Goldman Sachs and Nomura consistently confirmed a significant weakening in China's credit in 8: new RMB loans were only RMB 59-60bn and social financing was RMB 1.66tn, both far below market expectations. Stock social financing and loan growth rates fell to historic lows of 7.2% and 4.9%, respectively. Household loans turned negative to -RMB 203bn (consumption -122bn, mortgages -82bn), while corporate loans of +RMB 260bn remained below seasonal norms, indicating continued sluggish private sector credit demand.

The resilience in social financing mainly comes from government bonds (RMB 1.01tn, accounting for over 60% of new additions) and direct financing channels such as corporate bonds and equity financing. Fiscal policy and direct financing are compensating for weak private sector credit demand, determining the structure and sustainability of credit expansion.

Nomura observed marginal liquidity easing (DR007 daily average fell to 1.399%, close to the seven-day reverse repo rate; 10-year government bond yield at 1.698%; PBoC net injection of RMB 48bn) but this failed to translate into credit expansion, with constraints stemming from the demand side rather than the supply side. It expects fiscal policy to play a dominant role in the coming months, but net government bond financing in 8 falling to RMB 1,010bn has become a drag again, making it unlikely that fiscal measures alone can reverse the slowdown in loan growth.

8 activity data showed high imbalance: industrial production rose 5.2% YoY, beating expectations (driven by exports +25.0% and easing oil supply disruptions), but nominal retail sales grew only +0.4% (Nomura estimates actual at -0.4%), and fixed asset investment contracted by 10.6% YoY for the fifth consecutive month. Both manufacturing and infrastructure investment weakened, with strong exports masking weak domestic demand.

Goldman Sachs lowered its Q3 GDP YoY forecast to 4.4%, full-year 2026 to 4.5%, and 2027 to 4.6%. Nomura maintained its below-consensus forecasts of Q3 at 4.3% and Q4 at 4.5%. Both firms believe weak domestic demand poses downside risks to growth, with recent easing measures providing only moderate support.

Consumption categories showed clear divergence: smartphone sales growth accelerated from 20.4% to 27.3% (due to component cost inflation and premiumization), home appliances recovered to +2.3%, but gold/silver/jewelry fell by 17.5%, furniture by 7.9%, building decoration materials by 11.8%, and automobiles by 18.5. Discretionary and property-related categories remain under pressure; service consumption outperformed goods consumption (dining +1.1% vs goods +0.3%).

Deutsche Bank pointed out that China's actual GDP growth in 7 slowed to 4.1%, below the government's target range of 4.5%-5.0%, but signs of stabilization emerged in 8 (manufacturing PMI rose to 49.8, exports +25.0% YoY, seasonally adjusted trade surplus hit a record high of USD 112.8bn). Stronger fiscal issuance, structural relending, and interest subsidies will support demand; the CPI forecast for 2026 remains at 1.2%, while PPI was revised up from 2.9% to 3.0% due to higher oil prices.

Goldman Sachs calculated that seasonally adjusted, population-weighted new home prices in 70 cities saw their MoM annualized decline narrow to 0.4% in 8 (from 0.9% in 7), with YoY declines narrowing to 3.0%. Tier-one cities saw MoM annualized increases of 2.4% (led by Beijing), tier-two cities fell by 0.2%, and tier-three city declines narrowed to 1.9%. It is expected that housing prices in tier-one and tier-two cities will generally stabilize within the next one to two years, with Shenzhen and Shanghai potentially leading.

Weekly data showed divergence: new home GFA sales in approximately 75 cities fell by 4% MoM but rose by 5% YoY. It is inferred that contract sales for the top 100 developers in 9 may fall by 16% YoY (compared to -4% in 8). Meanwhile, second-hand home transactions in approximately 20 cities rose by 6% MoM, subscriptions increased by 14%, and visits rose by 13%. Brokerage price expectations improved for the third consecutive week. Subscriptions typically lead registered contracts by 1-2 weeks, potentially signaling short-term transaction improvements.

Nomura pointed out that real estate investment fell by 25.4% YoY, with sales, starts, completions, and funding all deteriorating comprehensively. Second-hand home prices fell by 0.31% MoM, and the Iceberg index declined by 0.60% in both 8 and the first two weeks of 9. Goldman Sachs estimated that commercial housing sales area in 8 fell by 14.7%, new starts by 31.0%, completions by 28.4%, and construction by 12.8%, but some major cities showed early signs of stabilization.

8-28 real estate reforms drive the industry's shift from high-leverage pre-sales to completed home delivery models. Safer home purchases and lower monthly mortgage payments support sales, but stricter financing will limit new projects, real estate investment, and local government land transfer revenues. Xiamen became the first major city to attach completed home sales requirements to land auctions, with most plots sold at reserve prices, cooling the land market.

Covered developers' valuations are at historic down-cycle lows: offshore and onshore discounts to estimated end-of-2026 NAV are 43% and 24%, respectively, with 2026E P/B at 0.4x, comparable to 2H2008, 2H2011, and 1H2014, indicating that pessimistic expectations are already priced in.

Goldman Sachs expects the FOMC to hike by 25bp in 9, the ECB to hike by another 25bp in 12, the BoE to hike by 25bp in 11, and the BoJ to hike by 25bp this week and again in 1 of 2027. Market expectations for Fed/ECB/BoE rate hikes through end-Q27 of 1 have increased by a total of 25bp, with the probability of hikes exceeding 100bp in the next 12 months at approximately 40% (approximately 20% in 8). Forecasts for end-2026 US 2-year/10-year yields are 4.30%/4.75%, Germany 10-year at 3.25%, and UK at 5.00%.

Nomura's tightening expectations are significantly lower than market pricing: It expects the Fed to hike by 25bp each in 9 and 12 of 2026 before holding until 2027, the ECB to reach a deposit rate of 3.00%, the BOJ to reach 1.75%, and India to hike by 50bp in 2026, while the BOE and BoC remain unchanged. This is viewed as a policy recalibration rather than the start of a long-term tightening cycle, as rates are near neutral, cost-push inflation is suppressing growth, and rising bond yields are already tightening financial conditions.

Deutsche Bank believes the 2020s have shifted from a demand-constrained, low-inflation environment to a supply-driven mechanism, where limited spare capacity makes inflation spikes more frequent. Traditional monetary stimulus is less effective in supply-constrained environments, while fiscal stimulus is more inflationary and more likely to crowd out private investment. Policy focus must shift to expanding the production frontier.

