〔Overnight Brief〕MS Lifts Q4 Brent to $105 as Mideast Premium Reflates; PIMCO Warns Yields Haven't Topped With One to Two Hikes Left; Rosenberg Flags 40-Year Bubble Extremes — Late-Cycle Energy Meets Crowded Risk Trades

0. Overnight Arc

The overnight brief carries one consistent message: late-cycle energy stays bid even as the bond trade splits three ways. Morgan Stanley pushed Q4 Brent to $105/bbl, +$5 from prior, on repeated Mideast risk ^[1]; Goldman argued TTF gas needs to clear €105/MWh in a tight-supply, normal-weather winter to choke off non-European LNG demand ^[2]. On rates, PIMCO's Daniel Ivascyn said US Treasury yields "haven't peaked" and the Fed has one to two more hikes to deliver ^[3], while BlackRock turned constructive on government bonds for the first time in six years ^[4] and KKR framed the selloff as an Asian IG entry near 5.8% yields ^[5]. Risk assets are now on their own warning track: Rosenberg called US equities the most dangerous setup of his 40-year career ^[6][7], and Goldman's Molavi said two of three AI-capex inflection signals have already triggered ^[8]. The falsifiable tests are the October Fed decision and the next hyperscaler capex print ^[8][9].

1. Energy: Mideast Premium Reflates the Complex

• **[NEW] Morgan Stanley:** raised Q4 Brent forecast by $5 to $105/bbl, citing Mideast risk ^[1].
• **[NEW] Goldman:** TTF gas winter peak must reach €105/MWh under a tight-supply, normal-weather scenario to incentivize enough non-European LNG demand cuts ^[2].
• **[NEW] Huatai Securities (research note):** flagged shipping-rate upside and courier unit-economics repair as two core October lines; high-dividend roads and rail passenger are the lead plays, with air-cargo and hazmat logistics as secondary ^[10]. Air-passenger expectations are low, and a pullback in oil could be the next catalyst ^[10].

2. Rates: Yields Haven't Peaked, but Contrarians Are Circling

• **[NEW] PIMCO CIO Daniel Ivascyn** (at Grant's Fall Conference): expects the Fed on hold this month but with one to two more hikes left in the cycle; Treasury yields have further to rise ^[3].
• **[NEW] BlackRock:** turned constructive on government bonds "for the first time in six years" ^[4]. Single-source headline; treat as a positioning flag.
• **[NEW] KKR's Dengzhao Pan, Head of Asia Liquid Credit:** Asia IG USD bonds average 5.8% yields, with longer-dated and subordinated paper higher; AUM $796B as of end-June ^[5].
• **[NEW] BofA's Michael Hartnett:** argues an October Fed hike would be the "biggest tailwind" for US equities — the disease is the long end (10Y touched 5.36%, 30Y at 2002 highs), not the front end ^[11]. Contrarian single-house view; print alongside PIMCO's ^[3].
• **[NEW] Morgan Stanley China chief economist Xing Ziqiang** (closed-door session): Q4 GDP path of 4.4-4.5% on the policy package; framed the mortgage-rate subsidy as a "breakthrough signal" of fiscal balance-sheet use ^[12]. Flagged K-shaped US consumption and the risk of mid-term election friction hitting AI data-center capex ^[12]. Single-source relay.
• **[NEW] Nobel-economics context (NPR):** sticky prices explain why recessions hurt and what policymakers can do about them ^[13]. Framing backdrop only.

3. AI Capex: Two of Three Inflection Signals Already Lit

• **[NEW] Goldman's Bobby Molavi:** AI ecosystem assets now $25-30T; two of three inflection signals have already triggered — price/real-earnings divergence and credit re-pricing of AI borrowers — with hyperscaler capex growth forecast to peak in Q3 2026 and decelerate quarter-by-quarter thereafter ^[8]. US long/short net exposure has dropped to ~46%, near the April-2025 sell-off low ^[8].
• **[NEW] Jinyuan Securities (RSI risk note):** recursive self-improvement is the next capex-debate phase; cited Anthropic's call for a third-party risk evaluator, backed by OpenAI and SpaceX CEOs but opposed by Trump ^[14].
• **[NEW] Macro theme research (Sina Finance):** US-China AI competition is settling into a "partial decoupling, partial deal-making, autonomous substitution" coexistence pattern; chip and compute gaps remain the binding constraint ^[15].
• **[NEW] AI hardware demand (single-source company note):** high-precision CNC machine tool orders activated by AI hardware; one company's pan-AI new orders +350% y/y, with traditional-segment demand also warming ^[16]. Company name not in headline.

4. Bubble Flags: Rosenberg's 40-Year Extremes

• **[NEW] David Rosenberg (Rosenberg Research & Associates):** US equities increasingly look like a bubble, with valuation, leverage and positioning at extremes not seen in his 40-year career; he has been buying US Treasuries and overseas equities, calling the US a "tech-boom center" investors are overpaying for ^[6][7]. Consumer reliance on further stock gains to fund lifestyle is a separate fragility he flagged ^[6].
• **[NEW] Truist's Keith Lerner, CIO and chief market strategist:** the four-year bull market has "room to run" with AI and tech still central; stocks bounced from session lows on speculation the Fed won't be forced into aggressive hikes ^[9]. Contrarian to Rosenberg ^[6] — print both, not a verdict.
• **[NEW] KBW's Chris McGratty, Head of US Bank Research:** large banks may still appeal to investors; consumer health, rates and consolidation are the three earnings drivers ^[17]. Single-source preview.

5. China & Asia Allocation

• **[NEW] Hong Kong:** Hang Seng reclaimed 24,000; allocation debate centers on dividends vs. growth, per China Securities Journal coverage ^[18].
• **[NEW] Yageo (MLCC):** planning a late-October to November price hike — high-cap products +50% or more, mid/low-cap +10-30% — likely to widen other makers' pricing room ^[19].
• **[NEW] El Niño 2026-2027 (Sina Finance macro research):** eastern-type super El Niño formed September 2026, expected to last through spring 2027; base-case food-CPI contribution +0.20pp in 2026, +2.51pp in 2027; rubber, copper and sugar supply chains flagged as the most exposed ^[20].
• **[NEW] Stablecoin vs. deposits (BIS research):** using weekly US data Jan-2019 to Jun-2025, deposit-rate rises predict slower stablecoin growth, supporting a substitute relationship ^[21]. Academic backdrop, not a trade.