Friday pre-open, before the 8:30 a.m. CPI print. Thursday's 10–20y buyback took $5.187B vs a $6B max (above the $4B falsifier). August PPI +0.4% m/m / +5.4% y/y; core +0.2%. FERC, 18A, WuXi still silent. Thin-vertical buckets #10 consumer and #11 transport: no net-new chain.

Event-day gates: Treasury took $5.187B, PPI printed in-line, CPI not yet, FERC/18A/CDMO still silent

Generated 2026-09-11 ~05:00 ET, before US cash open and before the 8:30 a.m. CPI print. Treat as raw material.

Summary

Treasury's first doubled long-end buyback took $5.187 billion of 10- to 20-year bonds on Thursday afternoon against a $6 billion maximum and $10.489 billion offered (bid-to-cover 2.02). That is above the machine-checkable falsifier (buyback_take_usd < 4e9) and below the max. Yields rose anyway — 30-year to about 5.36%, highest since June 2007. Step 1 of the twist chain is still confirmed. Do not emit gold. The next sized op is September 24 (20–30y).

August PPI printed Thursday: final demand +0.4% m/m / +5.4% y/y, matching the monthly consensus; core ex-food-and-energy +0.2% m/m (a tick under a 0.3% consensus in some prints). Energy +4.2% m/m on renewed US–Iran hostilities. CME FedWatch before that print was about 62% for a 25 bp hike at September 15–16. Futures odds are not a reaction function. CPI for August prints today, 8:30 a.m. ET — it has not printed as of this pass. Schwab's last disclosed scenario still assumed a December hike. Do not graduate SCHW.

FERC has not ruled on ER26-3380. Intel still has no named outside 18A high-volume order. WuXi's answer is still September 18. Twenty-first weekday on the Western-CDMO −15% contradiction.

Thin-vertical buckets today: #10 consumer shift and #11 transport / industrials. No net-new tradeable chain in either. The AI-infrastructure cluster is already overweight, so today's new-chain budget did not add another name there — the twenty-third weekday in a row.

Findings

Treasury took $5.187B. The $4B falsifier did not fire. Do not emit gold.

Gate / Treasury recap, September 10: accepted $5.187 billion; offered $10.489 billion; max $6 billion; bid-to-cover 2.02. Largest takes: November 2040 ($1.703 billion) and May 2042 ($1.294 billion). Yonhap Infomax and Rallies / Treasury match the take. Bloomberg: 10-year yield extended to 4.95%, highest since 2023. Next long-end op: September 24, 20–30 year sector.

treasury-buyback-twist-to-hard-asset-debasement Step 1 was already confirmed on the August 19 size announcement (sb0607). The dated test on this page was the take: buyback_take_usd < 4e9. $5.187B > $4B — falsifier did not fire. A take below the $6B maximum is not a miss of the doubled program. Yields going up on the take is the same "market went higher in yield anyway" pattern already on the page from the August 18 $2B op. Do not emit GLD/IBIT/XLE. Do not re-date a gold row that is not in the live feed. Next machine check: September 24.

Official results PDFs on TreasuryDirect's 2026 announcement index were not fetched this pass (JS table / prior-dated PDFs only in search). Secondary recaps of the Treasury statement are the citations. Flag: prefer the official BBR PDF on a later credentialed fetch.

PPI printed in-line. Core a tick soft. CPI has not printed.

Haver Analytics, September 10: PPI final demand +0.4% m/m (5.4% y/y) after an upwardly revised +0.1% in July; energy +4.2% m/m; goods +1.1%; services +0.1%. Core ex food and energy +0.2% m/m (4.6% y/y) vs a 0.3% consensus in that write-up. Core ex food, energy and trade services +0.3% m/m (4.7% y/y). Reuters via Journal Record: same headline; FedWatch ~62% hike ahead of the report; funds rate 3.50–3.75%. Energy rebound tied to renewed US–Iran hostilities.

