〔Day Digest〕Term-Premium Reset and AI Bond Supply Push 30Y Past 5.3% — Bessent's 'Put' Fades Into Jackson Hole, Nvidia, and the Midterms

0. Weekly Arc

Soft data (weak payrolls, soft inflation prints) has been overwhelmed by supply pressure: the 30Y US Treasury cleared 5.3% (highest since 2007) and the 10Y pushed into the 4.6-4.8% band, even with September hike odds near 40% and October cumulative at 65% ^{[1][2][3][4][5]}. The mechanism has flipped from rate expectations to term premium, driven by AI hyperscaler credit supply competing for marginal buyers, fiscal-deficit worry, and central-bank QT reducing the natural bid ^[2][3][6][5]. The Treasury's per-coupon buyback doubling to $4B is sentiment, not structure ^[7][3][8]. Catalysts: Jackson Hole (Warsh expected to pare forward guidance), Nvidia's print, the August PCE, and the November midterms ^[2][3][9].

1. Long End: Term Premium, Not the Fed

• **[ESCALATED] Supply repricing, not expectations:** Huaxin Securities, Zhonghang Securities and CICC all attribute the 30Y's 5.3% push to term premium — AI bond supply, fiscal worry, and central-bank QT are the common drivers ^[2][3][6][5]. The 30Y auction yield touched a 2001 high in the same window ^[1].
• **[NEW] AI credit crowding out:** Nomura's strategists (Aug 24) say major tech issuer spreads at cycle highs are competing directly with Treasuries for marginal dollars, eroding Treasury Secretary Bessent's long-end "put" ^[2]. The 10Y real yield has returned to 2.40% ^[2].
• **[NEW] Warsh preview:** Barclays' rates team expects Fed Chair Warsh to argue that forward guidance was a source of past policy errors and to offer limited near-term guidance at Jackson Hole; balance-sheet and productivity commentary likely ^[9].
• **[NEW] Gold through $4,600/oz:** Founder Securities attributes the breakout to the Treasury buyback signaling weak-dollar intent, plus soft US data, with structural support from AI inflation and US fiscal dynamics ^[7].

2. Cross-Asset and Geopolitics

• **[ESCALATED] Iran-US 60-day window expired without deal:** US suspended talks, Iran held its Strait of Hormuz closure stance, UAE suspended trade with Iran — risk premium is rebuilding into crude and shipping ^[10].
• **[NEW] Goldman (Aug 24):** European gas must rise above €100/MWh to balance winter demand if Middle East energy exports stay disrupted ^[11].
• **[NEW] Dollar ~99, RMB past 6.73:** The Treasury buyback initially supported the soft-dollar trade, but the long end rebounded by week's end and pulled the dollar back up; Huatai reads this as "the buyback moderates, but does not solve, the curve" ^[8].
• **[NEW] Commodities tape (Aug 18-23):** Energy > precious metals > coal chemicals > oil chemicals > agriculture > steel > nonferrous; Iran and weak dollar drove the front of the curve ^[12].

3. Political Risk: The Midterm Hedge

• **[NEW] Michael Hartnett (Bank of America):** Markets are pricing a benign mid-term outcome. If Democrats take the Senate, the deregulation and AI capex narratives both break, and the S&P faces a >10% drawdown by year-end with a weaker dollar and lower yields ^[13]. The Texas governor's race is the AI policy "referendum" ^[13]. Trump approval: 39% overall, 36% on the economy, 30% on inflation — all below pre-Iran-conflict baselines ^[13].
• **[NEW] Andrew Sheets (Morgan Stanley Global Fixed Income Research Head):** $40T US federal debt is a brake on activity, not a collapse trigger; the offset is healthier household and corporate balance sheets ^[14]. The break point, if any, is at the asset-allocation level when bond yields genuinely compete with equities ^[14].
• **[NEW] Barclays sectoral map (Venu Krishna, Riddhiman Dass and team, Aug 24):** Across the last five hiking cycles, the S&P 500 median fell 3.9% in the first quarter after the first hike, small caps 7.2%, financials 8.4% — energy was the only sector to print a positive return ^[1].

4. China-Specific Calls

• **[NEW] Soochow Securities / Feilong Co. (Buy):** 2026H1 net profit attributable ¥76M (-64.02% YoY) on FX and raw-material one-offs; liquid cooling orders exceeded expectations, with batch delivery imminent. 2026 NP maintained at ¥420M; 2027/2028 raised to ¥619M/¥943M. Current P/E 64/44/29x ^[15].
• **[NEW] Soochow Securities / Longxin General (Buy):** Wujue brand reached 5.79% combined market share in Italy, Spain, France, Germany, UK — the first Chinese brand above 5% in core EU markets. ATVs high-growth, exports as the second growth curve. 2026/2027 NP cut to ¥2.047B/¥2.423B on FX; 2028 forecast ¥2.925B. P/E 14/12/10x ^[16].
• **[NEW] Kaiyuan Securities / Zhuque-3 first-stage land recovery:** China's first private-rocket first-stage land recovery completed Aug 19; Beijing Stock Exchange lists 15 aerospace names, with Xingtu (920116.BJ) and Minsida (920394.BJ) the headliners ^[17].
• **[NEW] Huatai HK strategy:** Financials and AH premium still have room as USD weakness and RMB strength re-rate HK assets, but HIBOR likely to rise on overseas short-end control of long-end; balanced allocation with low-vol dividend (banks) as core, AI (foundries) as tactical ^[18].
• **[NEW] TFLN 8-inch wafer capacity gap:** A Chinese broker note highlights a beneficiary of AI optical interconnect upgrade and TFLN industrialization validation, with 8-inch wafer scarcity supporting the "sample validation → small batch → mass delivery" path ^[19].
• **[NEW] China Chengxin International:** 1H26 receivables ABS issuance 69 deals, ¥87.085B (+39.14% YoY); 1Y AAAsf median rate 1.65% (-25bp YoY); central SOE share of actual financiers 79.97%, down 13.31pp YoY ^[20].
• **[NEW] July fiscal data + equity tape:** Public budget revenue improved on the tax side; new local special bonds continued ramping; expenditure and infrastructure spending slowed; SSE Composite -0.56% to 3905.2, ChiNext -2.33% on the week ^[21][22].
• **[NEW] Domestic rates (Huatai):** 10Y CGB below 1.7% — don't chase, but direction still favorable; maintain CD + ultra-long combo; watch for amortized-cost bond fund restart ^[23].

5. What Would Falsify the Long-End Story

• Three mutually exclusive paths are live: the market prices one hike, Hartnett prices a policy-driven drawdown, and Morgan Stanley prices a soft landing with debt as a gentle brake ^{[1][2][13][14]}. The falsifiable tests are Nvidia 8/27, August PCE, Jackson Hole Q&A, and any Strait of Hormuz escalation ^{[11][2][3][10][9]}.
• Source quality: Hartnett's 10% S&P drawdown is a single-desk scenario ^[13]; Morgan Stanley's "no collapse" view is one senior voice ^[14]; Warsh's guidance paring is Barclays' expectation, not confirmed ^[9]. The 30Y 5.3% level is one print, not a regime — treat as a band, not a new floor ^[1][3][5].