Workspaces
Log in Sign up

5.29%: the 10-Year Ends September Above the 2007 Peak It Matched

Updated

5.29%: the 10-Year Ends September Above the 2007 Peak It Matched

The prior revision flagged the 10-year at 5.26% as matching its 2007 peak. September ended higher still: the 10-year Treasury (DGS10) closed 2026-09-30 at 5.29% (from 5.26%), with the fed funds rate at 3.75% after the September hike and unemployment at 4.1% (/markets/macro/2026/10/02).

This note's thesis — that AI capex now moves the rate that discounts it ("deficits and AI spending compete for the same pool of savings") — hardened another notch. The buildout's marginal buyer is now borrowing into it: Goldman notes hyperscaler capex is outpacing operating cash flow, pushing the complex toward debt and equity funding (The Macro Ledger Hardens: $9T Through 2032, a $4.2T Revenue Gap, and the 3–5% Productivity Bar Nvidia's Valuation Now Implies), while neoclouds and SPVs issue against GPU collateral (Private Credit Becomes the Buildout's Central Backstop: the WSJ Debt Trio, Anthropic's Four-Role Prospectus, and the AVGO Discount). All of it prices off the long end, and the long end keeps answering.

Against NVDA at $230.86 — a $5.51T market cap at 29.06x trailing earnings, 1.8% below its 52-week high (/markets/NVDA/2026/10/02) — a 5.29% 10-year is the highest steady-state discount rate the AI trade has faced this cycle1. The market is still paying 29x for it, but Goldman's count shows the multiple compressing anyway (AI infrastructure at 22x forward vs 32x in April).

The next FOMC decision (late October or mid-December; the Fed signaled one more 2026 hike, median projection 4.1%) is the second-order test — an existing watch on this topic. A second hike would push the discount rate further up and raise the cost of the debt-funded portion of the buildout; a hold would relieve the overhang.

  1. An instance of The cost of money, not customer demand, now gates the AI capex boom. — Zero cancelled orders alongside a compressing multiple (22x forward versus 32x in April) shows the gate on the boom has migrated fully to the price of money. ↩︎

Part of

This finding is an example of a pattern recurring across your work:

Revision history

  • 10-year closed September at 5.29%, extending above the 2007-peak level (5.26%) flagged in the prior revision.
    · by the agent
  • 10-year closed September at 5.29%, extending above the 2007-peak level (5.26%) flagged in the prior revision.
    · by the agent
  • 10-year closed September at 5.29%, extending above the 2007-peak level (5.26%) flagged in the prior revision.
    · by the agent
  • Update: 10-yr Treasury reached 5.26% on Sept 29 — matching the 2007-peak level the prior revision said was two basis points away. Cited from the Oct 1 macro market view.
    · by the agent
  • Rates update: 10-yr 5.24% (2bp from 2007 peak), 30-yr 5.56% (2002 territory), worst September since 2023, AI build-out named as a driver, S&P forward P/E 22→19.
    · by the agent
  • Discount-rate loop hardens: 30-yr closes at 5.489% (highest since 2004) with AI debt named a cause; hyperscaler spreads widen and 80% of issued bonds trade wider than issue price.
    · by the agent
  • Discount-rate loop hardens: 30-yr closes at 5.489% (highest since 2004) with AI debt named a cause; hyperscaler spreads widen and 80% of issued bonds trade wider than issue price.
    · by the agent
  • Discount-rate loop hardens: 30-yr closes at 5.489% (highest since 2004) with AI debt named a cause; hyperscaler spreads widen and 80% of issued bonds trade wider than issue price.
    · by the agent
  • Discount-rate loop hardens: 30-yr closes at 5.489% (highest since 2004) with AI debt named a cause; hyperscaler spreads widen and 80% of issued bonds trade wider than issue price.
    · by the agent
  • Revision: 10-yr printed 5.18% (Sept 24, highest since 2007) with a $570B AI bond wave cited as driver — direct confirmation of the note's self-referential-loop thesis
    · by the agent
  • Sept 18–19 update: Warsh names hyperscaler bond issuance as a partial driver of 5% yields; $121B 2025 issuance vs $28B avg; Fed AI task force due year-end; yields eased to 4.94%.
    · by the agent
  • Sept 18–19 update: Warsh names hyperscaler bond issuance as a partial driver of 5% yields; $121B 2025 issuance vs $28B avg; Fed AI task force due year-end; yields eased to 4.94%.
    · by the agent
  • Sept 18–19 update: Warsh names hyperscaler bond issuance as a partial driver of 5% yields; $121B 2025 issuance vs $28B avg; Fed AI task force due year-end; yields eased to 4.94%.
    · by the agent
  • Sept 18–19 update: Warsh names hyperscaler bond issuance as a partial driver of 5% yields; $121B 2025 issuance vs $28B avg; Fed AI task force due year-end; yields eased to 4.94%.
    · by the agent
  • Sept 18–19 update: Warsh names hyperscaler bond issuance as a partial driver of 5% yields; $121B 2025 issuance vs $28B avg; Fed AI task force due year-end; yields eased to 4.94%.
    · by the agent
  • Sept 18–19 update: Warsh names hyperscaler bond issuance as a partial driver of 5% yields; $121B 2025 issuance vs $28B avg; Fed AI task force due year-end; yields eased to 4.94%.
    · by the agent
  • New finding: the September 2026 FOMC delivered the first hike in three years with one more signaled — the macro wall now sits on top of the AI capex story.
    · by the agent