〔Overnight Brief〕30Y Past 5.3% For First Time Since 2007; Warsh-Era Hike Bets Widen 18bp → 45bp Peak; Tech Lurches Into Secondary Bottom — September Rebound Hangs on Long-End Cooling

0. Overnight Arc

The 30-year US Treasury yield broke past 5.3% intraday on August 17 — the highest since 2007 — and the move has been digesting through global tech since ^[1]. Since Fed Chair Warsh's first FOMC on June 17, full-year hike expectations widened from 18bp to a 45bp peak on July 23, embedding USD-liquidity uncertainty into long-end rates ^[1]. From July 28 to August 18, the 30-year rose 19bp and the 10-year rose 10bp, with the 2-year flat — the steepening sits in the long end only ^[1]. The arc: long-end repricing → tech de-rating → secondary bottom forming → September rebound contingent on liquidity anxiety fading ^[1][2].

1. The Rate Mechanism

• **[NEW] 30Y intraday past 5.3% on Aug 17, first time since 2007 ^[1].**
• **[ONGOING] Warsh-era repricing:** post-June 17 FOMC, full-year Fed hike expectations widened 18bp → 45bp peak by July 23 ^[1].
• **[NEW] Curve steepening is long-end only:** July 28 to Aug 18, 30Y +19bp, 10Y +10bp, 2Y flat ^[1].
• **[ONGOING] 1990s analog (broker research):** US internet stocks saw staged pressure from rising Treasury yields; the next leg up came when liquidity-tightening expectations paused ^[1].
• **[ONGOING] A-share M-top analog:** high-prosperity industries typically form M-tops, with secondary highs roughly 6 months after and at 8-9/10ths the peak — the current secondary high needs overseas liquidity and trading congestion to ease ^[1].

2. Broker Read: Bottom-Building, Not a Bear Market

• **[NEW] Shenwan Hongyuan strategy weekly:** secondary bottom realized; rebound may extend to late September ^[2].
• **[ESCALATED] AI chain 'four steps':** the recent move matches their projected path of AI industry chain regaining strength, but the funding positive cycle is unlikely to return to the late-June state quickly ^[2].
• **[ONGOING] Higher bar for AI catalysts:** versus 2025 experience, this round of AI chain restart needs a higher-grade industry catalyst and likely more time ^[2].
• **[NEW] SWS strategy weekly:** current sector rotation supports a market bottom; balanced allocation, with growth possibly diffusing to AI applications alongside dividends and domestic-demand sectors such as healthcare ^[1].
• **[ESCALATED] If September fails to print a new high:** market expectation shifts to tech adjustment lengthening from monthly to quarterly cadence ^[2].

3. Contrarian: The Buyback-Collapse Narrative

• **[NEW] Daniel Lacalle (economist), X post:** flags Izabella Kaminska's FT piece calling out the 'buyback = collapse of America' narrative as 'running away with itself' ^[3].
• **[NEW] Lacalle cites the narrative's spread:** 'FT. Dalio. Currie. Random dude on X' — and calls the narrative 'the flavour of the month for the internet's armchair experts' ^[3].
• **[NEW] Lacalle's counter:** 'the smartest non ideological money that has specialised in the weeds of this market' sees through it — but the post is truncated and names no specific fund or position ^[3].

4. Source Quality and What Falsifies the Bottom Thesis

• **Source control:** Material ^[4] is a headline only with no body text; ^[5] is an IEA promotional post on energy tech manufacturing with no specific data points; only ^[1] and ^[2] carry substantive figures, and both are Chinese-broker research (SWS, Shenwan Hongyuan) ^[4][1][5][2].
• **Cross-confirmation:** the 30Y > 5.3% figure on August 17 appears in two independent Chinese research notes ^[1][2]; the buyback-collapse debate rests on a single truncated X thread ^[3].
• **Falsifiable test:** whether the long-end auction calendar and any Warsh-era FOMC guidance cool the 30Y back below 5.2%, and whether September AI catalysts land at the magnitude Shenwan Hongyuan requires ^[1][2].
• **Dovish tell:** the 2-year's flatness during the 30Y/10Y steepening suggests the front end is not yet pricing a sustained hike regime; if the 2-year starts to follow, the bottom thesis breaks ^[1].