〔Day Digest〕Warsh's Jackson Hole Re-Credentials the Hawk as A-Share September Playbooks Cluster Constructive on 9-of-16 Historical Wins — Hormuz Capex Freeze Tests the Equipment Channel Behind Unitree's 54x 2027E PS

0. Weekly Arc

Three live tensions define the research tape: a Fed re-asserting hawkish credibility under Chair Warsh ^[1][2], A-share September playbooks that converge on a constructive bias ^[3][4][5], and a geopolitical-macro backdrop — Hormuz stalemate, AI-debt sustainability — that is starting to bite traditional capex ^[6][7]. The cross-cutting contradiction: a tighter-for-longer Fed is being read as net-supportive for A-shapes because it restores policy discipline ^[2], even as the same tighter regime plus a Hormuz freeze threatens the China mid-stream equipment channel that the September bull case implicitly relies on ^[3][6]. Single-stock risk is a third live wire: UBS's first coverage on Unitree asks whether a 54x 2027E PS leaves room for further re-rating ^[8].

1. Fed Communication Reset

• **[NEW] CICC's read on Warsh's Jackson Hole address — 'late hawk, credibility repair':** Warsh conceded inflation is still elevated, named the policy rate as the main tool, said the Fed will 'act as circumstances require,' and pinned 65 months of above-target inflation on the central bank itself, correcting July's 'let the market substitute for the Fed' framing ^[2]. CICC's punchline: hike odds are up, but liquidity is not the binding constraint — discipline and predictability are ^[2].
• **[NEW] ICBC International (Chief Economist Cheng Shi and macro analyst Zhou Ye):** a more measured read of the same speech — Warsh kept conditionality, preserved decision space, and is deliberately shrinking explicit forward guidance to enlarge market-led price discovery; the symposium theme ('financial innovation's impact on payments and policy') flags AI, payment rails, and the neutral rate as variables to re-evaluate ^[1]. Lower path transparency does not mean lower Fed dependence, but it does raise the bar for credibility ^[1]. Two opposite-leaning reads of one speech are themselves the story ^[1][2].
• **[ESCALATED] A-share translation:** September-strategy authors now treat a credible hawkish Fed as net-supportive for policy discipline rather than net-negative for risk ^[2], even as the macro tape prices Hormuz-driven rate upside ^[4] and AI-debt sustainability risk ^[7].

2. A-Share September Playbooks

• **[NEW] Huajin Securities (Sept strategy):** 9 of 16 Septembers for the Shanghai Composite since 2010 were up; drivers are policy/external events, fundamentals, and liquidity ^[3]. Constructive case: policy stays active, U.S.-Iran talks could settle, a U.S.-China summit may de-escalate relations, exports and high-tech manufacturing investment stay firm, and retail sales bottom out ^[3]. Caveat: property investment stays weak ^[3].
• **[NEW] Zheshang (Sept strategy):** the macro view explicitly attributes the upside surprise in rates to the Hormuz blockade; strategy shifts from a K-shape to a 'diamond'; the quant view argues AI's tech narrative pivots from far-end storytelling to mid-cycle supply-demand verification ^[4]. The named September gold-stock basket spans agriculture (China Youran Dairy), coal (Yankuang Energy), machinery (Slac), building materials (Honglu Steel Structure), pharma (WuXi Biologics), non-bank finance (China Pacific), transport (Milkyway Chemical Supply), tech (Sinosig Technology), and banking (Bank of Nanjing) ^[4].
• **[ONGOING] Weekly strategy (2026-08-24 to 2026-08-28):** indices pulled back then rebounded to close mostly higher; the 'retreat tested support' and a mid-line rebound toward higher highs is the base case; the note recommends adding on dips, avoiding chasing after sharp runs, and keeping a tactical tilt to innovative drugs, brokers, Hang Seng Tech, media, computer, non-ferrous metals, SOEs, plus property and agriculture on El Niño risk ^[5].

