Goldman Sachs’s credit strategy team tallied AI hyperscaler debt supply at $296 billion so far this year, well above the bank’s $250 billion forecast for 2026, MarketWatch reported on 6 November, citing a Goldman report from Thursday, 5 November.
By our arithmetic, that is about 18% over the full-year forecast (296 / 250 = 1.184), with roughly eight weeks of the year left at the time of the report. MarketWatch says Meta Platforms, Microsoft, Oracle and others borrowed another $67 billion since late October.
Business Pill · Credit Spread
An evergreen Business Engineer explainer on credit spreads. It teaches the general idea only and does not discuss this story.
The key insight: Goldman Sachs’s credit strategy team tallied $296 billion of hyperscaler debt supply this year against its own $250 billion forecast for 2026, MarketWatch reports, about 18% over by our arithmetic with roughly eight weeks left. BofA Global data put investment-grade AI bonds about 148 basis points over Treasuries, against 77 for the rest of the market, a 71 basis point gap by our arithmetic.
Goldman now expects another $440 billion in hyperscaler debt supply in 2027, according to MarketWatch. By our arithmetic, that is 1.49 times this year’s tally so far (440 / 296).
The Goldman team, led by Amanda Lynam, expects access to capital markets “to be somewhat uneven across issuers and highly dependent on the market tone”, MarketWatch quotes the report as saying.
MarketWatch says AI debt issuance this year “smashed Wall Street’s supply expectations” and “disappointed from a performance perspective”.
Hyperscaler debt supply as tallied by Goldman Sachs’s credit strategy team: $250 billion forecast for 2026, $296 billion so far this year (report of 5 November) and $440 billion expected in 2027. Our arithmetic: 296 / 250 = 1.184, about 18% over the forecast; 440 / 296 = 1.49. Data: Goldman Sachs, as reported by MarketWatch, 6 Nov 2026.
What Was Posted on X
MarketWatch: “With Meta, Microsoft, Oracle and others borrowing another $67 billion since late October, Goldman Sachs’s credit strategy team tallied AI “hyperscaler” debt supply at $296 billion already this year, in a Thursday report. That was well above the bank’s $250 billion forecast for 2026. Another $440 billion in hyperscaler debt supply is now expected in 2027, albeit with access to capital markets ‘to be somewhat uneven across issuers and highly dependent on the market tone.’ The supply deluge and volatility this year have pressured investment-grade AI corporate bonds, which have been trading at a large premium to the rest of the market. The difference was about 148 basis points above the risk-free Treasury rate on Thursday, versus 77 basis points for the rest of the market.”
As I warned my in my September report on “The AI Trade” (https://sageroadresearch.com/products/the-ai-trade): “The tech borrowing binge is far from over. There are no signs of US fiscal discipline to come. Inflation remains sticky and the chances of a Fed rate hike are rising. Market participants remain complacent about tech giants continuing to borrow at the rate they’re borrowing if borrowing costs keep rising. And if tech giants begin to pull back from the debt binge, it’ll ripple far beyond the bond market, through to everything from equities to consumer spending.”
A post by @trevornoren at 16:07 UTC on 7 November quoting MarketWatch’s passage on Goldman Sachs’s $296 billion hyperscaler debt tally, the $250 billion forecast, the $440 billion expected in 2027 and the 148 versus 77 basis point spreads. We read the MarketWatch article in full; the spread figures in it are attributed to BofA Global.
The Price Investors Charge
Investment-grade AI corporate bonds traded about 148 basis points above Treasuries on Thursday, against 77 basis points for the rest of the market, according to BofA Global data cited by MarketWatch. By our arithmetic, that is a gap of 71 basis points, and the AI spread is about 1.9 times the rest of the market’s (148 / 77 = 1.92).
AI exposure rose to 5.14% of the investment-grade corporate index as of late October, from 2.73% at the start of 2025, the same BofA data show. By our arithmetic, the AI share of the index has nearly doubled (5.14 / 2.73 = 1.88).
BofA Global expects AI-related issuance funded in credit markets to peak at about $685 billion next year, up about 31% from this year, MarketWatch reports. It says the yearly increase “would come almost entirely from private credit, which tends to be unrated”.
What the Issuers Filed
The issuers’ own prospectus supplements on SEC EDGAR show the size and price of their early-November deals. Microsoft’s supplement, filed on 2 November, covers $25 billion of notes in ten tranches due from 2028 to 2066, including a 6.150% note due 2046.
Oracle’s supplement filed on 4 November covers $10 billion of notes due 2029 to 2046, with coupons from 7.000% to 8.650%. By our arithmetic, the three dollar deals add up to $50 billion, plus Meta’s €9 billion; MarketWatch does not list the deals behind its $67 billion figure.
On the same 2046 maturity, Oracle’s 8.650% coupon is 2.50 percentage points above Microsoft’s 6.150% and 1.30 points above Meta’s 7.350%, by our arithmetic from the three filings.
