Markets are absorbing a violent sector rotation that began Thursday and extends into Friday's pre-market. The S&P 500 closed at 7,720 (-0.3% on the day, +0.7% on the week) but the equal-weight S&P outperformed by 1%, its largest beat in over two years, as money rotated out of megacap tech and into value, healthcare, and financials.

Nasdaq-100 futures are down 2.2% overnight, signaling a gap-down open Monday. The VIX jumped 5.5% to 18.15, its seventh straight weekly gain, signaling rising fear even as the index sits near highs. Spot Bitcoin ETFs saw $514 million net inflows over the last 14 days, but yesterday's $17 million was the smallest in the streak.

The bond market is telling a different story than the stock market. The 10-year Treasury yield at 5.05% (up from 4.99% Thursday) and the 2-year at 4.67% produce a 38-basis-point 2s10s spread, positive but compressed, the classic "bear steepener" driven by term premium, not rate expectations. The ACM term premium estimate has turned positive for the first time since 2022, now around 80 basis points versus a 20-year average of 40. That premium is the fingerprint of the AI capex supercycle: $649 billion of hyperscaler demand for capital is lifting long rates regardless of what the Fed does.

Credit markets are flashing amber. High-yield OAS widened to 3.14% (from 3.18% a week ago but up from 2.82% a year ago) while Treasury yields fell, a divergence that in every soft landing since the 1990s has signaled hardening credit conditions. Investment-grade OAS at 0.86% is also elevated. The Chicago Fed's National Financial Conditions Index ticked tighter to -0.487, the first sign this autonomous tightening is biting the real economy.

The dollar is caught in the crossfire. DXY at 101.7 is off its highs, but the broad trade-weighted index at 121.3 remains near multi-decade highs, pressuring multinational earnings and raising debt-service costs for USD borrowers globally. EUR/USD at 1.12 and USD/JPY at 157.3 are range-bound, waiting for the BOJ's next move and the Fed's December decision (87% priced for a hike).

Equity breadth is narrowing dangerously. Only 59% of S&P 500 components trade above their 50-day moving average (down from 62% Monday), and 61% above the 200-day. The Russell 2000 outperformed +2.7% on the week as small caps catch a bid from falling rate hopes, but the Semis ETF sits 21.5% below its 52-week high while the Software ETF is up 15.7% over three months, a split that didn't exist six months ago.

The Strait of Hormuz is effectively a war zone. Iran claims four tankers struck mines Thursday; the UK Maritime Trade Operations confirmed a limpet mine attack on a transiting vessel. This is the fourth straight week of projectile strikes and naval mining in the strait, alongside ongoing Houthi missile waves targeting Saudi energy infrastructure and Riyadh.

Saudi Arabia's East-West Pipeline (Petroline) restarted October 1 and moves up to 5 million bpd via Red Sea ports, bypassing Hormuz entirely, but Brent's $12 premium of dated spot over futures signals physical tightness the bypass hasn't solved. OPEC+ meets Sunday and is expected to hold December quotas unchanged, keeping the 2 mb/d baseline cut from 2022 through year-end.

In Ukraine, a Russian oil refinery was struck overnight, ending a two-week pause in Kyiv's refinery attacks that came under U.S. pressure. The strike defies Trump's push to halt such operations and adds another supply-side prop to oil prices. Meanwhile, Washington paused a new Russia sanctions package and agreed to purchase Russian diesel, moves criticized as aiding Moscow's war effort.

China's property sector continues its slow unwind. Evergrande liquidators have recovered only 1% of debt claims after 30 months. PAG agreed to buy a 51% stake in Evergrande Property Services for $513 million, a rare win for creditors but at a fraction of peak value. On the flip side, China's chip exports surged 136% year-over-year to $45 billion in Q3, driven by AI-related demand, confirming the semiconductor supply chain is rerouting around U.S. restrictions.

Marvell drops 13.2% in sympathy. MRVL fell to $218.34 despite no company-specific news. The chip complex is being repriced as a group: investors are questioning whether the AI infrastructure build-out can sustain current valuation multiples when hyperscaler FCF is evaporating.

Eli Lilly surges 6.6% on obesity drug dominance. LLY hit $1,373.70 (+18.7% on the month) after Retatrutide delivered double the weight loss of Ozempic's highest dose in a head-to-head trial. The GLP-1 franchise is becoming a cash machine that funds its own R&D, a contrast to AI capex with no visible payback.

