Reviewed 2026-07-10
AI Memory Pricing-Power Test
The AI memory setup is no longer only a demand story. The signal is whether HBM and DRAM suppliers can keep pricing power through the next round of earnings, capex guidance, and macro repricing.
Underlying demand
AI compute capacity that requires high-bandwidth memory, advanced packaging, server buildout, and customer willingness to absorb component cost.
Decision variable
HBM and DRAM availability, capacity allocation, packaging/test throughput, and whether customers accept another contract-price reset.
Capital relevance
Research-candidate layers include memory-linked capacity, packaging, testing, thermal/power infrastructure, and OEM margin exposure.
Evidence
Level 2 pricing and capacity indicators; Level 3 supply-discipline inference; Level 4 conditional repricing scenario.
Verify
Q3 DRAM/HBM pricing commentary confirms another reset; order books are described as constrained; hyperscaler capex language remains firm despite higher component costs.
Disprove
Pricing rolls over, customers resist contract resets, suppliers add capacity faster than demand, or hyperscalers push delivery schedules out.
Reviewed 2026-07-10
Hormuz Chokepoint Premium
A renewed Hormuz shock is not only an oil-price story. It tests whether a route treated as free infrastructure becomes a priced delivery gate across shipping, insurance, LNG, inventories, and energy-security capital.
Underlying demand
Energy delivery that reaches the buyer: barrels and LNG cargoes that arrive, not supply that exists only upstream.
Decision variable
The Strait of Hormuz as a physical and commercial gate, including routing control, war-risk insurance, tanker willingness, AIS behavior, port access, and bypass capacity.
Capital relevance
Research-candidate layers include non-Hormuz LNG supply, bypass corridors, port logistics, risk-bearing shipping capacity, insurance, inventory, and energy-security capex.
Evidence
Level 3 delivery-gate inference; Level 4 conditional repricing scenario; current shipping and insurance data still require refresh.
Verify
Transit counts remain below the pre-shock baseline; war-risk premiums, tanker U-turns, AIS gaps, or delayed LNG loadings persist.
Disprove
Transit normalizes, risk premiums fall, threat levels step down, and cargo behavior no longer reflects chokepoint risk.
Reviewed 2026-07-10
Two-Temperature Economy
A single headline economy can carry two operating temperatures: capital continues to fund strategic capacity while rate-sensitive households, small firms, and refinancing channels cool.
Underlying demand
Balance-sheet durability: cash flow, refinancing access, working capital, collateral quality, and pricing power through a slower consumption layer.
Decision variable
The financing gate between favored capex lanes and rate-sensitive balance sheets, including credit cost, refinancing windows, labor cooling, and collateral pressure.
Capital relevance
Research-candidate layers include resilient cash-flow operators, credit-workout services, essential discount channels, and infrastructure or defense lanes with non-discretionary demand.
Evidence
Level 2 macro and credit indicators; Level 3 capital-allocation inference; Level 4 conditional repricing scenario.
Verify
Strategic capex stays firm while labor, credit, housing, small-business, or consumer data weaken; spreads and delinquency measures diverge by borrower quality.
Disprove
Labor and credit reaccelerate broadly, or strategic capex slows enough that the economy no longer shows a two-temperature split.
Reviewed 2026-07-09
Large Load Interconnection Rules
FERC's June 2026 large-load action makes process design, readiness, cost allocation, and adequate generation part of the speed-to-power bottleneck.
Underlying demand
Reliable speed-to-power for large loads, with clear cost and readiness obligations.
Decision variable
Regional tariff and study processes that must separate credible projects from speculative requests while preserving reliability and assigning costs.
Capital relevance
Research-candidate layers include grid planning, engineering, flexible-load controls, transmission technologies, and generation-to-load integration.
Evidence
Level 1 FERC action; Level 3 enabling-layer inference; Level 4 conditional repricing scenario.
Verify
RTO and ISO filings adopt clearer study, readiness, cost-recovery, and flexible-service rules.
Disprove
Large-load forecasts fall materially, or new rules clear the process bottleneck without sustained enabling-layer pressure.
Reviewed 2026-07-09
AI Infrastructure: Deliverable Power
Official demand evidence supports rapid data-center electricity growth; the question is whether power, interconnection, equipment, and commissioning can convert demand into runnable capacity.
Underlying demand
Runnable compute capacity: powered, cooled, connected, commissioned, and available on a defined timeline.
Decision variable
Deliverable power at the required site and date, not electricity or generation capacity in the abstract.
Capital relevance
Research-candidate layers that control electrical equipment, interconnection execution, site enablement, and time-to-energization.
Evidence
Level 1/2 demand evidence; Level 3 bottleneck inference; Level 4 conditional repricing scenario.
Verify
Official load forecasts rise and are matched by site-specific power arrangements, permits, equipment delivery, and commissioning milestones.
Disprove
Demand forecasts fall materially, or interconnection, equipment, and commissioning timelines shorten enough for energized capacity to keep pace.