Why this bull market is unlike any we've seen before

The S&P 500 is up 12% year to date despite a 2% decline last week, when AI-linked stocks sold off amid concerns that borrowing to build AI networks may outpace demand; CoreWeave fell more than 25%, Micron dropped 15% and AMD fell 13%. The bull market has completed four years, with CFRA’s Sam Stovall noting double-digit advances in each year and an average fifth-year gain of nearly 21% for post-WWII bull markets. However, he called this the most expensive bull market of that period, at a forward P/E near 23; FactSet data showed S&P 500 Q3 earnings tracking 29% higher year over year.

Analysis

The historical fifth-year pattern is a poor timing signal: it may support dip-buying sentiment, but it does not offset the asymmetric risk created by a high starting multiple. The key near-term test is whether earnings revisions broaden beyond a narrow group of AI beneficiaries. If reported growth is concentrated while capital spending and borrowing rise faster than monetization, the market can de-rate before aggregate earnings actually contract.

The AI selloff also separates two exposures. CRWV is more directly vulnerable to investor scrutiny of debt-funded capacity: utilization, customer commitments, and funding terms matter more than headline AI demand. AMD and MU can be hit by the same capex doubts, but their earnings sensitivity depends on realized chip demand and pricing, not simply whether infrastructure is being built. A pullback in AI infrastructure spending could therefore hurt compute providers first, then spill into semiconductor estimates with a lag.

Over days, expect sentiment and positioning to dominate. Over 1–3 months, watch earnings revisions, AI customer spending plans, and credit spreads for infrastructure borrowers. Over 6–18 months, the decisive question is whether deployed capacity earns adequate returns. The bullish case is falsified by sustained estimate cuts alongside widening credit spreads; the bearish case is weakened if earnings breadth improves and capex converts into recurring revenue. Historical averages do not establish either outcome.

Key Decisions for Investors

• Do not add broad equity risk solely on the bull market’s age or historical fifth-year returns. Treat the forward multiple as leaving less room for earnings disappointments.
• Use index put spreads as a defined-risk, 1–3 month hedge if earnings revisions weaken while AI infrastructure credit spreads widen; avoid paying for protection if revisions stabilize and market breadth improves.
• Keep CRWV risk tighter than diversified semiconductor exposure until utilization, customer concentration, and financing terms support the growth case. A relative short versus a semiconductor basket is only an alert, not a trade, until those items and current valuations are verified.
• For AMD and MU, wait for company-specific guidance and estimate revisions before taking a directional position. Better-than-feared demand and stable pricing would challenge the current AI-demand skepticism; further estimate cuts would validate it.