〔Day Digest〕AI Platform Upside Meets a 15% CoreWeave Credit-Bond Yield and China's 15.5% Industrial-Profit Growth — The Bottom Is Still Unfinished

0. Daily Arc

• **[NEW] AI split:** UBS argues that markets are overpaying for AI spending burdens while underpricing the upside for internet platforms, given uncertainty over how AI will reshape discovery, content and enterprise workflows ^[1]. CICC provides the counterweight, warning that CoreWeave and listed neoclouds display the leverage and cash-flow traits of weak AI-financing borrowers ^[2].
• **[NEW] China split:** Industrial profits accelerated, but Guojin Securities says fear in three major equity indexes is near historical percentiles without full positioning clearance, allowing a short-term rebound while leaving midterm deleveraging intact if no strong external catalyst arrives ^[3][4].
• **[ESCALATED] Policy split:** The October Fed held rates despite two votes favoring a 25bp increase, while describing growth as strong and making financial conditions the main channel for implementing policy ^[5].

1. AI: Operating Upside, Credit Fragility

• **[NEW] Investment shape:** UBS says hyperscalers are committing infrastructure funds equal to roughly one year of core cash flow, while application-driven platforms can invest through R&D and retain more flexibility ^[1]. It argues valuations should separate temporary investment, recurring participation costs and permanent business changes, with capital, efficiency and ecosystems shaping the industry ^[1].
• **[NEW] Structural winners:** UBS favors capital-rich, full-stack platforms with mature customer touchpoints and scenario information, while leaving room for differentiated-technology companies or specialist laboratories ^[1].
• **[ESCALATED] Credit repricing:** CoreWeave credit-bond yields rose from around 10% in the first half to 15%; its five-year CDS spread exceeded 1,100bp, implying about a 60% cumulative five-year risk-neutral default probability ^[2].
• **[ESCALATED] Balance-sheet burden:** The four listed neoclouds carried roughly 7.5x net debt/EBITDA in 2026 versus about 0.1x for six large technology companies, while their 2025 asset-liability ratio was about 84% versus 45% ^[2].
• **[ESCALATED] Funding need:** CoreWeave had roughly $35 billion of interest-bearing debt as of Q2, with 29% in unsecured credit bonds, 39% in secured DDTL loans and 90% of all debt carrying parent recourse ^[2]. Based on 2.7GW of contracted pending capacity, $30 billion/GW of investment cost and 20% order prepayments, CICC statically estimated another $27 billion of financing was needed ^[2].
• **Falsifier:** The credit-risk case strengthens if yields and CDS keep widening, free cash flow stays negative and the financing gap cannot be closed; it weakens if funding access improves ^[2].

2. Policy and US Equity Crosscurrents

• **[ESCALATED] October FOMC:** The Fed held rates 10-2, with FOMC voters Cook and Hammack preferring a 25bp increase, and shifted its economic description from solid to strong ^[5].
• **[ESCALATED] Chair Warsh:** Fed Chair Kevin Warsh said stronger growth did not mean the economy was overheating, inflation had not reached target and no forward guidance was provided ^[6][5]. He identified tighter financial conditions as the main way policy is implemented and said interest rates still affect AI investment ^[5].
• **[ESCALATED] Contradictory path:** Prudential chief U.S. economist Blerina Uruci argued that the Fed's slow-hike strategy carries risk and that it may need to catch up with faster tightening in next year's first quarter ^[6]. This warning arrives through a short secondary flash and is thinner than the detailed FOMC recap ^[6][5].
• **[ONGOING] Internal split:** A BNP Paribas headline says Cook's first dissent and another hawk's retreat are changing the Fed's hawkish bloc, but the packet contains no supporting vote detail beyond that headline ^[7].
• **[ONGOING] Two equity scripts:** Deutsche Bank's Jim Reid, Binky Chadha and Parag Thatte cite S&P 500 data since 1928 and say a typical year's final low arrives around October 27 before a largely one-way advance to year-end ^[8]. BTIG chief technical strategist Jonathan Krinsky counters that the rebound lacks true capitulation and that complacent options positioning resembles pre-bottom periods, using the 2000 Philadelphia Semiconductor Index collapse as the bear template ^[8].

