Fed’s Cook sees AI inflationary push as a top 2027 risk

Fed Governor Lisa Cook identified AI-related investment demand and recurring supply shocks as key upside inflation risks for 2027, warning that AI buildout pressures may not ease quickly enough to be offset by eventual productivity gains. The Fed raised its policy rate 25bps last month, while its preferred inflation measure stood at 3.4% in August, still 140bps above the 2% target. Cook said Middle East conflict and other geopolitical disruptions could further constrain supply chains and may require a different monetary-policy response than traditionally looking through supply shocks.

Analysis

The relevant transmission is not a near-term change in AI demand but a higher-for-longer discount-rate premium applied to AI-linked equities if power, grid equipment, memory, and data-center construction remain capacity constrained. SMCI is particularly exposed to the capex-cycle side of this equation: strong order flow can coexist with multiple compression if customers face higher financing costs or if component inflation prevents gross-margin recovery. APP has less direct infrastructure exposure, but its valuation remains sensitive to real yields and to any ad-spend slowdown caused by tighter financial conditions.

Over the next 1-3 months, the cleaner read-through is dispersion within technology rather than an outright broad-market short: asset-light software with demonstrable productivity monetization should hold up better than hardware vendors dependent on constrained physical inputs. Over 6-18 months, persistent AI infrastructure inflation could benefit power and electrical-equipment bottlenecks more reliably than compute assemblers, while eventually creating a productivity offset for software and automation beneficiaries. The thesis is falsified if core inflation and long-end yields decline despite continued AI capex, or if hyperscaler earnings show infrastructure spending is being funded without pressure on margins or free cash flow.

Consensus may be too focused on AI as a secular earnings tailwind and insufficiently attentive to its sequencing: spending, power demand, and financing costs arrive before broad productivity gains. That makes a rate-driven selloff in high-beta AI hardware plausible even if unit demand remains healthy; conversely, a one-day hawkish reaction is not itself sufficient evidence for a durable de-rating without confirmation from yields and capex guidance.

Key Decisions for Investors

• Maintain a 1-3 month defensive hedge on AI-duration exposure via a long TLT position or call spread against concentrated SMCI/APP exposure; the hedge works if renewed inflation repricing pushes long yields higher and compresses growth multiples, but should be reduced if the 10-year yield breaks lower on disinflation data.
• Prefer a relative-value position of long APP / short SMCI over the next 1-3 months only if SMCI fails to demonstrate sequential gross-margin improvement; APP is still rate-sensitive, but SMCI carries incremental component-cost, working-capital, and hardware-cycle risk.
• Do not add outright SMCI length solely on AI-demand headlines. Reassess after the next earnings release for evidence that revenue growth is converting to gross margin and operating cash flow; a guidance raise without margin/cash-flow confirmation is a sell-the-rally risk.
• Set an alert around hyperscaler capex guidance and data-center power procurement disclosures over the next two reporting cycles. Broad capex acceleration alongside rising power and equipment costs favors shifting AI exposure away from assemblers and toward electrical-grid/power beneficiaries, but the specific trade requires confirmation of supplier backlog and valuation.