Mastercard gains 4.2%, Visa 2.5%. MA at $635.70 (+15.3% monthly) and V at $410.54 (+14.3% monthly) are hitting new highs. Payment networks are the toll booths on global commerce, they win whether AI pays off or not.

Healthcare sector leads. XLV rose 2.4% to $184.04 (+9.3% monthly). Defensive earnings, demographic tailwinds, and drug approval cycles (Weltruza for schizophrenia, Juvmo for Parkinson's, Jaypirca for blood cancers) are delivering while tech promises.

The Fed held rates at 3.75-4.00% Wednesday with a 10-2 vote, Governors Cook and Hammack dissented for a 25bp hike, the first split since the tightening cycle began. The statement upgraded the economy to "strong" from "solid" and noted inflation "remains elevated." Chair Warsh called financial conditions "not tight" and pushed back on any soft inflation target. The September dot plot median sees 4.1-4.4% by year-end 2026; CME FedWatch prices an 87% chance of a December hike.

Core PCE printed 3.12% year-over-year in September (0.3% monthly), released Thursday, still a full point above the 2% target. Q3 GDP came in at 3.4% annualized, beating the 3.2% consensus, driven by consumer spending and government outlays. The Atlanta Fed's GDPNow tracker for Q4 is already trending below 2%.

The Treasury auction calendar is light this week, 2-year, 5-year, and 7-year notes were sold Monday-Wednesday with solid demand. Next week brings the November refunding announcement (3-year, 10-year, 30-year) which will test appetite for duration at 5% yields.

The sharper edge: Professionals are debating whether this constitutes "private fiscal dominance", AI capex as a de facto fiscal stimulus that makes monetary policy impotent on the long end. The live question: does the term premium stay elevated until capex peaks (2027-28 per Anthropic's $40T TAM), forcing the Fed to choose between choking the real economy with high short rates or letting inflation run? Credit markets (HY OAS 3.14pp, IG 0.88pp) haven't priced this duration risk yet.

Watch: ACM term premium decomposition on the 10y (currently ~80bps vs 20y avg ~40bps), if it holds above 70bps through Q4 despite falling inflation, the AI capex channel is confirmed. Hyperscaler Q3 capex guidance (due next 2 weeks), any deceleration breaks the thesis.

Because the market is pricing a regime change, not a cycle. The Shiller CAPE at 40.88 (99th percentile since 1871) and S&P 500 trailing P/E at 26.33 (93rd percentile) look insane against current earnings, but they're not pricing current earnings. They're pricing the probability that AI delivers a step-function productivity surge that rewrites the neutral rate, the profit share of GDP, and the growth trajectory of the largest companies in history.

The evidence is mixed. Hyperscaler capex at $649B TTM (+88% YoY) is 90.5% of operating cash flow, the highest reinvestment rate in modern corporate history. Nvidia's data center revenue runs at a $127B annual rate with 66% operating margins. Microsoft's Azure grew 37% with AI contributing 12 points.

But Google's cloud grew 35% and the stock got crushed because FCF went negative. Meta's Reality Labs loses $4B+ per quarter. Amazon's AWS grew 37% but retail FCF is negative. The monetization curve is steepening for infrastructure (Nvidia, TSMC, Vertiv, Eaton) but flattening for applications.

What changed since last month: the market stopped giving "trust me" multiples. Alphabet's 12.6% drop was the inflection, the first time a hyperscaler was punished for capex intensity rather than rewarded for growth. Microsoft's 15.5% rally on similar spend shows the differentiator: visible AI revenue (Azure AI services, Copilot) and positive FCF. The new rule: show me the cash conversion, or I'll discount the spend at a high rate.

For a normal person's money: if you own an S&P 500 index fund, you own this bet at 26x earnings. The concentration is extreme, two-thirds of S&P 500 profit growth comes from roughly ten companies.

If the AI productivity surge delivers, the index earns its way to fair value. If it stalls, the multiple compresses 30-40% from here. There is no middle ground priced in.

The view changes if: (1) hyperscaler FCF inflects positive sequentially for two quarters, or (2) the 10-year term premium compresses below 50bps while capex holds, signaling the market believes the spend is productive. Neither has happened.