The U.S. equity market is in a mechanical repricing cycle. The convergence of the 2027 earnings-growth cliff, the discount-rate regime shift, and the AI capex circularity will compress concentrated tech valuations by 10–15% within the November 2026 – May 2027 window. The 20%+ tail is gated by a single variable: the AI capex → revenue conversion. If enterprise AI revenue lags infrastructure outlays, the repricing escalates to a bear market.
The market is pricing in sustained double-digit EPS growth, a benign discount-rate environment, and policy stability. All three are structurally shifting. The 10-year yield at 5.3% changes the discount-rate denominator. The earnings-growth cliff from 35% to 15% changes the expectation-reality gap. The AI capex circularity changes the tail-risk profile. The correction is the sum of these three repricing vectors.
The Discount-Rate Regime Shift
The 10-year Treasury yield at 5.3% — a 24-year high — and the Federal Reserve’s higher-for-longer posture end the decade-long cheap-capital regime. Long-duration growth assets are hyper-sensitive to discount rates. When the risk-free rate is 4–5% and rising, the TINA trade erodes mechanically.
| Variable | 2016–2021 Regime | 2026 Regime |
|---|---|---|
| 10Y Treasury yield | 1.5–2.5% | 5.3% (24-year high) |
| Fed funds target | 0–2.5% | 3.75–4.00% |
| Core PCE inflation | 1–2% | 3.4% (sticky) |
| Discount-rate environment | Cheap capital | Higher-for-longer |
The discount-rate regime shift is the macro-structural input. It changes the denominator in the valuation equation. When the discount rate rises, the present value of future cash flows falls. For high-valuation tech assets, this is a direct repricing vector.
The Earnings-Growth Cliff
The market is pricing in sustained double-digit EPS growth. Consensus projects earnings growth collapsing from 35% to 15% in 2027. In a high-valuation regime, decelerating growth triggers multiple compression — the gap between the priced-in expectation and the realized trajectory is the repricing vector.
| Year | S&P 500 Earnings Growth | Mag 7 Forward P/E |
|---|---|---|
| 2026 | 32–35% | 23–28x |
| 2027 (consensus) | ~15% | — |
The earnings-growth cliff is the expectation-reality gap. The market is pricing in a trajectory that consensus itself projects will not materialize.
Will the AI Capex Convert to Revenue?
The AI buildout is increasingly debt-financed and circular. Nvidia invests in OpenAI, OpenAI buys Nvidia chips, hosted by Oracle. This mirrors the 2000 fiber buildout: real infrastructure, unproven returns.
| Hyperscaler | Annual AI Capex |
|---|---|
| Amazon | ~$200B |
| Alphabet | $175–185B |
| Meta | $110–135B |
| Microsoft | $110–135B |
The capex is real; the returns are unverified. If enterprise AI revenue lags infrastructure outlays, the repricing will be sudden and severe. This is the tail-risk driver for a 20%+ bear market.
The Political Repricing
A Democratic House flip plus a contested Senate creates a policy environment where the market’s current assumptions — deregulation, tax cuts, aggressive industrial policy — face headwinds. Democratic control of the House means subpoena power, oversight, and spending pressure. This reprices policy-sensitive sectors and adds a non-fundamental volatility source.
The Mechanical Index-Concentration Argument
A 10% index decline requires only a 20–30% repricing of the largest companies, with the rest of the index flat. This is why a 10% correction is much more likely than a 20% bear market.
The correction concentrates in tech. The S&P 493 (ex-Mag 7) is at 17.5–19x forward earnings, near historical averages, providing a cushion.
The Calibrated Probabilities and the Gating Condition
| Outcome | Probability | Gating Condition |
|---|---|---|
| 10–15% correction | 60% | Earnings cliff + discount-rate shift + AI capex circularity |
| 20%+ bear market | 22% | AI capex → revenue conversion failure |
| No material correction | 18% | AI infrastructure profitability + earnings still strong |
The gating condition is the AI capex → revenue conversion. If enterprise AI adoption and revenue generation keep pace with infrastructure outlays, the market absorbs the deceleration without a major drawdown. If AI revenue lags capex, the repricing is sudden and severe.
Closing
The next decisive evidence is Q4 2026 hyperscaler earnings (January–February 2027) — specifically AI revenue disclosure and capex guidance — plus the November 3, 2026 midterm results. If AI revenue lags capex, the repricing escalates to a bear market. If AI revenue keeps pace with capex, the market absorbs the deceleration without a major drawdown. The political repricing adds a non-fundamental volatility source that amplifies the core mechanism.