Coupon on the note due 2046 in each issuer’s early-November deal, from the cover pages of Microsoft’s 424B2 (filed 2 November), Meta’s 424B5 (dated 3 November) and Oracle’s 424B2 (filed 4 November). The 2.50 and 1.30 point gaps are our arithmetic.
The Business Lever: Capex Moves to the Bond Market
MarketWatch describes debt financing as “now the lifeblood of the AI buildout”. Oliver Chambers, a senior portfolio manager at Clark Capital, told MarketWatch: “It won’t change the overall endgame — just the pace will likely slow a bit”.
Chambers said debt financing could start at a more moderate pace in the first quarter of 2027, as Wall Street tries to figure out “the rules of the road” after the midterm elections, MarketWatch reports. Denny Fish of Janus Henderson said in emailed comments that the locations and pace of the rollout “are now in question”.
The Structural Read
As we read the figures, the AI buildout is now funded in the bond market at a scale the bank that tracks it did not forecast: the year’s supply passed Goldman’s full-year number before November was out.
As we read the spreads, the 71 basis point gap is the price of that supply: investors are asking AI issuers for extra yield over the rest of the investment-grade market, which MarketWatch says reflects perceived AI risks.
As we read the three filings, the market is pricing issuers apart: on the same 2046 maturity, Oracle’s 8.650% coupon is 2.50 percentage points above Microsoft’s 6.150%, by our arithmetic.
Goldman Sachs credit strategy team, as quoted by MarketWatch
“to be somewhat uneven across issuers and highly dependent on the market tone”
Three Implications
FOR ISSUERS MarketWatch says borrowers have needed to pay higher interest rates to find takers for all the supply, and Goldman expects access to capital markets in 2027 to be uneven across issuers.
FOR BOND INVESTORS BofA data show AI exposure at 5.14% of the investment-grade corporate index as of late October, up from 2.73% at the start of 2025, which MarketWatch says makes it hard for any bond fund to avoid.
FOR THE NUMBERS TO WATCH Goldman expects $440 billion of hyperscaler debt supply in 2027, and BofA expects AI-related issuance funded in credit markets to peak at about $685 billion next year, with the increase coming almost entirely from private credit, MarketWatch reports.
The essay argues that “the capex race is no longer a corporate spending cycle” and that “It is a sovereign-scale financing problem dressed in tech-company clothing”.
As we read it, Goldman’s $296 billion tally puts a number on the financing side of that thesis for 2026, and the $440 billion it expects for 2027 is 1.49 times this year’s total so far (our arithmetic). The essay does not discuss Goldman’s figures or bond spreads.
What Is Not Established
We read the MarketWatch article, not Goldman’s report itself. The article does not name the deals behind the $67 billion or give Goldman’s split by issuer. The 148 and 77 basis point spreads are BofA Global’s figures, not Goldman’s.
BofA’s $685 billion covers AI-related issuance funded in credit markets, including private credit, a wider category than Goldman’s hyperscaler debt supply, so the two figures are not comparable. The coupons we compare are from the filings’ covers and do not show the spread each issuer paid over Treasuries.
This Week In The AI Supercycle
The Great Convergence of Capital, Power, and Geopolitics
This week’s issue: for two years the AI build-out was priced as a supply story, and last week as a credit story. This week it was priced as a permission story, and the answer arrived as a cost, not a ban.
Goldman’s credit strategists tallied $296 billion of hyperscaler debt supply this year, about 18% above their $250 billion forecast by our arithmetic, and expect $440 billion in 2027, MarketWatch reported. BofA data put investment-grade AI bonds at about 148 basis points over Treasuries, against 77 for the rest of the market.
A note on sourcing. The Goldman Sachs debt figures, the BofA Global spread and index figures and the quotations are from MarketWatch’s 6 November 2026 article by Joy Wiltermuth, read in full on 8 November 2026; we did not read the Goldman or BofA reports themselves. Deal sizes and coupons are from the cover pages of Microsoft’s, Meta’s and Oracle’s prospectus supplements on SEC EDGAR. Ratios, sums and coupon gaps are our arithmetic.
Gennaro Cuofano is a CRO and tech executive who has worked in AI since late 2015, starting with bringing NLP, natural-language generation, voice and chatbot products to the marketing industry. His background is in law and finance: he holds a Master's degree in Law and an International MBA with an emphasis on corporate finance (LUISS Business School and the University of San Diego, 2012), and worked as a financial analyst at a real-estate investment firm in San Diego and as an assistant controller. His work focuses on business model strategy, business engineering and, more broadly, structural analysis: how companies make money, read from their own filings. He created FourWeekMBA and leads research there and at The Business Engineer, his newsletter on AI and business strategy, with over 95,000 subscribers and more than 1,000 published analyses. His writing has also appeared on Entrepreneur (for example, How to Digitize a Family Business in Europe) and HackerNoon. Find him on LinkedIn and Substack. See how we source.