Deutsche Bank points out that the market has priced in 3.9 Fed rate hikes in the next 12 months, but risks lean towards Chair Warsh not endorsing a consecutive hiking path, supported by slowing core CPI. Key macro risks include Warsh not endorsing consecutive hikes or a shift in the Trump administration's policy towards Iran.

Goldman Sachs cross-asset allocation: Tactical defense for the next three months, mild risk-on bias for 12 months (overweight equities and Japanese government bonds, underweight credit, neutral on bonds and commodities). Expected total returns for S&P 500/Stoxx 600/MSCI Asia ex-Japan/Topix over 12 months are 9.5%/12.1%/28.6%/16.5%, gold at $5,275/oz (+20.9%), WTI/Brent spot returns at -27.9%/-25.5%, and US high yield expensiveness at the 99th percentile of the past decade.

Financial conditions are tightening: Goldman Sachs' US nominal FCI tightened by 11.1bp in one week to 98.49 (driven by rising 10-year yields), Japan's FCI tightened significantly in 9, and global (ex-Russia) FCI tightened by 6.1bp. However, the US Q3 GDP forecast remains at 2.5%, current activity indicators in 8 are +3.5% compared to 7, and global activity in 8 is +3.4%, still significantly above potential levels. The MAP Economic Surprise Index fell to +0.49, indicating weakening degree of data beating expectations.

Goldman Sachs views Germany as a relatively positive case: fiscal stimulus shifting from commitment to implementation, improving manufacturing and factory orders, and a €500bn investment plan supporting the economy in 2027-28. Consensus expects German EPS growth in 2027 to be approximately 17%, higher than STOXX 600. Preference is given to fiscal, defense, CAPEX, and domestic recovery themes, while maintaining caution on autos and chemicals.

A BofA survey shows nearly 80% of respondents expect the Bank of Japan to hike rates this month. Policy normalization has surpassed earnings to become a key theme for Japanese stocks. The USD/JPY level likely to trigger intervention has fallen from 165 to 160. Japanese holdings are concentrated in banks and semiconductors, with bank allocations reaching historic highs.

After Saudi Arabia shut down the 7 mb/d East-West pipeline due to attacks, Brent rose to near $110. Goldman Sachs believes that attacks on oil infrastructure mark a significant escalation of the conflict, with the scale of impact and repair duration being highly uncertain (assessments range from very soon to eight weeks). The latest attacks may threaten the remaining 2 mb/d of Yanbu exports.

Net loss in Persian Gulf oil exports is 6.7 mb/d (recent exports of 16-17 mb/d are approximately 70% of pre-war levels). A production loss of 7.1 mb/d in the Gulf in 8 shifted the global market from near balance in 7 to a deficit of approximately 3 mb/d (own balance estimate is approximately 1 mb/d). OECD SPR drawdown has slowed to 0.3 mb/d, entirely contributed by the US. China's crude oil imports are down approximately 4 mb/d year-on-year but have room for recovery; the sustainability of core buffers is questionable.

If Gulf production in 2027 is 4 mb/d lower than pre-war levels, Brent will exceed $120. The implied probability that the 2 contract expiry price is above $100/bbl has risen to 44%. If an energy ceasefire is reached, Russian oil exports may recover (refinery shutdowns remain at 4.6 mb/d), constituting a major downside risk.

Bernstein estimates that the closure of the Strait of Hormuz and damage to Ras Laffan facilities disrupt approximately 63 MTPA of LNG supply. Combined with a delay of 5 MTPA in Qatar's North Field expansion, effective global supply in 2026 decreases by approximately 20 MTPA. This assumes zero production for nine months in 2026 for Qatar and the UAE, representing a 76% reduction from a base capacity of 83 MTPA.

Europe's storage fill rate at the end of 8 was only 65%, expected to be approximately 73% by the end of 9. Under a normal winter, it could drop to 23% by the end of 3 in 2027, or to 8% in a cold winter. TTF weather scenarios: warm winter approx. US$20, normal US$23-25, cold winter approx. US$30/MMBtu. The monthly LNG peak for the 2026/27 winter could reach US$30. JKM average price is projected at US$22.8 for 2026 and US$15.7 for 2027, with clear normalization only in 2028.

In the long term, existing and under-construction capacity of 674 MTPA in 2030 creates an oversupply of approximately 50 MTPA against the required 619 MTPA. US operating and under-construction capacity exceeds 200 MTPA. The market expects a shift to oversupply starting from 2028, with long-term prices converging towards the full marginal cost of US LNG at approximately US$9-10/MMBtu.

Rising energy prices exacerbate inflation concerns: Brent +11.7% last week drove the energy sector up by +1.7%. Goldman Sachs raised its 4Q26 TTF forecast from €53 to €70/MWh. Higher energy costs suppress global equities through expectations of higher policy rates and government bond yields. European stocks fell -1.7% last week.

Goldman Sachs cites NOAA predicting a potential Super El Niño (temperature anomaly exceeding 2°C) lasting from 6 of 2026 to approximately 5-6 of 2027. The agriculture team expects it to be the strongest on record. Events with higher intensity typically produce the most significant global impacts during Northern Hemisphere winters. Droughts, floods, and transport disruptions in agricultural regions may tighten supply and amplify price volatility.

China has high import dependency and concentrated supply: palm oil nearly 100%, soybeans 92%, barley over 80%, sugar 31%. Concentrated in Indonesia (75% of palm oil imports), Brazil (74% of soybeans, 87% of sugar), and Australia (57% of barley). Supply concentration limits flexibility for alternative procurement during weather or logistics disruptions.

Net profit sensitivity for 2027E under a +10% cost scenario: CR Beverage approx. -12%, Jonjee approx. -7%, Nongfu approx. -3%, Haitian approx. -2%. Beverages, instant noodles, and condiments have the largest exposure. Hedging and cost-locking arrangements expire around the end of 2026. Nongfu and Haitian are more resilient due to margins, operational efficiency, and product mix.

F&B side: Brent/WTI rebounded nearly 50% from their 7 lows to approx. $105/$100. Global sugar prices rose nearly 30% in the previous two months. Sugar and oil-related PET/PP are the most notable cost risks for Chinese F&B. Chinese pork prices are expected to start rebounding in 2H26, with 2027 YoY +20%. Beef has recovered since the start of the year due to reduced supply, but the magnitude is small.

