Wall Street's largest banks are sounding increasingly urgent alarms over the scale of debt financing the artificial intelligence buildout, with Morgan Stanley estimating that global AI debt issuance will nearly double to $570 billion in 2026. The figure, reported by The Hindu BusinessLine, Tech Times, and ANI News, underscores how thoroughly bonds, rather than equity or operating cash flow, now fund the data centers, chips, and energy infrastructure underpinning the AI boom.
The warnings have grown sharper in recent days. Goldman Sachs has raised an internal alarm that an "AI debt tsunami" is looming as the market's capacity to absorb new issuance nears collapse, according to finance.biggo.com. Forbes followed with a report titled "Bond Investors Push Back As AI Debt Heads Toward $570 Billion," documenting growing resistance from fixed-income buyers being asked to underwrite years of spending before meaningful revenue arrives. For more context on this story, see our ongoing coverage of the AI industry's financial foundations. For more context on this story, see our ongoing AI news.
How Big Is the Number?
Morgan Stanley's estimate, first surfaced in June and reinforced through July, projects that AI-related debt issuance will roughly double compared with 2025, reaching around $570 billion this year. BusinessLine notes that bonds are now the primary mechanism funding the buildout, a shift that places the AI cycle's health squarely in the hands of credit markets rather than stock investors.
That dependency matters. Unlike equity, debt must be serviced with interest regardless of whether the underlying AI products succeed, meaning every quarter of delayed monetization tightens the squeeze on borrowers. Crypto Briefing and The Cryptonomist, reporting in mid-July, described investors as "wary" of the surge, citing uncertainty over whether the revenue required to repay the borrowing will materialize on schedule.
Goldman's 'Absorption Capacity' Warning
The most pointed caution has come from inside Goldman Sachs itself. According to finance.biggo.com, the bank's analysts warn that the bond market's absorption capacity, the volume of new debt investors are willing to buy at acceptable yields, is nearing its limit. When absorption capacity is exhausted, issuers are forced to offer higher yields, accept tighter terms, or postpone issuance entirely, raising their cost of capital precisely when they need it most.
For the AI sector, that dynamic is especially dangerous because the largest spenders, hyperscalers, chipmakers, and infrastructure startups, are raising debt against future revenue streams that remain speculative. Goldman's framing of a "tsunami" implies that the problem is not any single borrower but the aggregate volume hitting the market simultaneously.
Echoes of the Subprime Crisis
Analysts have begun drawing explicit parallels to the 2008 financial crisis. Moomoo, examining whether the AI rally carries echoes of the subprime mortgage meltdown, pointed to roughly $1.8 trillion in off-balance-sheet exposure tied to AI and data-center commitments, obligations that may not be fully visible on corporate balance sheets.
The comparison is imperfect. In 2008, the systemic risk came from derivatives built on deteriorating housing assets; in the AI cycle, the exposure is capital expenditure on physical infrastructure that retains long-term value. But the structural similarity, heavy reliance on debt to finance assets whose near-term cash flows are unproven, is what makes the parallel resonant, and what gives the Goldman warning its weight.
Why Bond Investors Are Pushing Back
Forbes reports that fixed-income investors are demanding better terms and clearer paths to repayment before committing fresh capital to AI-linked issuance. The pushback reflects a simple calculation: AI infrastructure projects typically require multi-year buildouts, while bonds come with periodic coupon obligations. If monetization, whether through enterprise AI subscriptions, cloud revenue, or model licensing, lags the projections used to justify the borrowing, the gap must be bridged with even more debt or equity dilution.
This is the core tension the banks are flagging. Morgan Stanley's $570 billion estimate describes the supply of AI debt coming to market. Goldman's warning describes the demand side buckling under that supply. Should both prove accurate, the second half of 2026 could force a reckoning in which only the best-capitalized AI spenders can keep borrowing at sustainable rates.
What It Means for the AI Industry
The immediate practical effect is a widening cost-of-capital gap between AI giants and smaller players. Companies with strong balance sheets, established cloud revenue, and investment-grade ratings can still issue debt at reasonable yields. Startups and speculative infrastructure builders face steeper premiums or outright market access constraints, accelerating consolidation around a handful of dominant firms.
It also reframes the competitive landscape between the United States and China. Chinese AI firms, operating with different financing structures and state-backed capital, are less exposed to Western bond-market discipline. As American labs contend with investor resistance to fresh debt, their Chinese rivals, several of whom are pursuing public listings of their own, continue building on capital terms that Western markets cannot easily match.
The coming quarters will turn on a single question: whether the revenue from AI products arrives fast enough to validate the $570 billion in borrowing, or whether the bond market's patience runs out first. Either outcome will reshape the industry, but Goldman Sachs and Morgan Stanley have made clear which one they fear.
---
Stay Ahead of AIGet the latest AI news, analysis, and breakthroughs — all in one place.
Read more AI news →
