We Are Witnessing the Stock Market Do Something for Only the 3rd Time in 156 Years, and History Is Clear About What Comes Next

The S&P 500 has returned roughly 21% annually in recent years, while its CAPE ratio is 40.5 versus a long-run average of 17.4, prompting concerns about elevated valuations and potential drawdowns. Goldman Sachs estimates hyperscalers could spend $800 billion on AI capex in 2026 and $1.4 trillion by 2028; a Wall Street Journal study says annual AI spending equivalent to 8.8% of U.S. GDP would be needed to justify the industry's investment. The article notes Oracle shares have fallen 50% over the past 12 months and advises taking some profits and diversifying away from stocks highly exposed to generative-AI spending.

Analysis

The key risk is not that AI investment stops abruptly; it is that the return on each incremental dollar falls below investors’ hurdle rate. That can pressure cloud buyers’ free cash flow and capital returns first, then flow back through the supply chain as order growth, utilization and pricing expectations reset. Semiconductor suppliers are not insulated simply because current earnings are strong: memory demand and pricing can be more exposed to an inventory or capacity digestion phase, while GPU demand may prove more durable if workloads continue to scale. The transmission is likely uneven, not a uniform “AI bubble” unwind.

The GDP-spending threshold cited in the article is a model assumption, not a clean break-even test: AI value can accrue through productivity or cost savings rather than a directly identifiable AI revenue line. Conversely, high aggregate capex does not establish attractive returns. Track hyperscaler capex guidance against AI monetization, data-center utilization and free-cash-flow conversion over the next 1–3 months; equipment orders and memory pricing are key 6–18 month confirmation points.

High CAPE raises vulnerability to disappointment but is a weak timing signal. A sharp short based on valuation alone risks fighting earnings momentum. Oracle’s reported share weakness is a warning about spending scrutiny, not proof that all AI buyers face the same economics. The thesis weakens if buyers sustain capex while improving AI-related revenue or measurable cost savings without deteriorating cash generation.

Key Decisions for Investors

• For portfolios concentrated in AI infrastructure, trim position sizing in NVIDIA (NVDA) and Micron Technology (MU) rather than liquidating solely on CAPE. Reassess after the next hyperscaler guidance cycle; restore exposure if capex is supported by improving utilization, monetization and cash-flow conversion.
• Avoid treating Oracle Corporation (ORCL) as a clean short proxy for the entire AI trade: company-specific execution and financing details need verification, and an already weak share price may leave less asymmetry. Watch its forward cash-flow outlook and spending guidance for evidence of broader buyer retrenchment.
• Consider a defined-risk 3–6 month put spread on a broad growth-equity proxy only if hyperscalers cut capex plans or supplier order commentary weakens; check implied volatility and premium before entry. This limits loss to the debit while targeting a de-rating catalyst, but offers poor risk/reward if earnings expectations remain intact.
• Falsification / alert: the cautious thesis is undermined by sustained buyer capex accompanied by accelerating AI monetization, stable or improving free-cash-flow conversion, and firm supplier order trends. Escalate the hedge if capex is maintained while cash generation deteriorates, or if memory pricing and infrastructure orders turn down together.