The growing reliance on debt to finance artificial intelligence (AI) investments could increase volatility in global corporate bond markets in 2027, as rising borrowing by technology giants heightens market concentration risks and potentially tightens funding conditions for companies, Fitch Ratings said on Wednesday.

The ratings agency said in a report that it expects AI-linked bond issuance by investment-grade (IG) non-financial corporates to remain elevated next year, although the pace could become more sporadic as investors become increasingly selective about their exposure to the sector and borrowing costs rise.

Fitch said the growing concentration of AI-related companies in the investment-grade corporate bond market could amplify market volatility, particularly if the credit profiles or equity valuations of major hyper scalers weaken.

AI-related sectors could account for 43 percent of outstanding US investment-grade non-financial corporate bonds in 2027, up from 38 percent in 2023, based on Fitch’s estimates.

The increasing share reflects the substantial financing requirements of hyper scalers as they expand data-center infrastructure and computing capacity to support the development and deployment of AI technologies.

Fitch said highly rated hyper scalers were increasingly using debt to fund a material portion of their capital expenditure (capex), changing the composition of the investment-grade corporate bond market.

The agency warned that greater concentration could heighten funding risks for AI-exposed sectors, particularly if investors become less willing to absorb large volumes of new debt without demanding higher returns.

This could also increase refinancing risks for companies with significant debt maturities in the coming years, especially in sectors such as healthcare, technology and utilities.

US non-financial investment-grade corporate bond issuance reached $940 billion in the ten months to end-October 2026, up 33 percent from the corresponding period last year, according to Fitch’s estimates, driven largely by a surge in AI-related borrowing.

The agency noted that debt issuance by major hyper scalers — Amazon, Alphabet, Meta, Oracle and Microsoft — had increased substantially, from less than $50 billion annually between 2020 and 2024 to about $108 billion in 2025 and approximately $277 billion in 2026 as at Nov 11.

The companies also issued another sizeable batch of debt in early November, including $25 billion and €9 billion by Meta, €6 billion by Alphabet, $25 billion by Microsoft and $10 billion by Oracle.

Much of the borrowing has been structured with long maturities, with about 61 percent of hyper scaler debt issued in 2026 carrying maturities of more than seven years, while 28% had maturities exceeding 20 years.

AI investment to sustain borrowing

Fitch expects hyper scaler borrowing to remain substantial in 2027 as major technology companies continue expanding their AI infrastructure.

Market consensus projections indicate that combined capex by the five major hyperscalers could exceed $1 trillion in 2027, compared with $772 billion in 2026.

The projected spending includes approximately US$322 billion by Alphabet, US$299 billion by Amazon, $212 billion by Meta, US$205 billion by Microsoft and US$91 billion by Oracle.

However, Fitch’s own capex projections are lower than market consensus estimates.

The agency expects the investments to be financed through a combination of operating cash flow, debt and equity, although pressure on hyperscalers’ free cash flow suggests that bond issuance will likely remain high.

Assuming 25 percent of their capex is financed through corporate debt, broadly in line with 2026 levels, Fitch estimates that the five companies could contribute another $280 billion in bond supply to the US investment-grade non-financial corporate market in 2027.

Fitch does not expect increased regulatory uncertainty surrounding US data-center development to trigger a major pullback in capex next year.

However, the agency cautioned that tighter development controls could slow new projects, particularly following the November 2026 US midterm elections.

Stricter controls on data-centre development were key campaign proposals among winning gubernatorial candidates in Texas, Georgia and Ohio, which together account for more than half of the US data-center development pipeline.

Delays in permits and approvals, as well as the possible removal of tax incentives, could extend project timelines and increase development costs.

Fitch said a slowdown in state-level approvals could weigh on AI-related sectors and reinforce its expectation that AI-related capex would peak in 2027.

It also expects a period of reassessment of capital spending plans in 2028 and 2029, amid weaker prospects for monetising AI investments.

Investor appetite and borrowing costs under pressure

Beyond the volume of new debt, Fitch identified investor appetite as another potential source of risk for the corporate bond market.

Although demand for hyper scaler debt remained strong in 2026, investors could become more selective as they seek to manage their exposure to AI-related investments.

The agency said large investment funds and asset managers were likely to monitor their exposure carefully, particularly if concerns over the returns on AI investments persist.

Such caution could eventually lead investors to demand higher risk premiums on debt issued by hyperscalers and other investment-grade companies with significant AI exposure.

Fitch noted that corporate bond spreads in the US and Europe remained relatively low by historical standards, although they had widened modestly since mid-September.

The agency also highlighted rising issuance costs among hyperscalers.

For instance, Meta’s notes issued in early November 2026, maturing in 2056, carried a yield premium of 2.2 percentage points over US Treasury securities, approximately 0.9 percentage points higher than an equivalent note issued by the company in April 2026, according to Bloomberg data cited by Fitch.

The agency said the development could indicate that the market’s capacity to absorb the substantial volume of debt issued by hyperscalers was being tested.

Higher borrowing costs could also influence companies’ capital allocation decisions, particularly those involving new investments, mergers and acquisitions (M&A), and financing strategies.

Fitch added that corporate borrowing costs were influenced not only by credit spreads but also by underlying sovereign bond yields, which had risen in both the US and Europe during 2026.

Under its base-case assumptions, the agency expects the 10-year US Treasury yield to remain at 4.8 percent at end-2027, unchanged from end-2026, while the 10-year German Bund yield is projected to ease slightly to 3.2 percent from 3.3 percent.

However, recent developments have increased the possibility of further US Federal Reserve rate increases, potentially putting additional upward pressure on Treasury yields.

Companies facing refinancing exposure

Companies with substantial debt maturities and significant AI exposure are likely to be particularly vulnerable if bond market volatility increases and investors become more selective, Fitch said.

Just over $400 billion of US investment-grade non-financial corporate debt is due to mature in 2027, followed by slightly more than US$500 billion in 2028.

Healthcare and pharmaceuticals, technology and utilities account for a combined $355 billion, or about 38 percent, of the roughly $925 billion in debt maturing over the two years.

These companies may also face higher refinancing costs as they replace existing lower-coupon debt with newly issued bonds carrying higher interest rates.

Nevertheless, Fitch said investment-grade issuers generally had sufficient financial headroom to absorb higher borrowing costs.

The agency expects earnings before interest, taxes, depreciation and amortisation (EBITDA)-to-cash-interest coverage for its rated investment-grade corporate issuers across North America and Europe, Middle East and Africa to remain relatively strong at approximately 8.8 times and 10.7 times respectively in 2027.

However, this financial buffer could narrow if higher interest expenses coincide with weaker earnings amid more challenging macroeconomic conditions.

Fitch does not anticipate such a deterioration in its base-case outlook, but warned that the combination of rising financing costs and weaker corporate earnings could reduce companies’ capacity to service debt.

Higher interest rates challenge APAC digital banks, says Fitch