Update: AI’s Share of USD IG Supply Jumps 18% → 23% as Private DC Deals Hit $200B

What changed since the August mid-year vintage

In August we expanded the theme perimeter beyond five names: AI-related debt near $489B mid-year, with hyperscalers only ~40% of the stack, AI’s share of US IG supply at ~18%, and non-USD hyperscaler mix at ~33%. That update answered the scope question — how large is the theme once you leave Amazon–Alphabet–Meta–Microsoft–Oracle? This Q3 vintage answers the absorption question credit desks trade next: what moved in the newest official / desk refresh, and is theme weight inside USD credit supply still climbing even while the dollar perimeter holds?

Three layers force the rewrite. First, market synthesis of the Goldman credit frame now puts AI-related debt at roughly 23% of USD investment-grade issuance and about 20% of US high-yield supply — up from the August mid-year ~18% / ~18% pair. Second, nearly $200 billion of private data-centre transactions since early 2025 turns the overflow story from dry-powder abstract into a measured channel. Third, long-dated Meta and Amazon paper are pricing below official AA (BBB-equivalent and single-A-equivalent respectively) — supply technicals, not a sudden default narrative.

The dashboard above is built as a vintage delta: dual IG/HY supply-share path, Aug→Q3 diverging bars, new Q3 disclosures, rating-vs-market scatter, funding-channel bars, and capacity×horizon overflow.

The headline table: Aug mid-year vs Q3 2026 update

| Metric | Aug mid-year vintage | Q3 2026 update | Δ | |--------|---------------------:|---------------:|--:| | AI-related debt YTD | ~$489B | ~$489B | Perimeter held | | Hyperscaler share of AI debt | ~40% | ~40% | Carried | | Hyperscaler IG YTD / FY path | ~$194B / ~$250B | ~$194B / ~$250B | Spine held | | AI share of USD IG supply | ~18% | ~23% | +5 pp | | AI share of US HY supply | ~18% | ~20% | +2 pp | | Private DC transactions (since early ’25) | Not scoped | ~$200B | New channel | | AI-tagged share of US IG stock | Flow lens only | ~15% | Stock lens added | | Big-5 path in major IG index | Not highlighted | >5% by YE2026 | Concentration path | | Meta / Amazon long paper | Spreads grinding | AA → BBB / A equiv. | Rating–market gap |

Two readings matter. First, the August dollar perimeter did not break — ~$489B YTD and the hyperscaler spine are carried. Second, the theme’s weight inside credit supply is the real vintage delta: five more percentage points of USD IG, two more of HY, plus a quantified private DC sleeve.

Theme weight: 18% → 23% of USD IG is the live number

Toggle the dashboard’s Supply shares lens. The dual-line path is the point of the update. Desk estimates moved from roughly 1% of US IG in 2024 to ~7% in 2025 to ~18% at the August mid-year print. The Q3 refresh prints closer to ~23% of USD IG and ~20% of US HY. That is not a restatement of the $489B stack; it is a statement that AI-tagged borrowers are claiming a larger slice of whatever the primary calendar is clearing.

High yield’s ~20% print matters for portfolios that thought they had escaped Big Tech by sitting lower in the capital structure. Theme risk is still thematic. Utilities, data-centre JVs, equipment names, and leveraged wrappers sit beside — and increasingly inside — the same absorption math as the hyperscaler IG calendar.

Goldman’s August desk preference still colours the trade: heavy supply produces modest spread widening, with a relative preference for BBB over AA where mega-cap tech supply is denser relative to market size. Q3 does not reverse that — it hardens it. When AA paper is the overweight supply bucket, the market charges AA borrowers as if they were lower-rated names.

Rating vs market: AA labels, BBB/A prices

Switch to Rating vs market. Meta’s 2036 bonds still carry an official double-A rating; they price like triple-B. Amazon’s long-dated paper, also officially AA, prices closer to single-A. The gap is the bond market’s way of saying volume — not imminent default — is the governing variable.

That matters for anyone holding “high-grade” passively. AI-tagged bonds already represent about 15% of the US investment-grade market on a stock basis in the August–Q3 synthesis, and the Big Five alone are on a path that could exceed 5% of major IG indices by end-2026. You do not need to buy a mega tech deal to own the theme; index weights and AI-tagged peers deliver it.

Oracle’s July move to BBB-minus remains the live fundamental case study — S&P citing mid-fifties-billion fiscal capex, negative free cash flow, and leverage heading toward the mid-4x range — but the Meta/Amazon pricing gap shows technicals can reprice AA names without a downgrade. Pair that with the capex intensity update: thinner free-cash-flow cushions make multi-year releveraging the base case, not a temporary funding bridge.

Private data centres: $200B turns overflow into a flow

When the dashboard switches to Funding overflow, the August story gains a concrete leg. Private markets dry powder near $4.5T and project-finance sketches near ~$300B in 2027 are still the capacity map. What is new is the realized private data-centre channel: roughly $200B of transactions since early 2025. That is not committed dry powder; it is deals that cleared outside the public IG calendar.