3. China Earnings, Commodities and Flows

• **[NEW] Industrial profits:** China industrial profits rose 15.5% in January-September and 14.4% in September versus 4.2% in August, helped by a low base and industrial-price repair ^[4]. Electronics profits increased 103.8% and contributed 9.6 percentage points, while raw-material manufacturing rose 44.9% and contributed 5.9 points; nonferrous metals grew 79.7%, chemicals 49.2%, coal 57.0%, and petroleum processing returned to profit ^[4].
• **[ONGOING] Quality of recovery:** Overseas AI-capex guidance diverged, with Microsoft cutting capital expenditure and Meta launching large-scale equity financing, while receivables and finished-goods inventories accelerated against weak demand ^[4]. A separate sector tracker said three-month rolling industrial-profit growth continued to slow under a high base and high PPI, favoring nonferrous metals, coal, petrochemicals, banks, gas and semiconductor equipment ^[9].
• **[NEW] Unfinished bottom:** Guojin Securities says fear in the ChiNext Index, CSI 1000 and STAR 50 is near historical percentiles, but chip and positioning clearance remains incomplete; short-term rebounds are possible, while midterm deleveraging persists without a strong external variable ^[3]. The CSI 300 equity risk premium reached its 10-year mean plus one standard deviation, with sustained high-intensity ETF inflows requiring the premium to rise further ^[3].
• **[NEW] Coal:** The 2026 thousand-yuan coal-price rally was supply-driven, reflecting tighter safety inspections, mine shutdowns, rectification work and delayed restarts rather than a demand surge ^[10]. Super El Niño could affect coal through greater southern hydropower substitution, winter heating volatility and reduced Indonesian exports, with some effects lagging into 2027 ^[10].
• **[ONGOING] Agriculture:** Geopolitical disruption, higher diesel costs, Black Sea logistics risk and historically low tradable stock-to-use ratios leave agricultural supply vulnerable to rapid price transmission ^[11]. El Niño has formed and is expected to peak in Q4, so its impact on production may not become visible until year-end or next year ^[11].
• **[NEW] Pricing influence:** Shanghai Futures Exchange commodity director Chen Ye said the Shanghai Price futures system has gained influence, with the exchange representing 23.04% of national futures volume and 35.5% of turnover in the first three quarters; the next debate is extending that influence into cross-border physical trade ^[12].
• **[NEW] Singapore property:** FCT, SUN and KDCREIT distribution-per-unit results slightly exceeded expectations, while KREIT proposed a S$626.6 million purchase of 51% of Keppel South Central at a 4.3% yield with lease-up revenue support ^[13]. Singapore REITs gained 2% for the week but recorded S$1.7 billion of institutional outflows year-to-date, while non-REIT property shares received S$50.5 million of inflows; DayOne is considering delaying or reducing its U.S. IPO as AI-related sentiment worsens ^[13].

4. Corporate Calls and Falsifiers

• **[NEW] Selected ratings:** Goldman Sachs cut Ford's target to $12 from $14 and retained Neutral ^[14]. Piper Sandler raised Mastercard's target to $755 from $710 after Q3 revenue grew 14% to $9.8 billion ^[15]. Evercore ISI cut Sandisk's target to $2,500 from $2,800, while Morgan Stanley cut Amazon's to $335 from $350 ^[16][17].
• **[NEW] China power:** Citi upgraded China Nuclear Power to Neutral with a RMB 10.50 target ^[18].
• **[NEW] Optical chain:** Guosheng Securities retained Buy on Xinyisheng after Q3 net profit reached RMB 3.84 billion, up 61.15% year over year but down 19.1% quarter over quarter; January-September net profit was RMB 11.37 billion, up 79.7% ^[19]. Guosheng says the company expects delayed shipments and revenue recognition in Q4, retained its 2027-2028 forecasts, and assigned 2026-2028 revenue of RMB 52.04 billion/RMB 109.41 billion/RMB 157.66 billion, net profit of RMB 19.64 billion/RMB 41.95 billion/RMB 61.06 billion and PEs of 16.2x/7.6x/5.2x ^[19].
• **[NEW] Testing and phosphates:** Zhongyou Securities kept Buy on Huace Testing after January-September net profit reached RMB 959 million, up 18.04%, and Q3 net profit reached RMB 395 million, up 14.52%; overseas revenue rose 68.34%, and 2026-2028 PEs were 21.70x/18.58x/16.12x ^[20]. Huayuan Securities kept Buy on Chuanheng despite January-September adjusted net profit of RMB 925 million, down 1.77%, and Q3 net profit of RMB 368 million, down 14.12% year over year, citing a 174.44% year-over-year rise in the domestic sulfuric-acid market average and incomplete downstream cost pass-through ^[21].
• **[NEW] Diversified consumer:** BOC International Securities kept Overweight on Lingnan Holdings after January-September net profit reached RMB 75 million, up 14.20%, and Q3 net profit reached RMB 18 million, up 15.04%; a new restaurant opened in August and LNLIFE extends the company into cultural-creative retail ^[22].
• **[ONGOING] Thin-source watchlist:** Citi's private-credit allocation call, First Financial's Hongfa order-and-pricing assessment and Huatai's preference for quality financial stocks are available only as headline or teaser summaries ^[23][24][25].
• **AI falsifier:** The bullish operating case weakens if CoreWeave and neocloud credit spreads continue widening, free cash flow remains negative and the estimated $27 billion financing requirement cannot be refinanced ^[2].
• **Bottom falsifier:** The durable-bottom case weakens if ETF inflows remain weak despite the CSI 300 risk-premium threshold, positioning fails to clear further or the external catalyst needed to arrest deleveraging does not arrive ^[3].
• **Profit falsifier:** The industrial-recovery case weakens if receivables and inventories keep accelerating, overseas AI-capex guidance remains fragmented and domestic policy fails to reach property, autos, power and other traditional industries ^[4].