3. Geopolitics and the Equipment Capex Channel

• **[ESCALATED] Macro note on Hormuz (sina/vReport):** May–July U.S.-Iran whiplash has visibly depressed traditional global firms' capex expectations — the dominant risk to mid-stream equipment demand over the next 6–12 months; the Trump 'TACO' trade has capped geopolitical risk at the headline level but not at the real-economy level ^[6]. The China-manufacturing-going-overseas bull case requires Hormuz shipping to migrate from 'shock-and-repair' to 'stable repair'; the report lists three pathways — a U.S.-Iran stable deal, a U.S.-China summit, or alternative routing ^[6].
• **[NEW] September-strategy cross-link:** Zheshang's macro view explicitly attributes upside rate risk to the Hormuz blockade, feeding back into the policy narrative above ^[4].
• **[ONGOING] Contradiction worth flagging:** Huajin's constructive September base case assumes U.S.-Iran talks succeed and a U.S.-China summit de-escalates ^[3], which is precisely the set of conditions the equipment note flags as not yet met ^[6].

4. Single-Stock Deep Dive — Unitree

• **[NEW] UBS first coverage on Unitree Robotics:** described as a hardware-first humanoid whole-machine leader that lifted annual unit shipments from 5 to 5,200+ over two years and captured ~30% of global shipments, one of the few profitable whole-machine vendors ^[8]. The framework decomposes capability into 'brain' (perception/planning), 'cerebellum' (motion control/balance), and 'body' (execution/reliability): Unitree leads on the last two, the 'brain' is an industry-wide bottleneck ^[8].
• **Valuation check (UBS):** at ~54x 2027E PS, the multiple already discounts the hardware lead and a 53% 2026-28 revenue CAGR; further re-rating needs clearer 'brain' progress and a sustainable path from revenue to profit ^[8]. Hard numbers: 2022-25 revenue CAGR 140%, gross margin from 43.6% to 60.3%, 2025 net profit 278 million yuan; IPO on 19 Aug on the STAR Market at 150.8 yuan, trading near 4x the issue price after a 6.099 billion-yuan raise; ~39% of IPO proceeds earmarked for embodied-AI R&D, with a DeepSeek strategic partnership signed in August ^[8].
• **Mix caveat:** 76% of Unitree's humanoid revenue comes from research and education, while industrial applications were only 3% of global humanoid demand in 2025 ^[8]. The 'brain' gap is the falsifiable test for the multiple ^[8].

5. Cross-Asset Plumbing and the AI Debt Question

• **[ONGOING] Stock–interbank funds 'seesaw' does not hold:** a strategy note shows the empirical link is unstable; the 63 BSE IPOs from Jan 2025 to Jul 2026 lifted GC001 by ~13.7bp on issue days but moved R001 and DR001 by only ~0.3bp and ~0.2bp respectively, so the plumbing runs through GC/R before DR ^[9]. Implication: equity volatility moves the policy tolerance for low funding rates, not the rate level itself; the right reads are PBOC operations, big-bank lending, and DR ^[9].
• **[NEW] AI debt sustainability (Zheshang Sept strategy):** the report argues AI's exponential-revenue phase is pausing — token prices have fallen, vendor revenue growth has slowed, and the debt/TOKEN ratio is back up — meaning the path of 'AI firms borrow today, growth repays later' is starting to fail ^[7]. The implication runs through gold: 2024's debt-sustainability fear decoupled gold from real rates, and the H1 AI-revenue rebound re-anchored it; if the AI-revenue path breaks, the real-rate link loosens again ^[7].
• **Cross-source tension:** Huajin and the weekly strategy are pro-risk on the China-equity side ^[3][5], while the Zheshang debt-degradation note is the strongest single-source hawkish input on the tape ^[7]. Source quality: items ^[1], ^[6], ^[2] are syndicated analyst columns; ^[8] is a single-broker initiation; ^[7] is a single-broker monthly strategy — treat the AI-debt claim as a hypothesis, not a consensus.