We read all 275 completed Treasury auctions of 2026: demand thins as the debt lengthens — and the single weakest spot on the curve is the 5-year note, not the 30-year bond

Every time the Treasury borrows, it holds an auction and publishes how many bids came in against how much it sold. That ratio — total bids divided by the amount accepted — is the bid-to-cover, the standard gauge of demand relative to supply. We read the complete 2026 auction calendar from the Treasury's own auctions dataset: every sale of bills, notes and bonds, including reopenings and cash-management bills. The 2026 calendar holds 282 auctions; 275 had settled with results through Aug. 20, and the other 7 were scheduled for Aug. 24–27 and had not yet been held. Across the 275 completed auctions, demand relative to supply is highest at the shortest end of the debt and thins as maturity lengthens. Bills averaged a 2.94 bid-to-cover across 207 sales; notes 2.60 across 50; bonds 2.56 across 18. Those 275 auctions accepted $20.76 trillion in total — gross issuance, not net new debt: bills are rolled over every few weeks, so the same borrowing capacity is auctioned again and again, and the figure dwarfs the year's net borrowing. The type-level slide from bills to bonds is clean, but the coupon curve underneath it is not. The soft spot is not the 30-year long bond that markets watch for demand scares — it is the belly. Comparing coupons like with like — each nominal fixed-rate tenor at its original term, with reopenings folded back in and TIPS and floating-rate notes set aside — the 5-year note averaged a 2.38 bid-to-cover across eight sales, the weakest of the seven coupon tenors: below the 10-year (2.49), below the 30-year bond (2.42), and below the 2-year (2.61). The 5-year also posted the single lowest cover of the entire year — 2.28 on July 27. At the other end, the strongest single result was a one-off 27-day cash-management bill at 4.60 on May 21; among regularly scheduled issues, the 52-week bill led at a 3.32 average.

The following is analysis, not fact. Read the type-level ranking as a verdict on trust — investors line up harder for 4-week paper than for the 30-year, so they must doubt the long-term credit — and you would be reading it wrong. Bills are a cash-management market: money-market funds and corporate treasurers park short-term cash there and roll it constantly, which structurally lifts a bill's cover regardless of any view on U.S. solvency. Coupons compete against duration risk instead. A 4-week bill and a 30-year bond are two different markets, and the gap between their cover ratios is mostly that, not a confidence gradient ^sspeculative. The read only the full-year pull surfaces is the within-coupon one: the weakest link in 2026 sits in the belly, at the 5-year, not at the long end. That is the comparison of like with like — coupon against coupon — and it inverts the usual long-bond-scare framing. External auction-watchers flagged the same belly softness in the spring: a late-March run of 2-, 5- and 7-year sales drew below-average covers, with primary dealers taking an above-normal share. Why the belly is soft — rate-path uncertainty, or thinner natural sponsorship than the 2-year (money funds) and the 10- and 30-year (liability-matching buyers) command — is an explanation the pattern is consistent with, not one these data isolate ^sspeculative. We report the ranking; the cause rides on top, labeled.

Bid-to-cover is one blunt gauge, and it is the whole of what we measured. It does not distinguish genuine end-investor demand from primary dealers backstopping a soft auction — dealers can absorb an outsized share and still leave the cover looking respectable, and the auction tail (how far the clearing yield runs past pre-auction levels) can flag weakness a cover ratio misses. The Committee for a Responsible Federal Budget's read of the March auctions makes exactly this point: the 5-year that month cleared a 2.29 cover with primary dealers taking 16% versus their 11% norm. We pulled the published cover ratio only, not bidder composition or tails. Two more cautions. The bills-versus-coupons comparison is not apples-to-apples for the structural reason above; the clean comparison is the within-coupon one (5-year versus 30-year). And the per-tenor counts are small — 8 five-year sales, 8 ten-year, 8 thirty-year on the original-term nominal basis — so an average built on single-digit auctions can move on one weak result, and this is a partial year (Jan. 5 to Aug. 20). Every figure here is scoped to that window.