However, most F&B companies' raw material cost indices have limited historical correlation with margins. Procurement price locking, inventory arrangements, menu adjustments, promotion intensity, and franchisee support are more decisive. Haidilao sources about half of its purchases as beef and mutton. In 2025, despite favorable raw material prices, gross margin contracted by 3 ppt, indicating that pricing strategy impact may exceed raw material costs.

Food segments: Imported nuts are most vulnerable (droughts in 2024-25 Africa increased cashew costs by +2%-12%, Qiaqia nut gross margin -4.4 to -7.5 ppt). Sunflower seed procurement costs were -18% YoY in early 9. Konjac processors buffered via efficiency. Fish surimi: Chinese fish prices were only +1% YoY. Most snack and frozen food companies have sugar/palm oil/soybean COGS exposure below 10%, lower than the sector, with Ligao Foods being an exception at 30% palm oil share.

BofA raised CY30 semiconductor TAM from $2.7tn to $3.2tn, and CY26-30 CAGR from 14% to 18%. Memory is the biggest driver (CY26 sales expected +327% YoY). CY26 WFE raised by 8% to $156bn, CY30 to approx. $360bn. Due to memory price inflation distortion, equipment demand is now measured by WFE corresponding to wafer starts per 12-inch wafer.

BofA sees no signs of slowing AI customer orders, long-term agreements, capacity commitments, or pricing. NVIDIA B200 rental price of $5.72/hour is less than 10% below the 3 peak. Sector 18x NTM P/E is below S&P 500's 19x, with expected EPS +139% YoY. However, maintaining tactical caution until midterm elections conclude and macro concerns ease.

Goldman Sachs confirms robust Chinese semiconductor demand: IC production in 7 +20.7% to 53bn units. Chinese semiconductor revenue +127.7% to $38.2bn. Total revenue of Taiwan-listed semiconductors in 8 +48.9%. IC import value in 8 +83.6% while volume only +6.7% (implied ASP +72.1%). Export value +129.8% while volume -7.9%. Value uplift rather than volume growth. However, inventory days in 7 rose to 66 days, higher than 57-58 days in the same period of 2023-25, which is a contradictory signal to track.

Deutsche Bank observed selective price hikes for mature chips (significant price hikes for TI/ADI analog products, MLCC entering shortage but not widespread tightness in 2021/22). STMicro core MCU lead times 40-50 weeks. Infineon constraints may persist until 2027. N2/N3 lead times 78-156 weeks. Foundry capacity insufficient at least until end of 2028. DRAM supply gap for 2026-2030 is 210k/507k/795k/575k/379k WSPM. HBM demand rises from 270k in 2026 to 1,208k WSPM in 2029. Non-hyperscaler DDR5 fulfillment rate below 30%.

Bernstein's ASML notes: AI demand spreading from TSMC/Hynix to Intel/Samsung/Rapidus. Enterprise AI is a key support for the infrastructure investment cycle. Revenue in 2026 already covered by orders. Most EUV capacity in 2027 pre-booked. Orders in 2028 include 50% prepayment. 110 Low-NA EUV machines are not the supply cap for 2028. System shipments in 2027 expected +30%. Effective wafer capacity +45%. Maintains Outperform €2,500 target price (80% upside).

Bernstein's BESI notes: Hybrid bonding is the long-term interconnect technology for logic and memory. Logic yield exceeds 99.9% while HBM is approx. 90% (gap due to complete process rather than bonder). Over 180 units shipped. Sole supplier at TSMC/Intel. First 50nm system sent to TSMC. Throughput to rise from 1,500 to 3,000 UPH. Large-scale HBM ramp-up in 2028-29. Maintains Outperform €320 target price (82% upside).

Bernstein's Arm notes: AGI CPU in final certification phase. Approx. 4,000 units undergoing final round of testing with Meta. CQ1 2027 first shipment approx. $100m revenue. Meta expected to contribute 90% of FY28 approx. $1bn opportunity. Demand exceeds $2bn but initial TSMC supply only approx. $1bn. 3nm essentially sold out. 2nm not widely allocated. Supply, not demand, determines near-term revenue cap. Agentic AI is a CPU-intensive workload. Arm can provide performance comparable to x86 at approx. half the power consumption. Outperform $480 target price (approx. 101% upside).

Nomura: Japan's rigid module substrate shipment value in 7 +68% YoY to record ¥31.3bn. Area +19%. Average price +41%. TSMC roadmap substrate size expands generationally from Blackwell 72×78mm to Feynman 120×128mm and subsequent at least 150×239mm. Shift to double/multi-layer core or glass core. Multi-layer core expected to be fully adopted in 2028 or later. Advanced packaging expanding to power delivery and heat dissipation functions drives drilling equipment demand.

Goldman Sachs: AccoTest's STS8600 has started small-batch shipments. Validated with local CPU/GPU customers with strong order backlog. 2Q26 revenue +39% beat expectations. 2026-28E revenue raised by 3%/7%/10%. Net profit raised by 15%/5%/3%. Operating margin lowered due to R&D investment. Maintains Neutral and Rmb630 target price (implying 81.8% upside). Upstream material supply tightness limits short-term shipments.

Goldman Sachs: Fujitsu began accepting orders for in-house developed 2nm FUJITSU-MONAKA CPU and MONAKA Server in 11 of 2026. CPU expected to ship in 1-3 of 2027. Start contributing to profits in FY3/28. Local deployment demand growing in sovereign AI scenarios. Already provided to 30 companies in Japan and Europe in advance. Maintains Buy ¥4,590 target price (14.5% upside).

Goldman Sachs lowered 2030 800VDC penetration in new US data centers from 25% to 21%, and in Europe from 20% to 17%. Because max power demand for AI inference racks is only approx. 150kW, traditional architectures remain viable. Experts believe 800VDC is unlikely to be applied in most inference scenarios. More likely upgrade direction is liquid cooling. NVIDIA Kyber NVL144 rack reportedly delayed from 2027 to 2028. Full 800VDC will not appear at least within the next five years.

The traditional architecture remains a larger and potentially higher-margin market: non-AI data centers account for two-thirds of the installed base by 2032 (approx. 17% CAGR, currently approx. 83%), while AI accounts for one-third (35% CAGR); Microsoft plans up to 38GW of installed capacity in 2032 (approx. 20% CAGR); Legrand, which has not announced SST development, is expected to benefit from sustained demand for traditional architectures.

