‘Uncharted territory’: The $40 trillion U.S. national debt just got uglier as interest payments rise to $1.25 trillion a year

U.S. net interest costs reached a record 18.5% of federal revenue in 2025, or roughly $1.25 trillion on a $40 trillion national debt, exceeding the prior 1991 peak of 18.4%. Debt held by the public has risen from about 44% of GDP in 1991 to more than 100% today, while the CBO projects interest costs could reach 25% of revenue by 2036. Analysts warn that $225 billion of AI-related corporate bond issuance in H1 2026 is competing with Treasury borrowing for long-term capital and pushing Treasury yields higher; the Treasury has doubled 10-to-30-year bond buybacks to at least $4 billion per operation.

Analysis

The investable transmission is a higher term premium rather than a mechanical rise in policy rates. Persistent long-end supply pressure raises discount rates for duration-heavy equities and rate-sensitive credit even if the Fed eases at the front end; this favors financials with asset-sensitive balance sheets over regulated utilities, REITs and long-duration software. The most vulnerable equity cohorts are those funding multi-year capex with incremental debt—particularly AI infrastructure beneficiaries whose valuations already assume cheap, abundant capital.

The claim that AI bond issuance directly displaces Treasury demand is directionally plausible but overstated: global reserve allocation, bank balance-sheet capacity, foreign hedging costs and Treasury issuance composition matter more than gross corporate issuance alone. The more important second-order risk is spread widening: if hyperscaler issuance must clear at wider spreads, the cost of capital rises simultaneously for suppliers, data-center developers and lower-rated issuers, potentially slowing the AI capex cycle 6-18 months out. That would pressure names with high revenue sensitivity to hyperscaler spend, including VRT, ETN, CEG and data-center REITs, despite near-term backlog support.

Over the next 1-3 months, auctions, term-premium measures and Treasury refunding guidance are the catalysts; a weak auction tail or renewed long-bond yield breakout can reprice equity duration quickly. The bearish duration thesis is falsified by sustained disinflation, materially smaller net coupon issuance, stronger foreign Treasury demand, or a decline in the 10y term premium despite heavy supply. Treasury buybacks improve market functioning at the margin but do not meaningfully alter the fiscal trajectory; treating them as yield caps would be a positioning error.

Key Decisions for Investors

• Maintain a 3-6 month curve-steepener bias: long IEF versus short TLT, sized modestly. The thesis is that front-end easing can coexist with long-end term-premium expansion; exit if the 10s30s spread compresses materially after refunding or auction demand improves.
• Pair trade for the next 1-3 months: long XLF / short XLRE or XLU. Higher long yields improve reinvestment economics for banks but compress REIT and utility valuation multiples; key risk is a growth scare that drives a broad flight to duration.
• Reduce exposure to highly levered digital-infrastructure vehicles and use VRT/ETN/CEG strength to trim rather than chase until corporate-bond concession data confirm funding remains benign. Re-enter only if new-issue concessions normalize and hyperscaler capex guidance remains intact.
• Set an alert around Treasury quarterly refunding and 10- and 30-year auction tails. If long-end clearing yields rise while investment-grade spreads also widen, add a tactical TLT short or TBT long; without current implied-volatility and yield data, options are a watch item rather than a recommendation.