Three signals.
One systems test.

Executive view

The headline is not the system.

A market label compresses many different structures into one phrase. “Private credit,” “AI demand,” and “high valuations” become useful only after the underlying mechanism is made visible. The engineering task is to identify the load path, the bottleneck, and the evidence that would change the decision.

01 · Private credit

Risk lives in the structure.

The Federal Reserve estimates that private-credit loans reached about $1.4 trillion in the second half of 2025—large enough to matter, but too varied for one conclusion. The relevant variables are borrower cash generation, debt-service capacity, collateral, seniority, documentation, sector concentration, and the liquidity offered to investors.

Traditional private-debt funds commonly lock investor capital for years. Newer semi-liquid vehicles permit periodic redemptions subject to limits. The Federal Reserve reported notable increases in redemption requests through early 2026; most managers capped redemptions, while requests remained manageable and first-quarter outflows only moderately exceeded inflows. The disciplined view is that liquidity pressure rose, but this evidence alone did not establish system-wide failure.

02 · AI infrastructure

Compute can scale faster than power systems.

The EIA reports that U.S. electricity demand grew about 1.7% annually between 2020 and 2025 after growing only 0.1% annually from 2005 through 2019, with data centers among the principal drivers. But aggregate megawatts do not identify a project’s critical path.

The binding constraint may be generation, transmission, an interconnection study, a substation, long-lead equipment, cooling, construction labor, or a customer commitment. Investment analysis should therefore distinguish requested capacity, contracted capacity, deliverable power, energized capacity, and economically utilized capacity.

Demand is a signal. Deliverable capacity is the system.

03 · Public markets

Price and earnings answer different questions.

The Federal Reserve’s May 2026 review found the forward price-to-earnings ratio of S&P 500 companies close to the upper end of its historical range. Separately, the Bureau of Economic Analysis estimated first-quarter 2026 corporate profits from current production at a seasonally adjusted annual rate of $4,426.5 billion.

Those observations can coexist. Aggregate profits do not show that every security is fairly priced, while a demanding valuation does not prove that current income is fictitious or that reversal is imminent. The useful test is company-level: revenue concentration, margins, reinvestment, cash conversion, balance-sheet obligations, and the growth already embedded in price.