Transition drives up equipment value: By 2030, the total power chain value for AI training is approx. $4mn/MW ($1.7-$1.8mn in 2025), and for AI inference approx. $2.6mn/MW ($1.4mn in 2025); in a favorable environment, the earnings CAGR of electrical equipment companies may remain at approx. 20-25%.

SST prices are approx. $400-$500k/MW with limited manufacturing scale; solid-state breakers require microsecond-level interruption and cost approx. $600k/MW (approx. five times existing levels). Hyperscaler SST RFPs are viewed as tests of price and scale sensitivity; after technology selection, evaluating orders still requires 12-18 months, with large-scale deployment no earlier than 2029; Delta is piloting 5MW SST in Asia, Eaton plans to provide 2MW samples by end of 2026, GE Vernova expects to launch SST in 2027, while Schneider/Vertiv/ABB offer transitional rectifiers and medium-voltage UPS.

Goldman Sachs forecasts the global >10MVA power transformer market will grow from $27bn in 2026E to $48bn in 2035E, driven by data center power demand, emerging market infrastructure, developed market replacement, and renewable energy integration; Chinese companies' overseas share rose from 5% in 2022 to 21% in 2026E and 25% in 2035E; Sieyuan's overseas share rose from 2% to 7%, with revenue increasing from $0.5bn to $3.2bn; Buy rating with a target price of Rmb204.5 (implying 4.0x P/S for overseas business).

Akamai has signed multi-year CIS commitments exceeding $2.8bn; contracted projects alone can accelerate growth from mid-single digits to high single digits next year, with project pipeline extending to 2028; large compute projects generate $0.50-$1.00 recurring revenue per $1 capex, operating margins of mid 20%-mid 30%, and payback periods of 2-3 years; power and hosting capacity, rather than demand, are the primary constraints; AI also expands the sales funnel for security businesses.

Deutsche Bank warns that power and industrial equipment bottlenecks may limit data center deployment: lead times for 15kV medium-voltage switchgear are 52-80 weeks, and for 5-50MVA transformers are 75-110 weeks; gas turbine backlog reached 116 GW in Q2 2026 while current annual production is approx. 20 GW, with data centers accounting for approx. 20% of the backlog.

Bernstein believes enterprise AI adoption is primarily constrained by data limitations, governance, risk, compliance, and internal implementation issues rather than frontier model progress; most enterprise applications use smaller specialized models; the SaaS sector, previously down due to AI displacement concerns, may benefit from slowing AI progress; Microsoft's training exposure is relatively small and it has invested in safety guardrail capabilities, while Oracle's non-cancellable contracts provide at least 5 years of protection, but its higher training exposure makes the stock sensitive to frontier AI news.

Deutsche Bank cites Bob Metcalfe's erroneous prediction of an internet crash in 1995 and his public retraction in 1997 to illustrate that even renowned experts struggle to predict technological development; within the year, AI-related search topics shifted from SaaSpocalypse to employment and then to extinction; the economic significance of the current AI boom depends on whether expectations and hype translate into tangible results, a judgment that is inherently uncertain.

JPMorgan forecasts hyperscaler capex to increase by +70% in 2027 to over $1.5T, cumulative free cash consumption of approx. $900B in 2026-2028, with 2027 being the peak loss year, and approx. $1.1T debt financing in 2027E-2029E; Google is expected to return to positive free cash flow in 2030, Amazon in 2031; however, AI returns are materializing: AWS annualized $169B, Google Cloud $100B, with backlogs of $520B/$497B/$47B for Google/Amazon/SpaceX, and compute demand still exceeds supply.

JPMorgan: AI-driven traffic YoY +62% in 7, conversion up 60%, revenue per visit up 53%, but only accounts for low single-digit share of third-party platform traffic; agents may disintermediate in the long term but still rely on existing platforms in the short term; Alphabet Overweight $420 (2027 revenue $612B, EPS $13.67), Shopify $185, Take-Two $310 (FY27 guidance implies approx. 34M GTA VI sales, conservative), Spotify $650.

Bernstein: AWS SSO network metrics show non-AI weakness persisting from 5 to early 9; Q3 non-AI revenue YoY accelerated by approx. 50bp, but Q4 could deteriorate by up to 100bp compared to the existing approx. 150bp resistance assumption; Twilio and Cloudflare face more pronounced resistance due to limited AI exposure; however, OpenAI revenue QoQ +18% in Q2, Anthropic ARR exceeded $65B in 7, accelerating AI-native token consumption can partially offset traditional workload weakness.

Bernstein's Perplexity notes: Positioned as an AI orchestration platform coordinating 24 frontier and open-source models, Perplexity Computer supports 600+ data connectors and answers over 15 hundred million questions monthly; auditability and enterprise controls are core to financial applications, automation and dynamic arguments have the greatest impact on investment teams, and Brain memory layer and Skills templates form reusable organizational assets.

Goldman Sachs: Meitu's MVLAND monthly ARPU reached Rmb220, high-ARPU overseas paying users drove subscription rate from 5.5% in 1H25 to 6.5%, photo/video design revenue +31%, expected 2H26 revenue +24%; however, net profit for 2026-28E cut by 7%/6%/7% reflecting lower traditional product revenue and weakening gross margins; maintain Buy, DCF target price lowered from HK$12.3 to HK$10.8 (implying 155.9% upside).

Goldman Sachs forecasts Chinese companies' overseas share in 11 case industries will rise from average 18% in 2026E to 31% in 2035E, with revenue growth of 3.6x, but annual average share gain will slow from historical 2.5ppt to 1.4ppt, as expansion targets core segments of existing enterprises, and brand, service network, and regulatory verification barriers make progress slower and harder.

Chinese companies' 2026E share: Emerging markets 24%, EU 16%, US 10; over 70% of industries started from emerging markets, over 80% entered from low-end segments; late-mover industries like autonomous taxis and surgical robots have the largest incremental opportunities, with combined revenue reaching 4.6x current levels.

Chinese companies entered with an average 30% price discount, but in 11 industries, prices rose in 7 after entry and fell in only 2, challenging the view that overseas expansion inevitably leads to deflation; price cuts are more closely associated with revenue contraction or share pressure rather than Chinese entry itself; robot vacuums and European cars are the most urgent deflation risk industries.

