〔Day Digest〕March 2027 25bp Hike Call, $63 Brent and a 5GW AI Delay: Liquidity Tightening Meets Oil Oversupply and a 2028 Supply Gap

0. Session Arc ^[1][2][3][4]

• **[NEW] Cross-current:** The research packet centers on whether resilient AI and semiconductor demand can coexist with tighter Fed policy, a renewed oil-surplus thesis and delayed data-center commissioning ^{[5][1][6][2][3][4]}.
• **[NEW] Source note:** The optical-interconnect, Morgan Stanley AI-timing, China Securities AI-strategy and Kaiyuan Securities 2027-strategy entries are visibly truncated, so this digest uses only their visible figures and stated conclusions ^{[7][2][8][9][10]}.
• **[ONGOING] Net read:** Near-term hardware research remains constructive, but the macro and project-timing case becomes less comfortable across 2027-2029 ^{[5][6][2][3][11][10]}.

1. Macro, Rates and Liquidity ^{[12][3][13][4]}

• **[NEW] Fed path:** Goldman Sachs expects another 25bp Fed hike in March 2027, while CICC retains a 25bp December hike call and says an acceleration is not strongly necessary ^[4].
• **[ESCALATED] Unresolved pace:** Fed Governor Christopher Waller supports a December hike and remains open to 50bp, Standard Chartered says a 50bp scenario has strengthened but cannot yet be concluded, and a Reuters poll finds December hike expectations firm while the tightening-cycle endpoint remains disputed ^[13][4].
• **[NEW] Market pricing:** Goldman Sachs says the December hike was fully priced and the 2-year Treasury yield had reached 4.61% before CPI ^[3].
• **[ESCALATED] Liquidity:** Goldman Sachs rate strategist Isabella Rosenberg says bank reserves are at the edge of the demand curve's steep section and system liquidity is “just enough...” but unable to absorb shocks, while Treasury cash rebuilding, larger Treasury auctions, repo-clearing effects and stalled money-fund growth may converge in H2 2027, when Goldman expects Fed cuts to begin ^[3].
• **[ONGOING] Dollar note:** A note titled “Will U.S. dollar liquidity become the core contradiction?” says dollar liquidity is not yet the main asset-pricing constraint, but SOFR has stayed above IORB since Oct 23, the three cited spreads have ranked above their 90th percentile since Nov 2021, and money-fund cash at ON RRP is nearly exhausted ^[12].
• **[ONGOING] Transmission:** Offshore dollar supply remains ample, while onshore pressure is concentrated in repo, bills and loans continue expanding, corporate issuance contracts and stress sits in CCC-rated borrowers ^[12].
• **[NEW] Test:** The source says the year-end check is whether repo pressure spreads into interbank markets or bond-market credit stress moves into higher-rated issuers ^[12].

2. Oil's Two-Sided Forecast ^[1][14][4]

• **[NEW] Oil reversal:** Brent briefly broke below $90/bbl overnight, its lowest since August, after losing more than 20% over 10 days; JPMorgan says global oil returned to a 1.1mb/d daily surplus in September and 1.4mb/d in October, while retaining a $63/bbl 2027 Brent average forecast ^[1].
• **[NEW] Deal scenario:** JPMorgan says a formal U.S.-Iran agreement could send Brent toward $80/bbl or the mid-$70s, with Middle East policy expert Dr. Trita Parsi putting the Trump administration's escalation-or-negotiation window at about three weeks ^[1].
• **[ONGOING] Fiscal clock:** The same source says U.S. federal funding runs out on Dec 11, congressional defense authorization may not pass until December or January, and House control changes on Jan 3, leaving the three-week window exposed to escalation ^[1].
• **[ESCALATED] Counterforecast:** The EIA raised its 2027 U.S. retail gasoline forecast from $3.56/gal to $3.73/gal and diesel from $4.49/gal to $5.03/gal because global fuel supply is expected to remain tight even if Middle East supply rises ^[14].
• **[NEW] Source contrast:** JPMorgan's surplus case rests on explicit September-October supply balances and inventory accumulation; Moody's expects global growth to stabilize in 2027 and oil prices to begin falling from mid-2027; the EIA's higher fuel forecast is more dependent on a forward global-supply assumption ^[1][14][4].
• **[NEW] Falsifier:** A formal U.S.-Iran agreement can validate JPMorgan's $80/bbl-to-mid-$70s scenario; absent one, the EIA's $3.73/gal gasoline and $5.03/gal diesel forecasts and the three-week political window remain the counterweights ^[1][14].

3. AI and Semiconductors ^{[5][7][6][2][11][10]}

• **[ESCALATED] Selective optimism:** Goldman Sachs says the Philadelphia Semiconductor Index's 21% one-month fall pushed the sector into oversold territory and named Nvidia, Broadcom and AMD as the most attractive allocation candidates, while BofA says the memory-chip upcycle can run through 2027 and avoid a downcycle in 2028 ^[5][6].
• **[NEW] Optical links:** A truncated industry report says AI optical-interconnect demand has not slowed, with domestic vendors advancing product breakthroughs and customer introductions while high-frequency, high-speed cables begin ramping ^[7].
• **[ESCALATED] Timing gap:** Morgan Stanley's “Buy the Drop: AI Investment Opportunities Post the US Midterms” estimates about 5GW of U.S. data-center capacity may move from 2028 into 2029, equal to 20% of annual U.S. additions and 10% of global additions, with U.S. GW additions then falling 4% y/y in 2028 and rebounding 38% in 2029 against global growth of 9% and 20% ^[2].
• **[NEW] Supply-chain gap:** Morgan Stanley warns of a 2028 “air pocket” for upstream cloud semiconductors, with weaker pricing power or inventory accumulation, slower orders, normalized margins and earnings cuts even if AI demand remains intact ^[2].
• **[NEW] Counterweight:** Kaiyuan Securities estimates the global cloud market reached $143.4bn in 2026Q2, up 43% y/y, and expects nine major cloud vendors' 2027 capex to exceed $1.5tn, while also noting expanding debt, equity and project financing and wider spreads at some highly leveraged vendors; it says backlog conversion depends on when new compute capacity enters service ^[10].
• **[ONGOING] Materials bottleneck:** China Securities' 2027 glass-fiber strategy sees LDK2 fabric as the largest-potential specialty application, with Rubin and later GPUs using at least about 2.8 times Blackwell's per-GPU quantity and the 2028 supply gap exceeding 90m meters ^[11].
• **[NEW] Falsifier:** Morgan Stanley's delay case is weakened if U.S. additions avoid the projected 4% 2028 contraction, while the glass-fiber supply case is tested by whether domestic looms reach mass application in H2 2027 ^[2][11].

