〔Day Digest〕Retail ETF Inflows Cushion AI Trades as Hot CPI Lifts 25bp December Hike Odds to 90% and 387 GW of Data-Center Capacity Looks Non-Credible

0. Daily Arc

• **[NEW] Near-term cushion:** JPMorgan strategists including Nikolaos Panigirtzoglou said global equity ETFs took in about $20 billion a day last week and $30 billion on Monday, while leveraged equity ETFs attracted nearly $5 billion on Nov. 5-6, making retail dip-buying critical to stabilizing the recently battered AI trade ^[1][2].
• **[ESCALATED] Rate shock:** October headline CPI rose 0.4% m/m versus 0.3% expected and core CPI rose 0.3% versus 0.2%, lifting implied odds of a 25bp December Fed hike from about 84% to 90% ^[3].
• **[NEW] Demand test:** Bernstein's screen of 8,858 North American data-center projects validated only 199GW of 586GW announced, under-construction or land-bank capacity, leaving 387GW—about two-thirds—as "Bragawatts" ^[4].
• **[ONGOING] Sourcing caveat:** The JPMorgan flow figures and Bernstein project screen reach the packet through secondary media summaries, while the Nvidia actions arrive through a compilation of bank views, so these research claims are not independently verified within the packet ^[1][2][5][4].
• **Arc:** Leveraged retail flows support AI-linked equities in the near term, yet hotter inflation and a large low-credibility capacity pipeline challenge the trade's price and demand assumptions ^[1][4][3].

1. Flow Support and Equity Positioning

• **[NEW] Flow mix:** Inverse equity ETFs saw roughly matching weekly outflows, and JPMorgan said receding call-option buying could further amplify the influence of retail fund flows; the two ETF items repeat the same research rather than providing independent confirmation ^[1][2].
• **[ONGOING] Passive backdrop:** The China Securities Index Company's first domestic whitepaper on indexes and indexed investing put global ETF assets at $19.8 trillion at end-2025 and annual net inflows at $1.87 trillion, while identifying AI-assisted index compilation as a future direction ^[6].
• **[NEW] Single-name targets:** Rosenblatt cut D-Wave Quantum's target to $30 from $43, Guggenheim cut Reddit's to $140 from $170, and B. Riley raised Firefly Aerospace's to $54 from $40; all three retained Buy ratings ^[7][8][9].
• **[NEW] Price action:** On the reported Nov. 11 close, Nasdaq rose 1.6%, S&P 500 1.31%, and Dow 0.49%; in China, Shanghai Composite closed at 3,680.64, up 0.46%, Shenzhen Component up 1.32%, and ChiNext up 3.11%, with turnover at RMB1.69 trillion, RMB171.9 billion above the prior session ^[10]. CPO and PCB led the Chinese computing-hardware move ^[10].

2. Inflation, Fed Repricing and Policy Divergence

• **[NEW] Inflation detail:** Headline CPI m/m slowed from 0.6% but exceeded the 0.3% forecast; core was 0.3% versus 0.2% expected, energy eased to 0.7%, gasoline fell from 9.1% to 1.0%, food rose from flat to 0.5%, and core services rose 0.3% ^[3].
• **[ESCALATED] Rates:** The 2-year Treasury yield rose 18bp on Tuesday after the release, the worst Treasury selloff since April 2025, while markets priced at least another 50bp of Fed hikes over the coming months after September's increase to 3.75%-4.00% ^[11].
• **[NEW] Size of next move:** Huaxin put the implied probability of a 25bp December hike at 90%, while Standard Chartered economist Steve Englander said the case for 50bp was stronger than expected but not settled; Fed Governor Christopher Waller had advocated 25bp, with more if price pressure accelerated ^[3][11].
• **[ONGOING] Snapshot mismatch:** CMB International's forecast note called for 0.3% headline and 0.2% core CPI and put December hike odds near 80%, whereas Huaxin's post-release note recorded 0.4% and 0.3% and lifted 25bp odds to 90%; these are separate research snapshots rather than one synchronized market quote ^[3][12].
• **[ESCALATED] Election-policy map:** CITIC Securities expects AI and data-center rules to tilt toward project relocation and developer cost-sharing rather than radical tightening, while congressional oversight of China technology policy may gain marginally even as executive sanctions remain the main instrument ^[13]. CMB International expects tighter approval and tax-benefit scrutiny to lengthen AI construction cycles and raise supporting inputs ^[12].
• **[ONGOING] Operating divergence:** Pengyuan International says the Fed uses a rate corridor and balance-sheet tools in a market-based system, while the People's Bank of China combines policy rates with reserve-requirement cuts, MLF, relending and regular government-bond trading in a bank-centered system; both face a constrained transmission "last mile" ^[14].