Capital markets punish existing enterprise revenue pressure faster than rewarding Chinese companies' overseas growth; selected Chinese companies' average implied P/Sales(2035E) is only 1.6x, white goods, European cars, autonomous taxis, and European e-commerce are below 1.0x, valuations have not fully reflected overseas growth opportunities.

Durable consumer goods: Non-China HVAC TAM grows from $149bn in 2026E to $228bn in 2035E, white goods from $137bn to $208bn (approx. 5% CAGR), Chinese share rises from 12% to 20% and 23% to 32% respectively; robot vacuum TAM grows from $6.4bn to $13bn (approx. 8% CAGR), developed market penetration is only approx. 20%, Chinese share rises from 63% to 82%, Roborock's share rises from 24% to 33% vs Ecovacs' 16% to 23%; Midea/Haier/Roborock Buy (Rmb103/HK$119, Rmb29/HK$31, Rmb190), Gree Neutral (Rmb41), Ecovacs Sell (Rmb56).

Automotive: European passenger car market sees only mild growth (16.4mn units in 2026E to 17.5mn units in 2035E), Chinese automakers' share rises from 17% in 2026 to 23% in 2035E, BYD and Leapmotor reach 8% and 3% respectively, with European revenue of $34bn and $14bn; BYD's European business is already priced at 1.7x P/S (overseas per-unit profit over Rmb20,000, approx. 21% of 2035E earnings) while Leapmotor is only 0.2x, not fully reflected; both are Buy (BYD A-share Rmb137/H-share HK$134, Leapmotor HK$55).

Grid equipment: Chinese companies' overseas transformer share rises from 5% in 2022 to 21% in 2026E and 25% in 2035E, Sieyuan's overseas revenue increases from $0.5bn to $3.2bn, rising overseas share is a key component of its valuation opportunity.

Goldman Sachs: Key NEV OEMs' combined orders in Week 37 of 2026 MoM +32%, YoY +1%, driven by new model launches and promotions; Xiaomi/Tesla/Geely orders MoM +724%/+67%/+38% (Skynomad N70/N90 launch, Model 3/Y incentives, Galaxy TT EV); NEV retail/wholesale penetration on 1-6 of 9 was 71.5%/75.1%, higher than 65.3%/63.4% in 8, while overall passenger car retail YoY -19% highlights NEV relative strength.

Price competition continues: As of 12 of 9, average NEV dealer discount 7.80% expanded MoM (lower than 8.49% a year ago), BYD 4.06%, ICE 19.98% both expanded MoM; battery-grade lithium carbonate fell to RMB143.5k/ton MoM -5.0% while cell prices remained flat, constituting a backdrop of upstream cost easing.

China's smart connected NEV 15th Five-Year Plan proposes NEVs to account for 70% of passenger car sales in 2030, promoting supply-side discipline, capacity control, exit of inefficient participants, and industry consolidation; slowing discount spiral provides tactical relief for German premium brands unwilling to chase volume at current prices, but consolidation may strengthen better-capitalized, vertically integrated local leaders, structural challenges remain unresolved.

European premium OEMs weak in China: Implied China retail revenue YoY for Mercedes/BMW/Porsche in 8 was -41.7%/-31.7%/-41.2%, with both volume and price falling; however, Mercedes GLC monthly sales approx. 1,500 units, localized GLE ramping up, BMW iX3 pre-orders positive (including Li Auto owner demand), order conversion depends on test drives and L2++ system credibility; Goldman Sachs maintains Buy on all three (€67/€82/€53).

Morgan Stanley: US premium truck and SUV demand not clearly weakening; GM expects earnings growth in 2027 (truck cycle, lower EV losses, warranty improvement, digital revenue, and defense), software and services with $3bn recognized revenue + approx. $7.5bn deferred revenue by end of 2026 is the biggest valuation release opportunity, Overweight $101; Rivian R2 demand not primary concern (Launch Edition conversion exceeded expectations) but Tier 2/3 supplier readiness is ramp constraint, Underweight $14; Scout differentiates with EREV (87% pre-orders are EREV) delivering in 2028.

Goldman Sachs maintains Buy on BYD (A-share Rmb137/H-share HK$134) and Leapmotor (HK$55), noting risks including intensifying EV competition, slower-than-expected overseas expansion, external battery supply below expectations, Stellantis partnership uncertainty, and overseas policy risks such as tariffs.

HSBC believes Biopharma sector valuations have risen to decade highs (excluding Lilly and Novo); returns over the next 12 months should rely more on bottom-up stock selection rather than broad industry re-rating. The firm lowered the additional risk premium for the sector from 75bp to 25bp and raised target prices broadly, but emphasized that low valuations do not necessarily mean cheapness; the key lies in whether companies can improve their medium-term growth trajectory.

HSBC points out that peak sales of drugs facing patent cliffs in 2025-29 exceed $200bn in aggregate, and the LOE scale in 2030-34e is even larger. If replacement revenues lack franchise fit, consensus may underestimate downside risks. New products in adjacent therapeutic areas can leverage existing commercial infrastructure, physician networks, and payer access, making limited incremental sales generate disproportionate earnings and stock price sensitivity.

HSBC's analysis of over 250 M&A and licensing deals since 2019 found limited evidence of consistently beating expected clinical success rates. M&A is not a panacea for patent cliffs; the firm prefers assets that can be integrated into existing therapeutic moats. It expects recent deal flow to remain primarily from China, with regional manufacturing deals potentially increasing. It also warns against hot-hand bias, noting that clinical success rates should first be evaluated by therapeutic area and trial design.

HSBC individual stocks: Maintains Buy on AbbVie/J&J/Merck/Bayer/Sanofi ($315/$320/$172/€65/€100), citing manageable patent cliffs, accelerating growth, pipeline refreshment, litigation resolution potential, and undervalued transformation options, respectively. Maintains Reduce on Eli Lilly ($850 raised to $940, oral obesity drug assumptions too optimistic, intensifying competition). Downgrades Amgen from Buy to Hold ($425, key pipeline drivers only materialize in late 2027). Upgrades Novartis from Reduce to Hold (CHF110).

