〔Overnight Brief〕Goldman Calls Hike Pricing 'Too Many' as 10Y Climb ~40bp Tests Equity Bulls; Peters Defends 'Super-Cycle,' Barclays Holds, Morgan Stanley Maps Next AI Wave - USD/JPY Cut to 150

0. Overnight Arc

The overnight tape reframed the rates debate: Goldman's note headlined "Why Markets May Be Pricing in Too Many Fed Rate Hikes" landed the same session the 10-year U.S. Treasury yield posted a ~40bp month-to-date climb that JPMorgan's Grace Peters says is still below the two-standard-deviation level that would truly shake equities ^[1][2]. Net: a credit-supply, oil-and-issuance bid for yields, but a fast-spreading equity-bull rebuttal built on earnings breadth, AI capex, and Barclays' three-reason defense of stocks ^[1][3][4]. Forecasts swung hard: USD/JPY 12-month cut to 150, December Brent raised to $85/bbl, Meta Buy reiterated at $810 ^[5][6][7].

1. The Rates-vs-Equities Debate

• **[ESCALATED] Equity bull case — JPMorgan's Grace Peters:** the 10Y's ~40bp monthly move has "not yet reached two standard deviations," so it cannot truly disturb equities; she prefers stocks on a widening earnings "super-cycle" ^[1]. The causal chain she names: strong growth data, AI-infrastructure bond supply, and oil above $100/bbl driving inflation expectations ^[1].
• **[NEW] Counter — Goldman Sachs:** argues markets may be pricing in too many Fed rate hikes ^[2]. Sourcing is thin in the packet — only a Google News headline is captured — so the thesis is attributable but the specifics are not ^[2].
• **[ONGOING] Bond camp — Federated Hermes' Karen Manna:** last week's 25bp Fed hike confirms the multi-year-high yield regime; resilient growth, inflation concerns, persistent deficits, and a reassessment of the "normal" rate level all matter; near-term price pain, long-term opportunity ^[8].
• **[NEW] Former Dallas Fed President Robert Kaplan:** aggressive hike pricing reflects investors not yet having read Fed Chair nominee Kevin Warsh's reaction function, so markets pre-load hikes as a safety buffer ^[9].
• **[NEW] Single-source caution — JPMorgan relay:** "Not going against the Fed with 100bp hikes" is captured only as a headline with no mechanism — flag as thin ^[10].

2. Forecast Revisions and Stock Calls

• **[NEW] USD/JPY — Goldman (strategist Karen Reichgott Fishman):** 12-month cut to 150 (from 165); 3-month 158 (from 162), 6-month 155 (from 163) ^[5]. Driver: improved Japan policy backdrop plus the possibility of capital repatriation to the home market ^[5].
• **[NEW] Brent — Goldman:** December forecast raised to $85/bbl on rising Asia-Pacific energy prices ^[6]. This sits in tension with the Gelonghui note that Saudi capacity restoration and Trump's allowing Iran's UN delegation have brought spot Brent back below $100/bbl since Sept 16, 2026 ^[6][11].
• **[NEW] Meta (META) — Bank of America Securities:** Buy reiterated, $810 target, after Meta Connect 2026 and the launch of the personal AI agent Muse ^[7]. Partnerships cited with Walmart, Best Buy, Shopify, Expedia, and Instacart; future integration with Meta smart glasses and Reality Labs; commission monetization remains early-stage ^[7].
• **[NEW] United Airlines Holdings (UAL) — UBS:** target raised from $137 to $146 ^[12].
• **[NEW] Edison International (EIX) — UBS:** target cut from $63 to $57 ^[13].

3. Thematic Research: AI, Cloud, and Humanoids

• **[NEW] Cloud capex — Morgan Stanley executive director Zhi Tian Yan, GMIF2026 (Shenzhen):** 2026 global cloud capex may top $900bn with storage at 48% (ex-HBM); 2027 projected at $1.2T with storage at 53%; 2028 could reach $1.6T ^[14]. The 2026 capex/EBITDA ratio at Google, Amazon, Microsoft, and Meta has already crossed 70%, with some 2027 figures set to exceed EBITDA; 2027 HBM consumption estimated at ~$44bn ^[14].
• **[NEW] AI positioning — Morgan Stanley (via MarketWatch):** AI hardware still has room to run, but the next wave calls for diversification into industries starting to realize AI benefits ^[3].
• **[NEW] Barclays cross-asset:** three reasons to stick with stocks despite rising yields — the strategists recommend exposure to U.S. technology, media, telecommunications, and industrial stocks ^[4].
• **[NEW] Humanoid robots — Morgan Stanley supply-chain note:** OEMs are ratcheting 2026 shipment targets by several multiples; supply chain is being retooled for thousand-unit and higher production cadences; one North American OEM is targeting a weekly production milestone by July 2026 ^[15]. The bottleneck has shifted from "can it be built" to consistent yield, cost control, and global delivery ^[15].

4. China A-Share, Sectors, and Banking

• **[NEW] Sector rotation — Sinolink Securities:** model balanced across cyclical (coal, non-ferrous, utilities), financial-property (real estate, non-bank financials), and tech-manufacturing (electronics) ^[16]. Real estate rises to #1 on consensus and momentum, with ETF flows as a drag; non-bank financials #2 on earnings improvement; coal slips to #3 ^[16].
• **[NEW] A-share weekly — Sina Finance relay:** for Sept 17-23, Shanghai Composite +1.15%, ChiNext +2.06%, CSI 500 +1.41%, CSI 300 +0.83%; sector leaders real estate, pharma, beauty-care; laggards utilities, non-ferrous, banks; two-market turnover 9.83T yuan, daily average 1.97T yuan, up 226.4bn yuan vs the prior five sessions ^[17].
• **[NEW] New infrastructure tracking — Sina Finance relay:** the traditional "steel-cement + heavy machinery" yardstick misreads new infrastructure; the proposed rewrite tracks "chips-servers + power-energy" via special bonds and ultra-long treasury allocations, IT and power-heat industry investment growth, computing-center project counts, 5G tender volumes, and IT industry electricity use ^[18].
• **[NEW] Capital-market reform — Ping An strategist review:** A-share dividend-plus-buyback exceeded equity issuance for the first time in 2022; 2024-25 the ratio was ~3.15x; reform direction is duration-matching between funding and assets ^[19].
• **[NEW] Banking — Liu Xiaoshu:** the "value bank" framing is under pressure from loan growth slowing to 4.9% (Aug 2026) from 12.3% (end-2019), net interest margin compressing to 1.41% from 2.2%, and the loss-grade share of NPLs rising versus 2019 ^[20].

5. Falsifiable Tests and Source Quality

• **[NEW] Oil-recession signal — research relay:** the historical pattern that oil surges precede U.S. recessions broke down around 2010 as the U.S. shifted from net oil importer to net exporter, which reframes the inflation channel Peters cites ^[1][21].
• **Source quality control:** the Goldman "too many hikes" thesis is captured only as a Google News headline, so the argument is attributable but the specifics are not ^[2]; the JPMorgan "100bp" item is a one-line headline with no mechanism ^[10]; the Meta target, USD/JPY cuts, UBS target changes, and Brent revision are crisp and directly attributable ^{[5][6][7][12][13]}.
• **Falsifiable test:** the next 10/30-year Treasury auction tail and any Fed-speaker reinforcement of the Warsh reaction function will determine whether the ~40bp 10Y move is absorbed as a "better entry" (Peters' read) or breaks the equity bull case ^[1][9]. Print the band, not a point — divergence between Peters' "not yet two sigma" and Goldman's "too many hikes" is itself the content ^[1][2].