3. China Policy, Capital Spending and Solar

• **[ESCALATED] Stabilization stance:** CMB International says Chinese policymakers have sent strong market-stabilization signals, with the central bank intervening and stocks becoming the first policy priority; exports still support growth, while consumption, credit and property remain weak ^[12].
• **[NEW] Liquidity tools:** After the People's Bank of China, National Financial Regulatory Administration and China Securities Regulatory Commission optimized the swap facility and share-buyback and shareholding-increase relending on Nov. 2, two swap-facility batches had involved 28 securities and fund companies for RMB105 billion by end-October, while 31 firms applied for more than RMB100 billion in a third batch ^[15].
• **[NEW] Mechanism:** The swap facility lets institutions exchange assets for high-liquidity assets without selling holdings, while buyback relending is intended to ease cash-flow constraints and add active demand; the latter is moving from a temporary arrangement to a permanent policy ^[15].
• **[ONGOING] Global allocation:** Allianz Investment Asia-Pacific head of equity products William Russell said China assets have become more strategic globally and the question is how to allocate them, while Bank Negara Malaysia Beijing chief representative Mohd Faizal bin Md Fathil said diversification requires a large allocation to assets of the world's second-largest economy ^[16][17].
• **[NEW] Domestic compute:** Tencent and Alibaba's combined 2026 Q2 capex exceeded RMB120 billion and doubled quarter over quarter; Tencent expects the increase to broaden as domestic ASIC chips arrive monthly, while China Mobile said 2027 compute investment would exceed RMB100 billion ^[18]. CITIC Securities flagged IDC, connectors and switches as potentially underestimated domestic-compute links ^[18].
• **[NEW] Industrial split:** Solar remains oversupplied, with polysilicon inventory above 570,000 tonnes, November output expected near 95,000 tonnes and wafer inventory around 27GW amid weak cell demand, while CITIC Construction Investment sees strategic-metal prices supported by defense stockpiling, AI-compute expansion and supply-chain sovereignty ^[19][20].

4. AI Demand Credibility and Falsifiers

• **[NEW] Low-credibility pipeline:** Bernstein's Oct. 11 model update found the credible project share at 33% initially, 35% in August and 34% in October even as announced capacity rose from 411GW to 586GW ^[4]. At $7-$15 billion of construction cost per GW, Bernstein's arithmetic put the 387GW low-credibility pipeline at up to $3-$6 trillion ^[4]. The team argued that state-level project pauses were not the main risk to focus on ^[4].
• **[ESCALATED] Nvidia governance risk:** A secondary roundup said Bernstein cut Nvidia from Outperform to Market Perform and its target from $400 to $200, Baird cut to Neutral with a $200 target, and Morgan Stanley removed Nvidia from its top semiconductor picks ^[5].
• **[ESCALATED] Caution conditions:** Baird analyst Tristan Gerra cited possible congressional hearings, government action, investigations, fines and unresolved CEO-tenure risk, while Morgan Stanley analyst Joseph Moore said Jensen Huang's leadership was critical to Nvidia's strategy ^[5].
• **Source-quality flag:** The Nvidia ratings and targets appear in one secondary compilation rather than the original bank notes, so they warrant verification against primary research ^[5].
• **[NEW] Historical warning:** Goldman Sachs strategist Bobby Molavi compared Nvidia's rise from $1 trillion to $5.8 trillion in market value and fall to an 18-month low with the 1840s British railway mania, arguing that technology can persist while value shifts away from builders toward the application layer and the broader economy ^[21].
• **[NEW] Macro check:** The People's Bank of China said AI investment's demand, growth and inflation effects are already visible, while its productivity, cost-reduction and disinflation effects still require observation ^[22].
• **Falsifier:** The model test is whether credible project share rises above 34% and the 387GW gap begins to be removed from forward demand forecasts, while the flow test is whether the ETF cushion holds without renewed call-option participation ^[1][4].