Strong demand, robust issuer fundamentals and clearer communication should help the market absorb the expected surge in artificial intelligence-related bond issuance.
There’s a moment in the Wizard of Oz where the terrifying wizard turns out to be a nervous man pulling levers behind a curtain. Once revealed the wizard seems a lot less imposing. We think the same trick applies to one of the credit markets‘ current favourite scare stories: the coming wave of investment grade (IG) bond supply driven by spending on artificial intelligence (AI).1
Granted, the headline numbers do look frightening. Hyperscalers are set to spend around $6 trillion on AI infrastructure over the next five years. Some of that will be funded with debt, and some of that debt will land in IG bond markets. Adding it up you get forecasts of elevated net supply for years to come. Cue nervous conversations about spread widening, market indigestion and the return of the „bond vigilantes“.
We have taken a good long look at this and believe that, once you pull back the curtain, the numbers are far less alarming than the popular narrative suggests.
Debt just needs to keep pace with earnings
Over time, corporate debt tends to grow broadly in line with earnings, as management teams, rating agencies and bondholders constrain leverage. If earnings double, debt can roughly double while the ratio between them remains stable. This is the normal pattern of a growing economy.
For the IG bond index, we expect earnings (EBITDA) to grow by more than 10% annually over the next couple of years. If leverage remains flat, that implies roughly $1 trillion of net new supply each year to the US market. Including refinancing, gross issuance could reach $1.75 trillion-$2 trillion annually. These are large numbers, but they largely reflect bigger company balance sheets and a growing economy.
Figure 1 shows our forecasts for EBITDA growth and leverage for the dollar IG universe. Despite forecasting a modest deleveraging, we still expect more than $1.8 trillion of net new supply between FY25 and FY27.
Figure 1: Dollar IG EBITDA and debt growth ($ billions, LHS) and leverage (RHS)
Source: Columbia Threadneedle Investments’ analysis of Bloomberg data, October 2026
The AI bit is smaller than it looks
Against this backdrop, we estimate that hyperscalers will fund around $4.5 trillion of AI spending from free cash flow and $1.5 trillion through debt. Only about $1 trillion is likely to pass through IG markets. Spread over five years, that is roughly $200 billion of additional annual tech supply annually, albeit with some front-loading. This is a meaningful amount, but manageable for a market that already absorbs trillions of dollars of issuance each year.
More important than the overall volume is its concentration. The supply will come from a small group of very large technology companies, increasing their weight in corporate bond indices much as the “Magnificent Seven” have come to dominate equity indices. That shift warrants attention, but concentration is not inherently dangerous if the underlying businesses remain strong and issuers communicate clearly.
There is also useful historical context. Over the past decade, technology earnings grew much faster than debt, reducing sector leverage from 1.8 times earnings to 0.8 times (Figure 2). Had borrowing kept pace with earnings, these companies might have issued something like an additional $1.1trillion of bonds. AI-related borrowing therefore looks less like a new imbalance than a partial return towards a more normal relationship between earnings and debt.
Figure 2: Technology sector debt ($ billions, LHS) and leverage (RHS)
Source: Bloomberg, as of October 2026
History offers a useful playbook
Previous supply surges offer a useful comparison. After the global financial crisis (GFC), banks issued debt to meet regulatory requirements; telecom companies borrowed to fund network investment; and energy companies funded the shale boom. Each sector increased its share of the bond index by a similar amount, but market outcomes differed. Bank issuance, as required by the regulators, was accompanied by clear, well-signalled disclosures and caused little spread disruption. In telecoms and energy, faster leverage growth and less clarity contributed to sharper widening.
The lesson is that uncertainty matters more than supply alone. Investors are generally more receptive when companies explain how much they plan to borrow, why they need the capital and how they will manage leverage. If technology issuers follow the banks’ example, the market should absorb the additional bonds with less disruption than current fears imply.
Encouragingly, technology companies are becoming more disciplined in their communication with bond investors, including setting explicit guardrails for issuance. Greater clarity makes us more comfortable with the risk and more inclined to view any associated spread widening as a potential opportunity rather than a warning.
Who is buying all these bonds?
The debate also tends to overlook demand, which remains strong. Exporting economies in Asia continue to recycle savings into US bonds. Better-funded pension schemes are natural buyers of long-dated corporate debt, while insurers need high-quality bonds to match growing liabilities. Household inflows into IG bond funds provide further support. In aggregate, demand appears capable of absorbing the expected supply.
So, is the ‘wizard’ actually dangerous?
Not obviously so. AI-driven issuance is real, but it looks less threatening once placed in the context of normal balance-sheet growth and strong demand. The greater risk is not the volume of bonds itself, but whether issuers are transparent about their plans. With clear communication and disciplined leverage, the outcome could resemble post-GFC bank issuance: a substantial, well-signalled increase that the market absorbs without major disruption.
The wizard’s power depended on perception. We suspect concerns about IG supply may also prove overstated once the numbers are examined more closely. The scale is substantial, but the market has absorbed comparable shifts before; this one should be judged on issuer fundamentals, leverage and transparency rather than headline figures alone.
