase in exports, led by IT components including semiconductors used in AI infrastructure. Its three-year government-bond yield rose 8.8 basis points on the release. 3

The transmission channel is different from the U.S. corporate-debt story. Stronger semiconductor exports can lift Korean growth, investment and inflation expectations. Investors may then expect the Bank of Korea to keep policy tighter or raise rates, which pushes government-bond yields higher.

Bloomberg reported that South Korean local-currency government bonds had lost 7.5% in 2026 by early June, the weakest performance among 44 markets it tracked, while the benchmark three-year yield reached about 3.9%. 1 Analysts cited the semiconductor boom and its inflationary implications as reasons yields could rise further. 2

This is an important distinction: in South Korea, the AI boom can pressure bonds not only through global competition for capital but also through stronger domestic growth and expectations of tighter monetary policy.

Effects on companies, mortgages and growth stocks

Higher benchmark yields raise the hurdle rate for many forms of borrowing. U.S. mortgage rates are influenced by Treasury and mortgage-backed-security markets, while corporate borrowers typically price debt relative to government benchmarks. A persistent rise in long-term yields can therefore increase financing costs for households and businesses.

The effect may be uneven. Highly rated hyperscalers can still attract deep demand, but they may have to pay more than investors previously expected. Smaller companies, highly leveraged borrowers and projects with uncertain future cash flows can face wider spreads, larger issuance concessions or reduced market access.

For companies financing their own AI expansion, higher rates create a second test: whether the expected revenue and productivity gains will justify the cost of the infrastructure. Pimco has warned that the scale of the investment cycle is large while the ultimate size of the buildout remains uncertain.

Higher yields also affect growth-stock valuations. The value of a company depends partly on the present value of its expected future cash flows. When discount rates rise, distant cash flows become less valuable today, placing particular pressure on companies whose investment returns are expected far in the future. A debt-funded AI strategy adds a balance-sheet question to that valuation question: will future earnings grow fast enough to offset the new obligations?

The stronger counterargument: fiscal deficits are the main driver

The most important alternative explanation is that U.S. fiscal deficits and Treasury supply—not corporate AI borrowing—are the dominant sources of upward pressure on long-term yields.

That argument has a strong foundation. The federal government issues vastly more debt than any individual technology company, and long-term Treasury yields also reflect inflation risks, expectations for Federal Reserve policy, the term premium and investor demand for safe assets.

Pimco’s own separate analysis has argued that the recent rise in longer-dated Treasury yields was driven more by shifting policy expectations than by a meaningful AI-related repricing of the term premium. Reuters’ analysis similarly concluded that AI debt should not be treated as the sole or dominant explanation for bond-market weakness.

The most defensible interpretation is therefore that AI borrowing is an amplifier. It adds a large, cyclical source of corporate duration supply to a market already dealing with heavy sovereign issuance and uncertainty about inflation and policy.

Why Pimco sees a potential opportunity

The near-term effect of heavy issuance is uncomfortable for existing bondholders. When yields rise, prices of previously issued bonds generally fall, creating mark-to-market losses for investors holding longer-duration securities.

But higher yields also improve the starting income available to new buyers. If issuance eventually slows, inflation moderates, growth weakens or the market absorbs the supply, yields could decline. Bond investors would then have the potential to earn both coupon income and price gains.

That is Pimco’s opportunity thesis: the AI financing boom may create short-term volatility and pricing pressure, but it can also reset bond-market returns at more attractive levels. The key is selectivity. Investors must distinguish financially resilient issuers from companies or projects whose leverage depends on aggressive assumptions about AI demand, cash flow and future refinancing.

The broader lesson is not that AI has single-handedly caused the bond selloff. It is that the technology boom has become large enough to affect the supply, pricing and risk structure of fixed-income markets. For now, “too much, too fast” describes a market-absorption problem. Over a longer horizon, that same problem could leave bond buyers with more income and better opportunities—provided they can withstand the adjustment.