Deutsche Bank believes the greater structural opportunity in AI pharma lies in R&D (target selection, molecular design, trial design, patient identification). AI projects show significantly higher Phase I success rates than non-AI projects, and in 2023-2025, this no longer comes at the cost of poor Phase II results. However, there are too few cases entering late-stage conversion to assess Phase III success rates. The termination of two AI-designed drugs by Exscientia and BenevolentAI in Phase II suggests that 80-90% Phase I success rates do not eliminate late-stage failure risks.

European Big Pharma AI deployment: GSK becomes the first company to officially announce partial AI-driven cost cuts. Roche acquires PathAI with an upfront payment of $750m and up to $300m in milestones. Sanofi improves target identification efficiency by 20-30% and halves mRNA design time. Isomorphic Labs completes $2.1bn Series B and launches IsoDDE (claims accuracy more than twice AlphaFold3 but not formally peer-reviewed). Open-source Boltz-2 correlation coefficient is 0.65 with 1,000x improvement in cost and speed.

Nomura WCLC 2026 data: DualityBio/BioNTech's DB-1311+BNT327 showed ORR of 70.4% and DCR of 93.0% in 71 SCLC patients (first-line ORR 92.3%). Leads Bio's LBL-024 first-line NSCLC ORR increased from 65.0% to 71.0%, squamous subgroup 87.1%, 6-month PFS rate 86.7%. RemeGen's RC148 squamous ORR was 90.0%, with high response rates even in PD-L1 TPS<1% patients. Maintains Neutral on RemeGen (A-share CNY119.71/H-share HKD89.44).

Morgan Stanley Healthcare Conference: Pharvaris' oral deucrictibant shows potential for efficacy close to injectables; 1Q27 CREAATE top-line and NDA combined. Monopar will complete rolling NDA for ALXN1840 and receive Rare Pediatric Disease designation. Atea's C-BEYOND met SVR12 primary endpoint (8-week regimen cure rate 94-95% non-inferior to 12-week Epclusa), NDA target mid-2027. Adagene expects to announce randomized Phase 2 of ADG126 combined with pembrolizumab in 1H27 (small sample response rate 15-36% vs historical PD-1 monotherapy approx. 0%).

Goldman Sachs is bullish on large banks: Slowing balance sheet expansion helps preserve capital and limits RWA growth. Large banks have completed recapitalization, enhancing capital positions and supporting dividend sustainability. Additional capital will not translate into significantly faster loan growth in the current demand environment. Maintains Buy on China Construction Bank and Bank of China.

Deutsche Bank reiterates Buy on Robinhood, raising target price from $136 to $138 (based on 2027 EPS and forward P/E of 90% premium to S&P 500). Designated as top online broker growth pick (implying approx. 23% upside). Mixed signals in 8 metrics: Crypto trading volume rose MoM (App $7.4bn, Bitstamp $10.1bn), options and event contracts run-rate above expectations, but net inflows of $4.0bn, margin clients of 28.6mn, and net securities lending income of -$8mn missed expectations. 3Q26 EPS raised from $0.57 to $0.58, and 2027 from $3.48 to $3.53.

Robinhood Chain and Arbitrum revenue sharing agreement increases sequencer revenue retention to 70%-85%. By 4Q, it could become a $100mn annualized business, giving the company approx. 14 business lines with annual revenue over $100mn by end of 2026. Combined with prediction markets (pending SEC review), international expansion (UK, Canada, Singapore, Indonesia), and AI agentic trading, these constitute diversified long-term revenue sources.

Goldman Sachs: SCHW 8 month NNA of $64.7bn equates to approx. 6.0% annualized organic growth, exceeding forecast by 141bp. Client cash stable excluding long-short flows, margin balances +4.5% MoM and +92% YoY. Reiterates Buy on $128 (approx. 18x Q5-Q8 EPS). However, tightening of long-short account requirements starting 9 16 (minimum deposit raised from $1m to $10m) will slow NNA growth, with full impact reflected in 10, posing a temporary headwind to AQR wealth channel flows.

Goldman Sachs: Consumer and corporate spending stable through 8. Visa US payment volume +10% last quarter, cross-border +14%; Mastercard 8 month metrics strong. Agentic platforms like Meta Muse center on bank cards, allowing Visa/Mastercard to benefit directly via fraud prevention and tokenization, temporarily alleviating concerns about card payment disruption. Prefers card networks (Visa $438, Mastercard $701), BNPL (Affirm, Klarna), and consumer fintech (Block added to Conviction List, Chime $37, Toast Buy). Maintains Sell on PayPal (brand checkout growth may slow to 1-2%), Shift4 Neutral with target price lowered to $47.

JPMorgan forecasts global investment banking 3Q26 market business revenue +10% YoY (US +18%, Europe +2%). Equities business +21%, FY26 equity sales & trading revenue +29% hitting record high (driven by prime brokerage), FICC flat YoY. Management guidance is conservative with upside potential. UBS and Deutsche Bank added to European preferred list (2028E P/E 10.7x/7.6x vs US 14.5x gap expected to narrow). However, 2H26 market business revenue expected -23% QoQ.

Goldman Sachs Walmart notes: Overall consumer resilience, solid back-to-school demand, but low-income households under pressure from SNAP cuts, high oil prices, and beef inflation; behavior starts changing when national average gas price exceeds $4. In 2Q, qualified for $2.9bn tariff refund and received most of it. Launched over 11,000 price reductions in US business (1Q end 7,200 items). Unit sales inflection point driving share gains; member spending is 4x non-members. Maintains Buy on $132 (21% upside).

Goldman Sachs Global Consumer Conference: Consumers resilient but more cautious and selective. US environment more constructive than EMEA and LatAm. Purchase decisions revolve around specific goods and value orientation. Pricing divergence—some companies reinvest tariff refunds and lower interest rates into prices to strengthen value gaps, while others offset inflation through price hikes and surcharges. Fuel, freight, crude oil, metals, and gold constitute key emerging cost headwinds.

Companies emphasize diversified growth engines: Aramark expects data center-related Nexus business to grow to approx. $2-3bn in coming years. Chewy Autoship accounts for over 84.6% of sales. Trex reaffirms 2030 $2bn revenue target. Capital allocation increasingly focused on growth reinvestment and share buybacks, with higher M&A hurdles.