BLS CPI for August is today, 8:30 a.m. ET (BLS CPI schedule; BLS CPI home). This pass is before that print. July CPI-U was +3.4% y/y (BLS archive, August 12). Wall Street Times still had 60–65% hike odds into Friday. Conversion for warsh-higher-for-longer-to-brokerage-nii-rerate remains the FOMC statement, September 15–16, not a PPI print that matched estimates. Do not graduate SCHW.

Intel 18A: still no named high-volume outside order

Socionext 18A-P remains an ASIC-house process add (TrendForce, August 20). InvisibleHill: AWS framework and a Microsoft unnamed 18A design are the public names; volumes unconfirmed. SK Hynix "weighing" HBM4E base-die work is unconfirmed. October PDK 0.9 remains the dated test. Question stays open.

FERC has not ruled on ER26-3380

No Commission order in overnight search. PJM still wants action by September 29 so the Reliability Backstop Procurement can open September 30–October 21, results December 2 (PJM Inside Lines; July 31 filing). Silence is not a Constellation re-rate. Do not re-rate CEG. Do not re-date the CEG signal.

Western CDMO: still no 2026 realized price series — twenty-first weekday

Last primary is still the August 7 injunction / National Law Review restatement. Answer September 18. OMB BCOC list December 18. A law-firm restatement is not a 2026 realized mAb fill-finish series. Human call, not a silent downgrade. Twenty-first weekday.

Consumer shift (bucket #10, thin vertical): no new 72-hour forcing function

NielsenIQ / World Data Lab, September 8: US FMCG as a barbell — premium and value grow, the middle shrinks; shoppers trade up and down in the same basket. That is the existing iran-fuel-shock-consumer-bifurcation / consumer-trade-down-to-private-label-manufacturer-treehouse spine, not a new chain. Walmart Q2 FY27 (August 20) and TJX Marmaxx 1% comps are already on those pages. No new named manufacturer. Do not graduate WMT. Do not rank TJX (low, live −21.2%). Do not draft.

Transport / industrials (bucket #11, thin vertical): rates still tight after Labor Day — existing spine

DAT dry-van, September 8: spot linehaul $2.21/mile, +33.6% y/y, load-to-truck 11.47 vs 6.68 a year ago. FreightWaves / Yahoo, September 9: volumes −15% post–Labor Day; National Truckload Index $3.44/mile, +47% y/y. Capacity trickling back is not a contract-rate collapse. Corroborates driver-supply-removal-to-truckload-contract-rate-inflection; Q3 prints still late October. Do not re-rate WERN. Howmet: no new aerospace primary overnight; Tuesday's dump is still tape. Do not chase HWM.

SemiAnalysis BTM (filed separately) — attach, do not mint

What is So Hard About Behind-The-Meter Power For Datacenters? Part 1 (September 10): 75 GW firm BTM OEM orders; ~20 GW in Q2 2026; ~3 GW US IT BTM by year-end; recip count 12 → 22 manufacturers. Microsoft / Google / Anthropic / OpenAI named. That is hardware-demand color for ai-power-gap-to-genset-bridge-power and the phantom-load / energized-vs-announced pages — not a tenth AI-infrastructure chain. Partial (OEM/BoP winners behind the paywall). Do not auto-draft. Do not re-date CAT.

Contradictions and open questions

CUDA Step 2 is still an unmerged pair (AgentX vs Jalapeño). The BTM post does not merge them.
Energy-oil cluster label still measures CCJ/CF/NTR, not the refiners sitting near the highs. Twenty-second consecutive week flagged, unfixed.
Western-CDMO −15 percent series vs. thesis: twenty-first weekday, still a human call.
Does this morning's CPI, then the September 15–16 statement, move Schwab — or is the conversion still Warsh's words? Do not graduate off the calendar.

Provenance

Method: in-vault WebSearch + WebFetch (DAILY.md Skill("autoresearch") fallback; compact dated-gate pass + thin-vertical buckets #10 and #11). Priors skipped (headless). Generated: 2026-09-11

Web sources: