Long-end yields are resisting Treasury support; $100 oil is shifting value toward shipping and refiners; AI demand is real but delivery and financing separate winners; memory, optics and chip testing are the next capacity bottlenecks.
The expanded buyback still left the 10-year yield at its highest since 2023, while a proposed $5,000 election dividend would add another inflationary fiscal impulse. A near-consensus Bank of Japan hike and a restrictive ECB path reduce the room for global duration relief. Long bonds need softer inflation or credible fiscal restraint, not another small liquidity operation.
Commodity vessel traffic through Hormuz fell to single digits as Bloomberg said the strait may remain effectively closed for months. Russian diesel exports are at decade lows, while physical-market specialists see China bidding for marginal barrels. The cleaner equity expression is refinery and tanker scarcity, but a diplomatic reopening or materially weaker Chinese demand would break the setup.
Morgan Stanley can still model 25%-50% returns on AI infrastructure, and Nebius says its order book extends into H1 2028. Yet Crux now gives projects only a 50% chance of meeting delivery dates, while customer prepayments can conceal weaker contract economics. Demand is not the scarce variable; permitted power, on-time megawatts and financing quality determine who earns the return. bubbleboi adds a useful demand-side extension: the higher-value endgame may be AI labs selling compute-backed R&D partnerships to industrial experts, not $200 consumer subscriptions.
TSMC's 53.3% monthly sales growth confirms demand, but the useful signal is where capacity is now rationed. Nvidia is reportedly paying premiums for probe-card and test-socket capacity, MOCVD lead times have stretched to 8-12 months, and Chinese AI-chip prices are rising with HBM shortages. Testing, optical tools and memory retain pricing power until new capacity actually ships.
TradingWarz reports a $50,000 MSFT share position and is adding 2028-plus $600 calls after the options doubled.
seedy19tron added BHVN in the $12s, arguing the regulatory hold is not an MAE under the merger agreement and SK lacks a walk right.
A Reddit investor initiated ONWD as ARC-EX moves into commercialization, citing 240% revenue growth and home-use expansion while acknowledging cash-burn and clinical risk.
ParadisLabs highlights visibility into H1 2028, while Northwise says customer prepayments help only if contract pricing and credit quality remain healthy.
The $1,999 iPhone Duo may be a niche product as memory costs rise, although another analyst sees pricing power and scale supporting an upgrade cycle.
Lumentum says the $8 billion OCS market estimate is too low, but long MOCVD lead times make capacity installation and revenue guidance the real checkpoints.
A Reddit author took a bearish Reverse LEAPS position, arguing Nike sacrificed scarcity and pricing power for volume.
A WallStreetBets commenter holds IWM puts, citing fragile companies and pressure from rising yields and oil prices.
__con_ disclosed a 3x PONS short while otherwise arguing for higher-beta crypto exposure, making this a specific hedge rather than a broad crypto exit.