Goldman Sachs Nielsen data: US beauty & personal care sales +2.6% in four weeks ending 9 5 (volume -2.8%, price/mix +5.6%). Skincare +5.3% accelerating while health category -2.7%. L'Oréal +3.7% driven by volume +4.0% (L'Oréal Paris +8.0%, La Roche-Posay +32.2%) outperforming market. Beiersdorf -6.3% (Coppertone -20.6% vs sunscreen market +9.8%). Henkel laundry -2.4% continuing to lose share (Persil -5.2%, Purex -8.4%) confirming Sell logic. Nielsen covers only approx. 40% of market, potentially underestimating actual growth.

Bernstein believes Louis Vuitton's core constraint is affordability rather than appeal: Full-size handbag entry price €1,550, ASP approx. €2.2k exceeds most aspirational consumers with annual spending below €1.5k. Only 6% of SKUs online are below €1.5k (Gucci 27%, Burberry over half). Due to paused China recovery and Molly Tea IP dispute triggering consumer backlash, lowered F&LG 3Q26E organic growth from +1% to -4%, FY26E to -0.6%. EPS cut by 2.5%/3.5% for two years. Target price lowered from €570 to €520 but maintains Outperform.

Bernstein: WTI closed at $100.26 (week +9.6%), Brent $104.69 (+8.7). Refinery utilization 97.8%, distillate inventories below same period last year by 11.9%. Diesel and jet fuel crack spreads expected to remain elevated; relief relies on imports or demand destruction. Distillate crack spread widened week-on-week by $1.70 to $100.91 while gasoline fell by $12.31 to $31.22. Should not use blended 3:2:1 crack spread to adjust transportation models; should re-evaluate Q3 fuel based on current forward curve.

Truck spot rates rising across all equipment types, but fuel accounts for over half of box truck increase and entire flatbed increase; not viewed as evidence of improved pricing power. Freight rate recovery driven by supply factors like capacity contraction rather than true demand. Rail traffic +13.8% weekly mainly reflects low base (YTD +3.6%). Models should maintain constrained tonnage and load factors.

8 PPI +5.4% YoY above 5.3% consensus. 2-year US Treasury 4.56%, 10-year 4.96%. Transportation companies face dual pressure of fuel costs on earnings and higher discount rates on valuations. Rail stocks underperforming for third consecutive week attributed to interest rate duration sensitivity rather than deteriorating weekly operational data. Interest rates, not merger news, are the more critical variable for rail stock entry timing.

UP-NS Merger: 3 Class I railroads and 11 shortline railroads, among 19 respondents, raised 21 demands, many of which are mutually competitive, reiterate remedies previously rejected by the STB, or shift profits between operators. Bernstein believes no demand is high enough to jeopardize approval; the primary risk is the regulatory timeline (including contingent matters such as Norfolk Southern's option on CPKC's Dallas terminal), with the complete application deadline on 2026/11/18.

During the week when crude oil rose 9.6%, the four major US airlines fell an average of only about 1%. This resilience stems from prior valuation de-rating and positioning rather than fundamental improvement; the Delta-Hyatt reciprocal membership partnership is expected to have minimal impact on short-term RASM but can reinforce its premium revenue positioning at very low incremental capital cost, with membership advantages more pronounced relative to American and United.

The Alibaba-DHL AI logistics memorandum of understanding targets cross-border SME trade (the high-yield international business served by UPS and FedEx). It is not viewed as a near-term earnings event but creates structural pressure on international export yields. In upcoming results, yield rather than volume will be the more important competitive metric.

Deutsche Bank forecasts Asian GDP growth of 5.5% in 2026 (higher than 5.4% in 2025), driven by the AI and semiconductor cycle, strong exports, and equipment investment. It raised India's FY27 forecast to 7.3%, Malaysia to 5.4%, Taiwan to 11.5%, and Singapore to 5.2%; however, domestic tech gains are uneven: Malaysia has translated this into strong investment, ASEAN faces a large trade deficit with China, and Vietnam and Thailand are constrained by foreign investment dependence and high import content. Vietnam's Q3 growth of nearly 9% still faces inflation, dong pressure, and bank financing constraints.

Regional tightening cycle expands: Bank Indonesia and BSP have reversed part of their easing. RBI is expected to begin hiking in 12 2026 after liquidity normalization, reaching a terminal rate of 6.25%. BNM and Bank of Thailand are expected to hike in 2027. Bank of Korea and Taiwan CBC will tighten to levels unseen since 2008. Continued fiscal support may delay disinflation and keep yields elevated, while higher US rates exert pressure. Sri Lanka is expected to hike by 50bp to 9.25% by the end of 9.

Nomura expects AI-driven semiconductor and electronics demand to support outperformance in Taiwan (2026 GDP +12.5%), South Korea (8 month semiconductor exports +209.0%), Malaysia, and Singapore. China's weak domestic demand (Q3 GDP +4.3%) and K-shaped economy limit stimulus effects. India's FY27 GDP forecast was raised to 7.0%. A reversal of the AI boom or memory chip shortages triggering commodity inflation are key risks.

BofA 9 Survey: Respondents expect Asia (ex-Japan) equities to rise +6.3% over the next 12 months (historical 89 percentile) and Japan +6.4% (94 percentile). Net 55% expect improved APAC corporate earnings (higher than 45% in 8). However, concerns about Fed re-hiking have weakened growth expectations, and China's economic outlook deteriorated significantly in 9.

BofA Survey: Four-fifths of investors stated that proven substantial AI monetization best boosts confidence in adding positions. 55% believe the positive impact of AI is already sufficiently priced in (37% in 8). Hedging against AI downside involves rotating into defensive sectors; software and platforms are seen as having the best AI risk-reward. Expectations for a stronger semiconductor cycle rebounded to 35% (still below 60% in 7). Taiwan and the US are tied as the main beneficiary markets for the next phase of the AI cycle.

Nomura Fund Flows: From 7 to 11 9, foreign funds flowed into US equity thematic funds by $2.4bn ($4.9bn month-to-date, $10.7bn in 8); bond fund outflows narrowed to $59mn. Emerging market ETF inflows rose from $1.2bn to $1.5bn, concentrated in equity funds (+$1.7bn). However, Korean retail investors net sold $490mn of US assets (equities -$667mn), and inflows into US dollar bond ETFs by Taiwanese investors were near zero, indicating declining willingness for regional retail cross-border allocation.

