AI-related borrowing is reshaping the U.S. corporate bond market

Key takeaways

• Technology companies have significantly increased capital expenditures to build the data centers, computing capacity and power infrastructure needed to support artificial intelligence (AI).
• As this spending consumes a growing share of internally generated cash flow, many companies are increasingly turning to capital markets. Technology companies are therefore becoming a larger part of the U.S. corporate bond market
• Technology-sector credit spreads, the additional yield over Treasury securities to compensate for default risk, have widened relative to credit ratings on the bonds. This suggests investors are demanding more compensation for the sector's changing financial profile.
• This shift does not necessarily signal broad credit deterioration. Many large technology companies benefit from strong cash-flow generation, robust liquidity, diversified businesses and access to capital. Instead, the repricing may reflect greater uncertainty around future capital needs, leverage, execution risk and the ultimate returns on AI-related spending.

Technology companies are turning to bond markets to help finance AI investment

U.S. technology-related companies are investing hundreds of billions of dollars in AI infrastructure, including data centers, semiconductors, equipment and electricity. As capital expenditures consume a growing share of operating cash flow, many companies are supplementing internally-generated funds with equity, bonds and other financing arrangements.

This represents a meaningful shift for a sector historically associated with asset-light business models, strong free cash flow, and conservative balance sheets. AI investment is making parts of the technology sector more capital-intensive and, in some cases, more reliant on external financing.

New bond issuance has increased the technology sector's representation in the U.S. corporate bond market. The sector now accounts for roughly 10% of both the $7.5 trillion investment-grade market and the $1.5 trillion U.S. high-yield bond market, as shown below.

Off-balance-sheet obligations are also growing

Traditional debt measures may not fully capture the extent of technology companies' financial commitments. In some cases, long-term contractual obligations are larger than reported balance-sheet debt. These commitments may include plans to lease data centers, as well as agreements to purchase inventory, components, infrastructure, cloud-computing capacity, and electricity.

Although these arrangements differ from funded debt in their legal structure and accounting treatment, they may still represent likely future cash outflows. Credit rating agencies evaluate some of these obligations qualitatively or through adjustments rather than treating them fully as debt. Differences in how rating agencies and investors assess these commitments can therefore lead to diverging views about a company's financial profile. This may become more important as the AI investment cycle matures, particularly if AI demand falls short of expectations for some companies.

Rising debt and investment uncertainty are putting pressure on credit spreads

Bond investors have taken note and are demanding more compensation for the technology sector's changing financial profile. Investment-grade technology credit spreads began to widen relative to the sector's A/A- average credit ratings in late 2025, although they remain tighter than those on BBB rated bonds, as shown below.

This widening suggests that the bond market is pricing risks that are not fully reflected in credit ratings. However, wider spreads do not necessarily mean that market participants expect widespread rating downgrades. Instead, they may reflect a repricing from very strong starting points and recognition that some technology companies may be taking on more capital-intensive and leveraged financial profiles. In addition, higher bond issuance itself can contribute to wider credit spreads by increasing the supply of bonds that markets must absorb, even when fundamentals remain sound.

Importantly, borrowing to fund productive investment is not inherently negative for credit quality. For companies with strong financial profiles, additional borrowing may represent an efficient use of their balance sheets rather than evidence of stress.

A key question is whether the additional investment ultimately generates sufficient incremental cash flow to support larger financial commitments. AI infrastructure spending may create both opportunities and risks, with outcomes likely to vary widely by company. Those with leading market positions, recurring revenue, diversified businesses, substantial liquidity, and demonstrated ability to monetize AI investments may be better positioned for the next phase of the investment cycle. Companies with greater leverage, concentrated business models, and reliance on aggressive demand assumptions may have less room for error.

As the AI investment cycle matures, we expect markets to increasingly distinguish between companies that are spending on AI and those that can earn attractive returns on those investments. Early signs of that differentiation may already be visible in widening gaps in share-price performance. This shift could raise the bar for companies with aggressive investment plans. Large technology companies may have the balance sheets, cash flows and access to capital to sustain elevated spending, but investors appear more likely to scrutinize the timing of returns more closely over the months and years ahead.