J.P. Morgan’s absorption framing — about $300B of high-grade capacity for AI and data-centre bonds over the next year against roughly $1.5T of funding needs over five years — sits beside Goldman’s earlier ~$510B bank-comparable IG room case study. Access to capital is still not the binding constraint. Where, at what tenor, and at what concession are. Private DC deals are how the stack clears when US dollar books thicken and concessions refuse to shrink.

Non-dollar markets remain open overflow valves. August’s ~33% non-USD share of hyperscaler supply is carried — CAD, sterling, Swiss franc, euro, Aussie, and yen packages are still the waterfall Amanda Lynam described, not a closed channel. EUR credit remains relatively under-contributing versus US IG in that reading, which means European books are still an option when New York saturates.

Equity and ETFs still price the narrative, not the PP&E

The equity sleeve from the July research and August update still matters as sentiment capacity. FactSet’s 2025 US thematic ETF inflows near $43.5B, with QQQ alone around $21.7B, remain the public-markets willingness meter. Alphabet’s mid-2026 equity raise sits beside the bond calendar as proof issuers will reopen equity when credit books thicken. ETF flows do not pour concrete; primary debt and private DC capital do. The dashboard’s equity panel is carried deliberately — so readers do not confuse secondary ownership with issuer proceeds.

Who is exposed under the Q3 vintage

Exposed: real-money and insurance books that assumed “underweight Big Tech bonds” meant underweight the AI financing theme; passive IG holders who inherit ~15% AI-tagged stock and a Big-5 path toward >5% index weight; AA-rated hyperscaler longs priced as if they were A or BBB; HY and JV buyers who thought leaving the five-name IG calendar exited the theme.

Relative winners under current rules: desks that can rotate into BBB where AA supply is densest; private credit and infra vehicles clearing the ~$200B DC channel; issuers with multi-currency programs that can open EUR/GBP/CHF/JPY when USD cover ratios compress; investors who mark theme share of supply rather than name count.

What would change the story: a Q4 print that pulls AI’s USD IG share back toward the mid-teens while concessions compress; a restated AI-debt perimeter well above or below $489B that forces a full-year rethink; or a broad rating migration that turns today’s technical AA–BBB gap into fundamental downgrades across the cohort. None of those reverse signals are in the Q3 desk refresh.

Caveats and methodology

1. Perimeter vs share. The ~$489B AI-related debt YTD figure is carried from Goldman’s August mid-year estimate (hyperscalers ~40%). The Q3 23% / 20% prints are theme shares of USD IG / US HY gross supply from market synthesis of that framenot a new dollar tally. Do not annualize $489B into a full-year run-rate.
2. Desk estimates, not a single SEC series. Supply-share percentages and private DC ~$200B are research / market tallies with theme-tagging judgment. Labels differ across houses.
3. Rating-vs-market scores on the dashboard are ordinal viz aids (AA≈5, A≈4, BBB≈3). They illustrate the Meta/Amazon pricing gap; they are not agency ratings or option-adjusted spread z-scores.
4. Stock vs flow. ~15% AI-tagged share of the US IG market is an outstanding / stock lens. ~23% is a flow share of issuance. Do not add them.
5. Overflow capacities (JPM ~$300B / ~$1.5T; GS ~$510B bank-comparable; ~$300B project finance 2027; $4.5T private powder) are case studies and aggregatesdirectional maps, not committed pipelines.
6. Hyperscaler FY path (~$250B; ~33% of capex) remains the July–August research ratio for the five-name IG spine, not the share of the $489B stack.
7. ETF flows remain secondary-market ownership metrics.

Primary sources: Goldman Sachs Exchanges — How AI Debt Is Reshaping Credit Markets (Aug 2026); Dark Side of the Boom synthesis of the IG/HY share and private DC channel (Jul 22, 2026 frame, Q3 desk read); ACF/StockWireX rating-gap and IG stock-share context (Aug 11, 2026); prior theme posts AI financing mid-year update and AI financing research. Capex intensity context from the August intensity update.

What to watch into year-end 2026

Three coincident signals will tell you whether the Q3 share print stabilizes or climbs further: (1) AI’s share of USD IG / HY supply — if the theme stays in the low-to-mid twenties into Q4 while concessions refuse to shrink, saturation is binding even with private DC overflow; (2) whether Meta/Amazon long paper keeps pricing below AA — a closing gap means technical indigestion easing; a widening gap means supply still dominates ratings; (3) private DC and project-finance secondary performance — if the ~$200B private channel and JV prints keep trading wide to issue, indigestion has already spilled past the IG calendar. Until those three clear, the live number is not August’s $489B perimeter alone. It is ~23% of USD IG supply — with hyperscalers still the visible 40% of the dollar stack, and theme weight inside credit benchmarks still rising.