Deutsche Bank: Despite significant repricing of rates and energy, gold remains about 10% above its 6 low, supported by official sector and institutional allocation demand. Retail and ETF flows are the main channel for Fed expectations to transmit to precious metals. The persistent decoupling of gold from real rates has reduced macro fund participation. A break below $4,220/oz could trigger CTA selling of approximately 14% of maximum algorithmic positions.

Copper: The pause in US stockpiling momentum caused prices to fall, but unresolved uncertainties may recur in coming months. The market appears crowded, but directional holdings by autonomous funds are limited; unless macro headwinds intensify significantly, the risk of a sharp decline is low. If LME three-month copper breaks below $13,850/t, CTAs will stop rebuilding long positions.

Palladium: Spot inflows reached 26% of historical highs, offsetting approximately 8% of CTA selling. However, another round of approximately 9% CTA selling is expected at current levels. Price support depends on whether spot inflows can remain high; the flow balance is fragile.

Deutsche Bank FX Framework: BIS estimates daily FX turnover in 2025 at $9.5tn (swaps account for 42%, the largest instrument). In the base case, approximately 40% of adjusted turnover is profit-seeking trading. Electronic trading and internal matching improve transparency, but market fragmentation creates an illusion of liquidity. Effective spreads, slippage, and price impact better measure true liquidity than displayed depth.

FX determination is organized by horizon: short-term order flow and positioning sentiment, medium-term interest rate differentials and external balance capital flows, long-term PPP (deviation half-life 3-5 years). UIP has been rejected in the long run; arbitrage returns may compensate for negative skewness and crash risk. Backtests of carry, momentum, and value factors show DBCR annualized excess returns of approximately 4% and Sharpe ratio of 0.77. 88% of FX trades involve the USD, and 99% of circulating stablecoins are pegged to the USD; stablecoins may expand USD demand.

Bernstein: Global TV production market size is $251bn, growing about 2% in 2025. Streaming and ad-based video-on-demand are the main future growth drivers. It is estimated that streaming will account for 40% of global content spending in 2026. The industry is highly fragmented; execution capability can directly translate into market share, and scale can diversify portfolio risk across different genres.

Financing models determine risk-reward: European producers mostly use pre-financing and cost-plus (stable profitability but limited upside), while US studios use deficit financing (higher IP upside but greater recovery and profitability risks). Rights retention is a core strategic variable; content budget constraints have improved producers' negotiating position relative to streamers, and relinquishing rights would weaken long-term library and secondary revenue potential.

AI is an incremental margin and library monetization tool, not a replacement. Post-production accounts for 10-20% of production budgets, representing the main opportunity for automation efficiency. Companies with large content archives can benefit from AI-driven tagging, search, and distribution efficiency.

Ratings: Disney and Netflix Outperform, Paramount Skydance Underperform, ITV, Banijay, RTL, TF1, M6, ProSieben, and Warner Bros. Discovery Market-Perform. Paramount Skydance plans to acquire WBD for approximately $110bn; post-merger 2026 revenue is expected to exceed $60bn, but execution and integration risks exist.

Deutsche Bank maintains Buy rating on SpaceX with a target price of $235.00 (price on 11 9 2026 was $151.21). The significant gap between the target price and current price indicates that Starlink user growth is not yet fully reflected in valuation.

As of the end of 8 2026, Starlink's global consumer subscribers are estimated at 13.2mn, with net additions of 601k in the month. North America had 5.4mn (over 40% of global, net additions 258k), the largest regional market, and the US had 4.0mn (net additions 226k), the largest single market. Service coverage expanded to 200 countries in 8 (Vietnam newly added). Smaller markets like the UK saw MoM growth of +13%, and Tajikistan saw MoM growth of +67% due to a low base.

Morgan Stanley: With high-end model production up 18% YoY, iPhone 18 Pro/Pro Max delivery times in non-China markets are flat or slightly longer YoY, indicating early demand is healthier than market fears. China market delivery times shortened YoY (Pro Max 27.0 days vs 30.5 days, Pro 21.0 days vs 30.5 days). International markets: Pro Max 25.0 days, Pro 18.1 days. Regional differences reflect different supply and demand structures.

iPhone 18 full-line prices increased by $100 YoY, with larger increases in international markets. The 5-week delay in iPhone Duo launch may cause high-end consumers to wait before purchasing, suppressing delivery time signals. However, the installed base expanding by 7% YoY is concentrated in the two launch models, carrier subsidies increased by $100 YoY, and extended leasing options help mitigate the impact of price hikes.

Historically, early delivery times have no correlation with iPhone cycle strength, iPhone revenue, or stock performance beyond 3 months post-launch. Verification requires channel inventory, third-party sales, and production adjustments. Maintain Overweight rating; $360 target price based on 9.4x CY27 EV/Sales (implying ~35x P/E on CY27 EPS of $10.30).

Bernstein: Following a €460m non-compliance fine, Google fully launched the EEA hotel search DMA overhaul in 9. This benefits OTAs in the short term (higher search exposure, free participation in aggregator modules, more traffic directed to OTA ecosystems, potentially lower bidding acquisition and discount spending). However, aggregator modules display only three hotels and cannot filter by price or rating, and supplier modules provide less information, representing a significant drop in service quality.

Revenue distribution is uncertain: New ranking is said to be based on relevance and utility rather than bidding economics, potentially weakening Booking's scale and bidding advantage. In ten-city aggregator rankings, Booking.com ranked first four times, Google three times, and Expedia and Hotels.com once each. Smaller eligible OTAs may gain disproportionate benefits. Airbnb benefits marginally as Google Vacation Rentals was removed in Europe; previous DMA adjustments are estimated to have reduced direct hotel booking traffic by about 30%.

Instinct completed a Manhattan hotel search (recommending Arlo Midtown among 6 options) and proceeded to Booking.com checkout (all-inclusive price $2,138.02). Agentic hotel booking is feasible, but 5-10 minute response times and text-based output still lag behind OTA planning experiences. Agent monetization is a core strategic risk; user-paid models prioritize traveler outcomes, intensifying direct booking and price competition. Meta Muse, xAI Grokbot, and Claude Agent show AI shifting from search recommendations to transaction and workflow layers.

Ratings: Airbnb and TripAdvisor Outperform, Booking and Expedia Market-Perform. Expedia target price is $310, with DCF valuation incorporating AI disruption risk scenarios. Rating differences reflect relative risk-reward of each intermediary under AI disruption and DMA overhaul.