AI infrastructure may "burn" up to $5.5 trillion; JPMorgan: Bond market can absorb bond issuance surge, tech giants can still increase leverage

According to Zhitong Finance APP, as tech giants ignite a bond issuance boom to build AI data centers, the market is increasingly concerned about whether the US investment-grade bond market can absorb the surging debt supply. However, Stephanie Aliaga, Global Market Strategist at J.P. Morgan Asset Management, believes that the current leverage ratios of these hyperscale cloud companies remain relatively low. Strong demand for AI computing power continues to support future cash flow, so the bond market is fully capable of absorbing the new issuances. J.P. Morgan estimates that the six largest hyperscale cloud providers could even add roughly $1.5 trillion in debt on top of current levels without putting significant strain on their financials.

Currently, bonds issued by these six major cloud companies account for about 5% of the US investment-grade bond index, which has doubled over the past two years. As investment in AI infrastructure continues to grow, these tech giants are rapidly gaining influence in the global bond market.

Aliaga said in an interview on Tuesday: “We believe they will continue to issue bonds.”

The Share of Six AI Giants’ Bonds Doubles in Two Years, Still Able to Add Another $1.5 Trillion in Debt

Building AI data centers requires massive capital investment, so tech companies are increasingly turning to the bond market to raise long-term funds. Currently, the six largest hyperscale cloud companies account for about 5% of the US investment-grade bond index, double the proportion from two years ago. The rapid increase in bond supply has also led some investors to worry that tech companies may exert growing supply pressure on the investment-grade bond market.

However, J.P. Morgan Asset Management believes that these companies’ balance sheets still have ample room for expansion. Aliaga pointed out that leverage ratios for the six hyperscale cloud providers remain significantly lower compared to the overall investment-grade bond market. J.P. Morgan estimates these companies could relatively easily add about $1.5 trillion in additional debt compared to current levels.

The firm expects that as investment in AI infrastructure continues, large tech companies will continue to utilize the bond market for financing. Aliaga stated that debt itself is not necessarily a bad thing. For hyperscale cloud providers building data centers expected to remain operational for five, ten, or even more years, debt can in reality be a very attractive way to finance.

US Bond Yields Reach Highest Levels Since 2008; Tech Bond Issuance Boom Sparks ‘Buyer Shortage’ Concerns

Aliaga’s comments come as the global bond market is under clear pressure. Amid market concerns about persistently high US inflation and expectations that major central banks will further raise interest rates, US government bond yields have climbed to their highest since 2008. At the same time, the boom in AI infrastructure construction is prompting large tech companies to accelerate the issuance of investment-grade bonds.

The simultaneous expansion of government and tech sector financing needs has raised worries among investors about whether the bond market has sufficient buying power to absorb the new supply.

But Aliaga believes such worries may be overstated. She said, “We think the market is fully capable of absorbing this new bond issuance. If anything, this could actually help make the AI boom more sustainable.”

In other words, if the bond market can continue providing large tech companies with long-term, relatively stable financing channels, then building AI data centers will not have to rely entirely on corporate cash flows, helping extend the current AI capital expenditure cycle.

Anthropic Has Over $175 Billion in Compute Contracts, AI Infrastructure Spending Keeps Growing

AI computing power remains one of the most scarce resources in the global tech sector. For example, Anthropic, an artificial intelligence company, has already committed to over $175 billion in cloud computing power contracts, underscoring the huge scale of AI industry demand for computational infrastructure.

With large tech companies, AI labs, and cloud providers all competing for GPUs, servers, storage, networking, and data center resources, capital requirements across the AI supply chain are rising rapidly.

This financing need is not only pushing tech companies to issue more bonds, but also starting to compete with the US government and other sovereign issuers for funds in the global fixed income market.

However, Aliaga believes that AI demand itself is also providing important support for this debt.

Currently, operating cash flows of hyperscale cloud providers roughly cover their capital expenditures. More notable, however, is that the contract backlog signed by the three largest hyperscale cloud companies is growing faster than their capital expenditure.

Aliaga believes this is a positive sign, as it shows that huge investments in AI infrastructure are increasingly backed by future customer demand, raising the likelihood these projects will ultimately generate returns.

Global AI Infrastructure Investment Could Reach $5.5 Trillion by 2030

J.P. Morgan estimates that by 2030, total global investment in AI infrastructure could reach a staggering $5.5 trillion. Such enormous capital needs mean that even the world’s largest tech companies, with their strong cash-generating abilities, would find it difficult to fund all investments relying solely on operating cash flows.

Aliaga noted that hyperscale cloud providers’ own cash flows can only cover part of the $5.5 trillion in investment needs, so debt financing and other forms of external capital will play an increasingly important role in the future.

In addition to the public bond market, alternative capital sources such as private credit may also become important in financing AI infrastructure. This means the AI investment boom could increasingly spill over from the tech equity market into the bond and private credit markets.

For investors, evaluating the AI capex cycle will no longer simply be a matter of judging how much tech companies are willing to invest, but also whether the global capital markets can continue to provide enough financing for these projects.

AI Capex Will Eventually Slow, Profit Margins and Free Cash Flow Likely to Improve

Aliaga believes the current rapid growth in capital expenditure by hyperscale cloud providers will not persist indefinitely. As AI infrastructure is gradually built out, capex growth will eventually slow down. At that point, profit margins and free cash flow pressures for big tech companies are expected to ease.

She commented that when capital expenditures eventually decelerate, “it should provide some relief to profit margins and free cash flow.” However, at least in the foreseeable future, tight supply of AI computing power will remain difficult to resolve completely.

Therefore, the truly critical question for the next stage of the AI investment cycle may not be whether demand exists, but rather who will be first to overcome compute power bottlenecks and when new capacity will actually come online.

Aliaga specifically highlighted that memory supply constraints are one of the main bottlenecks currently facing AI infrastructure expansion. For investors, it will be important in the future to assess which companies are first to overcome supply limits in critical components like memory and bring new AI computing power to market.