4. Corporate and Strategy Research ^{[15][16][17][18][19][20][21][8][9][22][4]}

• **[NEW] Sinopec:** Changjiang Securities keeps a Buy rating as refining spreads recover; Q1-Q3 attributable profit was RMB38.988bn, up 30.0% y/y, and Q3 profit was RMB13.361bn, up 57.2%, with 2026-2028 EPS of RMB0.34/0.42/0.46 and P/E of 16.93x/13.80x/12.67x at the Nov 9 close, while the report also cites H1 A- and H-share buybacks and a payout ratio above 50% ^[15].
• **[NEW] Yangtze Power:** Guosen Securities keeps Outperform despite Q1-Q3 attributable profit falling 6.56% to RMB26.344bn and Q3 profit falling 23.44% to RMB11.588bn, citing lower financing costs and steady cash flow, with forecast P/E of 22x/20x/19x ^[18].
• **[NEW] Shenhua:** Western Securities keeps Buy after Q1-Q3 attributable profit rose 3.8% to RMB44.403bn and Q3 profit rose 3.2% to RMB15.688bn; coal output fell 2.3% to 382m tonnes, sales edged up 0.1% to 461m tonnes, and 11 of 12 acquired targets were consolidated ^[19].
• **[NEW] NetEase:** CLSA keeps Outperform and a $150 target as it forecasts Q3 revenue rising 6% to RMB30.2bn and adjusted EBIT rising 31% to RMB11.6bn, while flagging disappointing performance for a new game ^[20].
• **[NEW] Target cuts:** UBS cut Infineon's target price from €64 to €55, while B. Riley cut AST SpaceMobile's from $65 to $48 and kept Neutral ^[16][17].
• **[NEW] Autonomous driving:** Huatai Securities says L3 road-test authorization and responsibility boundaries are becoming clearer, with Volkswagen China obtaining a license in Hefei, L3 capability potentially delivered as early as H2 2027, and Momenta's L3/L4 cycle potentially inflecting in 2028, while higher redundancy and compute requirements lift execution- and decision-system value per vehicle ^[21].
• **[NEW] Consumption:** Guojin Securities finds that above-quota retail has underperformed overall because social-group consumption is contracting rather than because the resident base has collapsed, with resident goods consumption still growing about 3.6% and necessities stable, discretionary spending resilient and big-ticket categories weak ^[22].
• **[NEW] AI strategy:** China Securities says its 2027 preference shifts from model capability toward application-deployment opportunities as large models enter commercial scenarios, while UBS says foreign Chinese-equity positioning is broadly unchanged and remains bullish on Chinese AI technology hardware ^[8][9][4].

5. China Market Structure and Decision Map ^{[1][2][12][3][23][14][13][24][4]}

• **[ONGOING] Duration:** Public funds' bond duration reached a near-three-year high and several pure-bond products returned more than 10% this year, but institutions cite low long-end valuations and crowded trading, prompting some rotation from ultra-long to mid-long bonds ^[23][4].
• **[NEW] External trade:** China's October exports rose 35.2% y/y and first-ten-month exports rose 22.7%, while machinery and electrical-product exports increased 46.06% and integrated-circuit and automatic-data-processing-equipment exports doubled in October ^[13].
• **[NEW] Unlock scale:** December lockup expiries total RMB474.2bn across 105 stocks versus RMB116.6bn in November, with median unlock value rising from 5.1% to 51.5% of free-float market cap, the ten largest accounting for RMB325.1bn or 68.6% with each above 50%, and the 39 stocks with market caps below RMB5bn representing only 8.8% of value but a 61% median free-float ratio ^[24].
• **[NEW] Caveat:** Dongwu Securities cautions that unlocking does not equal selling because restricted shares merely gain sale eligibility and actual pressure depends on shareholder intent and reduction rules ^[24].
• **[NEW] Rates decision:** The stated policy fork is a 25bp base case versus a 50bp scenario in December, with Goldman separately forecasting a 25bp March move; the later macro test is whether repo and credit pressure spread beyond their current pockets ^[3][13][4].
• **[NEW] Oil decision:** A formal U.S.-Iran agreement is the event that can validate JPMorgan's $80/bbl-to-mid-$70s scenario; without one, the EIA's $3.73/gal gasoline and $5.03/gal diesel forecasts and the three-week political window remain the counterweights ^[1][14].
• **[NEW] AI decision:** The timing test is whether U.S. data-center additions avoid a 4% 2028 contraction, whether the projected 5GW slips into 2029, and whether cloud backlogs convert as commissioned capacity comes online ^[2][10].
• **[NEW] Corporate decision:** Sinopec's refining-margin recovery, Yangtze Power's inflow-sensitive generation and Shenhua's post-consolidation coal production are the company-level variables behind the respective Buy and Outperform calls ^[15][18][19].