Weighing higher bond yields against influx of AI debt

[SINGAPORE] Until recently, the artificial intelligence narrative has consumed the equity market. But now, AI is pursuing the bond market, where debt financing among hyperscalers and companies involved in the AI infrastructure build-out is surging.

This raises challenges for investors, particularly those who invest in fixed income as an anchor in portfolios.

The big question is: Will the debt pipeline become a structural feature of credit markets?

For now, few are concerned about credit quality, as the hyperscalers themselves are mostly rated investment grade. Still, if the issuance pipeline is structural in nature, what risks are raised when corporate bond indices begin to skew towards technology?

Between 2020 and 2024, five hyperscalers combined – Alphabet, Amazon, Meta, Microsoft and Oracle – issued an average of roughly US$35 billion of debt each year, said investment-management company Vanguard. In 2025, that figure jumped to US$93 billion.

Vanguard estimates total AI-related debt issuance for FY2026 – including chipmakers, data-centre developers and utilities – may range between US$300 billion and US$570 billion.

This is occurring amid a spike in yields in the sovereign bond markets, as investors reckoned with rising budget deficits in the developed world and governments’ need to raise even more capital to fund various objectives, including defence.

These factors, alongside the ongoing US-Iran conflict and inflation concerns, recently pushed Group of Seven yields to their highest levels in decades.

In the US, 30-year US Treasury yields touched 5.33 per cent in August.

German 30-year yields reached post-2011 highs, and French 30-year yields also rose to their highest since the global financial crisis. In Japan, 10-year yields have hit a 30-year high.

Over the weekend, US Federal Reserve chairman Kevin Warsh signalled a determination to cool inflation, raising the likelihood of a quarter-point rate hike in September. Two-year Treasury yields edged higher; 30-year yields now hover at around 5.27 per cent.

How should retirement portfolios with substantial fixed-income allocations navigate this environment? As bond yields rise, what are the possible knock-on effects on equity markets?

Daryl Ho, senior investment strategist at DBS, said some factors may be pushing global equilibrium interest rates higher, including elevated fiscal deficits and rising capital demand from AI investments.

“This does not necessarily lead to a reset in risk. Rather, the hurdle rate of return for expectations should naturally be higher when the ‘risk-free’ rate is also elevated,” he added.

“For equities, the impact would likely be uneven rather than universally negative. Companies with genuine earnings growth and pricing power can offset higher discount rates. Conversely, companies whose valuations depend heavily on distant future cash flows are far more vulnerable.”

Silver lining in higher yields

Pictet chief investment officer for Asia, Kelvin Tay, said higher yields reflect a “broader structural reality” as investors demand higher compensation to lend over the long term.

“Persistent government budget deficits and ongoing fiscal largesse exacerbate this situation further and translate into higher government term premia,” he added.

“Structurally, higher government borrowing is likely to compete with companies’ efforts to finance their investment needs – a trend that has already begun.”

Ho said most of the pain inflicted on bond investors this year was not from credit risk, but duration risk. That is, bonds’ sensitivity to rates has mattered more than the risk of default.

He noted that in the current higher-yield environment, investors should temper their expectations of capital appreciation from bonds.

“In a world of structurally higher rates, bonds may not deliver the same capital gains that investors became accustomed to in the decades after the (global financial crisis). Portfolios should hold bonds primarily for consistent income generation and not capital appreciation,” he said.

Cyrus Ng, senior analyst at FSM Global’s fixed-income team, said investors may be demanding more term premia.

“We encourage balancing risk and return across any portfolio. It is important to be deliberate about duration. Investors should balance any added duration and (ensure) the yield pickup is fair for the additional duration risk.”

He preferred short-to-medium over long-term bonds, adding: “There is not much pickup as we go into longer tenors, especially with heightened duration risks and larger mark-to-market fluctuations.”

Yields rise when bond prices fall. Existing investors incur a paper loss when this happens, but interim price volatility should not matter for those who intend to hold a bond to maturity.

Bonds with long maturities carry a relatively higher duration risk – that is, they are more sensitive to interest-rate changes.

James Cheo, managing director and chief investment officer for Southern Asia and Australia, global wealth management at UBS, said there is an important offset to higher yields.

“Higher starting yields improve prospective bond returns, and provide a larger income cushion against further increases in rates,” he noted.

He added that as yields on the long end could be volatile, “there is a case for concentrating more exposure in the two to five-year part of the curve”.

“This can capture much of the available income, while taking less interest-rate risk than long-duration bonds. For retirees, this is important, because the objective is not simply maximising total return,” he said.

“It’s about generating predictable income, while limiting the risk of having to sell assets after a large mark-to-market decline.”

AI-related issuance

Large hyperscalers are generally fundamentally strong, said Cheo, with high investment-grade ratings, substantial free cash flow, significant liquidity and relatively low leverage.

“For these issuers, the immediate risk is less about default and more about supply, spreads and relative value.”

Being selective will be key. “As issuance accelerates, even high-quality borrowers may have to offer larger concessions to attract investors. Good credit does not mean good value at every price,” Cheo said.

But even as the AI investment cycle is creating opportunities, it is also reshaping corporate financing, noted Boh Hui Ling, Bank of Singapore’s head of fixed-income strategy, “making traditional leverage metrics less effective at capturing the full extent of credit risk”.

Investors, she added, will have to look beyond balance-sheet debt and assess issuer fundamentals, cash-flow generation, capital requirements and potential off-balance sheet obligations.

“Ultimately, successful investing in the AI era will depend less on broad thematic exposure and more on rigorous credit analysis, selective issuer positioning and disciplined portfolio construction,” she said.

Kylie Soh, Fullerton Fund Management’s client portfolio manager for fixed income, said the key question is whether AI investment translates into sustainable cash-flow generation and returns on invested capital.

“Large AI-technology companies generally retain strong balance sheets and healthy liquidity. But the credit analysis becomes more demanding when capex rises rapidly, financing shifts towards debt or an increase in off-balance-sheet obligations, such as lease and purchase commitments,” she said.

Ho sounded a note of prudence.

“Bond investors should remain well-diversified, investing beyond just hyperscaler debt, to prevent any one instance of a negative credit event putting capital at risk, given that fixed-income instruments typically cap upside, while exposing investors to the full downside in capital invested.”

He also suggested investors should move from bond exchange-traded funds, which track indices to actively managed bond funds.

“Market cap-weighted bond ETFs are typically ‘loser-takes-all’ structures, in the sense that the most indebted companies get higher weights in the indices, exposing bond investors to concentration risk relating to the least creditworthy issuers.

“Active management would allow more discretion in allocation, giving investors a more intentionally diversified portfolio.”