The artificial intelligence infrastructure boom is unleashing an unprecedented supply wave on credit markets, and the current calm is merely a fleeting quiet within the storm's eye. In a recent report, Goldman Sachs fixed income credit trader and investment-grade credit specialist Jeffrey Papai warned that after roughly $300 billion in AI-related bond issuance this year, fourth-quarter supply will slow notably, offering temporary relief for AI credit spreads. Yet this reprieve is extremely brief—he projects that bond issuance from hyperscale cloud providers and chipmakers will climb about 40% in 2027 versus 2026, reaching roughly $340 billion, when the market will face an even larger impact.

The warning comes against a backdrop where AI debt expansion has begun to exert competitive pressure on the U.S. Treasury market. According to JPMorgan's Michael Cembalest, debt issuance from the five largest hyperscale cloud providers plus Nvidia totals roughly $320 billion so far this year (including special purpose vehicles for data center leases), with the longer-duration portion equating to about $303 billion on a 10-year equivalent basis—approximately 68% of the new long-duration borrowing taken on by the U.S. Treasury over the same period.

Q4: A brief calm within the storm's eye

Papai noted that the roughly $300 billion in AI supply this year has made it the most significant theme in the investment-grade credit market. Heading into the fourth quarter, he expects just one remaining hyperscale cloud senior bond deal and a handful of data center transactions—the lowest issuance period since the AI buildout began, roughly 50% below any comparable prior stretch. Against this backdrop, he holds a short-term tactical bullish stance on the Goldman Sachs AI credit basket (GSUCIABK), seeing room for spread-tightening rebounds over the next one to two months—the basket currently sits less than 10 basis points from its widest level on record, and slower supply should provide support.

Papai stressed, however, that this call is purely tactical in nature. He advises investors to trim positions into strength during this rally rather than chase gains, because structurally, AI credit spreads remain in a long-term underweight state. Over the past year, AI spreads have widened by more than 50 basis points, and with a fresh supply wave approaching in 2027, renewed pressure is nearly a foregone conclusion.

2027: A larger supply wave is already on the way

Papai laid out a rough estimation framework for 2027 bond issuance from hyperscale cloud providers and chipmakers, explicitly noting this is not an official Goldman Sachs forecast but rather an extrapolation based on current trends. Key assumptions include: hyperscale cloud capital expenditures of roughly $930 billion in 2027 (a midpoint of Goldman Sachs Research and Bloomberg projections); the debt-financing share of capex rising from about 30% in 2026 to 37.5%, driven by an expected contraction in equity issuance from this group; the U.S. dollar funding share declining from roughly 80% to 70%; and senior chip debt increasing from about $35 billion in 2026 to between $50 billion and $75 billion.

Under these assumptions, average quarterly issuance in 2027 would match the pace of the busiest quarter to date. What is more striking: maturing debt for hyperscale cloud firms and chipmakers in 2027–2028 amounts to only about $45 billion, far below high-issuance sectors like banks (which have roughly $180 billion in maturities excluding 2029 TLAC callable bonds). This means refinancing demand will provide very limited backfill, and nearly all new supply will need to be absorbed by the market.

Papai further noted that including data centers and other AI-related issuance would expand the total even more. Structured chip financing, in particular, could be the single largest incremental area, with potential size exceeding $100 billion—but due to opaque structures and unpredictable timing, no specific estimate is possible at present.

Non-AI investment-grade bonds: Relative pressure in the near term

In contrast to the AI segment's brief respite, Papai sees non-AI investment-grade bonds as relatively fragile in the near term. September is expected to bring roughly $230 billion in predominantly non-AI investment-grade supply into the market, pressuring that segment. That said, he also pointed out that yield-driven demand for investment-grade bonds overall remains robust, and supply duration is expected to be short. He therefore recommends investors treat any supply-shock-driven dip in non-AI investment-grade bonds (GSIG30NH) over the coming weeks as a buying opportunity.

From a broader perspective, investment-grade credit spreads have widened through the summer but have yet to break out of their range—the investment-grade index spread sits around 80 basis points, above the 77 basis point year-to-date average, while the GS100 spread is near 99 basis points, above its 96 basis point average.

Structural shift in credit markets: Risk transfer volumes expanding sharply

Papai also highlighted in the report that the scale of risk transfer within credit markets is rising significantly across multiple product lines, a trend that will accelerate further in 2027. The credit default swap (CDS) market is experiencing a notable revival. Papai noted that several relationship managers who last traded CDS before the global financial crisis have recently been fielding fresh inquiries, signaling a clear rebound in demand for hedging tools. After nearly 15 years of declining volumes, CDS trading is projected to grow about 15% in both 2025 and 2026.

Meanwhile, basket total return swaps (TRS) continue to grow rapidly, and cash bond trading volumes have hit record highs, with both investment-grade and high-yield trading up 10% to 15% year over year. Papai believes that as structured chip and data center financing expands—featuring shorter duration and higher hedging needs—these trends will intensify further, and CDS traders could become one of the most sought-after roles on Wall Street in 2027.