The Fed: No Surprises, Loads of Uncertainty

The Fed held the policy rate at 3.50%–3.75% and the SEP now forecasts one 25bp cut by year-end and one in 2027, with 2026 PCE inflation revised up to 2.7% (from 2.4%). Markets sold off (all three major indexes down >1%, Dow -1.63%) as Powell emphasized uncertainty and a spike in natural gas after an attack on an LNG site in Qatar raised energy-price inflation risks. Hyperscalers (Alphabet, Amazon, Meta, Microsoft) plan roughly $650bn of capex in 2026, creating an AI infrastructure-driven capital cycle that is largely interest-rate insensitive and stressing demand for copper, memory, and energy infrastructure. Conversely, Morgan Stanley warns direct-lending defaults could approach ~8%, highlighting rising stress in private credit and software-backed loans and warranting review of exposure to private credit managers, BDCs, and software-heavy lending portfolios.

Analysis

The hyperscaler-driven capex cycle is creating a multi-year, supply-constrained investment loop that behaves structurally different from rate-driven liquidity cycles. Because contracts and build schedules are long-dated, the marginal buyer is not levered to short-term rates — that shifts pricing power to upstream suppliers (copper, power transformers, specialty memory) and to logistics nodes with long lead times; expect material lead-time dislocations to persist for 12–36 months.

Private-credit and software lenders sit on the opposite side of this bifurcation: portfolios were underwritten to steady SaaS cash flows that AI can compress and to funding models that assume low redemption stress. The key second-order failure mode is not a single corporate default but a liquidity spiral — gating or markdowns trigger forced selling of illiquid loans, steepening secondary loan discounts inside quarters rather than years.

For portfolio construction this implies asymmetric opportunities: overweight durable, balance-sheet-rich hyperscalers and their constrained suppliers while tactically short credit-exposed managers/BDCs and selected legacy software franchises that face direct workflow automation risk. The biggest external risks to this stance are a sudden normalization in energy availability/cost (which can stall builds quickly), or rapid macro easing that re-prices credit-sensitive names before defaults materialize; both can meaningfully shorten the window of outperformance for infra-linked equities.

Key Decisions for Investors

• Buy 9–18 month call spreads on GOOGL/GOOG (LEAP-style financing) to capture upside from continued hyperscaler capex; fund by selling a higher-strike call to cap cost. Target return: 2–4x premium if AI buildout accelerates; loss limited to net premium. Monitor quarterly capex cadence and adjust at earnings/guide points.
• Pair trade: long AMZN (AWS leverage) or MS (cloud exposure) 6–12 months vs short OWL and BX equity or buy put protection on those names. Rationale: capture secular cloud demand while extracting asymmetry from private-credit liquidity risk; size shorts at 25–50% of longs to limit beta mismatch; take profits if loan-market spreads tighten by >150bps.
• Short ADBE via outright equity or buy 6–12 month puts—target companies with high recurring-revenue multiples and direct workflow automation exposure. Reward is downside if AI compresses renewals/ARPU; risk is product re-monetization or strong AI-enabled growth — stop-loss at 20% adverse move or on positive product cadence surprises.
• Add commodity/memory exposure (copper miners or targeted memory suppliers via ETFs/selected names) on pullbacks as a hedge on sustained capex-driven input inflation. Time horizon 12–36 months; expect a run if lead times remain constrained, but set tranche sell targets at 30–50% gains and hedge with short-call overlays as supply responses materialize.