CNBC The Exchange
2026-11-06 · Hosted by Kelly Evans · CNBC
Executive Summary
The Nasdaq fell as much as 5.3%, more than 10% below its record high set last month, as the AI trade "suddenly imploded" a day after a post-election relief rally. The SOX semiconductor ETF fell nearly 11% — its worst day since COVID hit in March 2020 — with Sandisk down 20%, AMD down 15%, and Qualcomm down 11%. Treasury yields fell sharply in sympathy, with the 10-year down a full half point from its highs a month ago to 4.18%. Larry Adam of Raymond James argued this looks like the tariff-turmoil pattern from last year — indiscriminate selling that prices in worst-case outcomes before fundamentals are even known — and said he'd use the sell-off to rotate out of richly valued healthcare and back into tech, semiconductors, and hyperscalers.
Key Stories & Changes
1. AI-Driven Market Rout
Nasdaq down as much as 5.3%, more than 10% below its record high from last month
S&P 500 down 4% to 7,381; Dow down about 3%; Russell 2000 down 4.5%
Nvidia down almost 9% to $182; Qualcomm down 11%; AMD down 15%; SOX ETF down nearly 11% (worst day since March 2020)
Sandisk down 20%, Seagate down 15%, Micron down 15% — memory names hit hardest
CoreWeave down 13%, Iren down 16% — NeoClouds extending recent declines
10-year Treasury yield fell to 4.18%, down a full half point from highs a month ago
2. Larry Adam (Raymond James): "This Reminds Me of Tariff Turmoil"
Argues the sell-off mirrors last year's tariff panic, where investors "shot first and asked questions later," pricing in worst-case scenarios that later proved overblown
Sees AI infrastructure names trading at a roughly 30% discount to historical levels with some of the healthiest next-year earnings estimates, calling this a buying opportunity
Trade idea: rotate out of healthcare, calling valuations "the most expensive in 20 years" and pricing in an unlikely full reversal of ACA/Medicaid cuts; rotate into mega-cap tech, semiconductors, hyperscalers, and cybersecurity
Expects energy prices to fall toward $65/barrel over 12 months on an expected Iran détente, benefiting industrials and transports
Acknowledged one dissenting data point: an industrials analyst who correctly called the Caterpillar downgrade over the summer based on on-the-ground political pushback to data centers
3. Stacy Rasgon (Bernstein): Fundamentals Still Solid, Sentiment Collapsed
Says chip fundamentals have actually improved since the June highs even as the stocks fell — this is "a very rapid pivot in narrative and sentiment," not a fundamentals-driven move
Notes much of the political rhetoric is bipartisan and focused on guardrails/community conditions rather than halting investment outright, which could actually smooth out the AI capex cycle rather than kill it
Weakness has been concentrated in PCs and smartphones (Intel hit hard), ironically because AI demand is consuming the memory supply those products need — a sign of AI strength, not weakness
Nvidia reports November 17; Broadcom reports early December — seen as key tests of whether data-center demand commentary can stabilize sentiment
On AMD's outsized 15% drop despite strong earnings: Rasgon called it surprising given solid data-center CPU/GPU results and strong 2027 visibility
4. Microsoft-Enscale Headline Rattles NeoClouds
Reuters reported Microsoft declined to expand its cloud computing deal with Enscale at a Texas data center
John Ford's "sushi" analogy: NeoClouds built with leverage need fast utilization to generate cash flow to service debt; if a major anchor customer like Microsoft won't expand orders, "who's going to eat it?"
Diversified players (hyperscalers, SpaceX) can absorb a slowdown because they have other revenue sources; NeoClouds specializing purely in data-center capacity cannot
5. Private Credit and Alternative Asset Managers Hit Hard
Goldman Sachs down 6%, Morgan Stanley down 6%, Blue Owl down 7% — alternative asset managers tied to AI financing sold off sharply
Research from Americans for Financial Reform: private equity firms back an estimated 25-30% of active US data-center capacity; 78 PE firms back 13-17 gigawatts across 670 facilities, with 26.8 more gigawatts announced — over half of the 50 gigawatts planned by 2030
Largest private-equity data-center owners named: Brookfield, Ares, Blue Owl, Digital Bridge, Apollo
Leslie Picker: delays or moratoriums directly raise the cost of capital and extend repayment timelines for these backers
6. Anthropic IPO Valuation Compresses
Kate Rooney: Anthropic's floated IPO valuation has fallen from roughly $2 trillion to closer to $1.5 trillion given the tech discount and data-center policy pushback
About 30% of new AI/data-center projects are located in Texas, now under a Democratic governor, adding uncertainty for AI labs reliant on growing compute capacity
OpenAI released a new policy blueprint today promising community engagement and dropping NDAs — described as "damage control," though OpenAI says work began in August
Risk flagged: the gap between AI labs (reliant on new compute) and incumbent big tech (which can retrofit existing data centers) could narrow, reducing the labs' competitive edge
7. Victoria Green (G Squared): Buying the Dip Selectively
CIO Victoria Green frames this as an "overreaction" and "recalibration," not a fundamental deterioration, citing the S&P's 200-day moving average near 7,020-7,120 as key support
Notes buying after a 4%+ drop has historically returned about +180 basis points the next day in 57 instances since 1950
Favors memory stocks (Sandisk) given persistent memory-crunch commentary across earnings calls; sees chips moving into oversold RSI territory
Does not see a recession signal yet; expects data centers to still get built, just "more orderly, more planned, and a little more regulated"
8. Carter Worth: Technical Context for the Decline
Carter Worth (Worth Charting): a 4%+ single-day S&P drop has happened only 45 times in 30 years (about 0.6% of trading sessions), with 30 of those 45 instances clustered around the 2008 financial crisis and COVID
Notes market breadth had already been deteriorating for weeks — 35-40% of S&P constituents were already down more than 25% from 52-week highs even as the index itself was only down 4-5%
Warns that defensive-sector spikes (staples, utilities) "can't continue" as a substitute for leadership from the hardest-hit sectors
Trends Identified
1. Sentiment Collapsed Faster Than Fundamentals Changed
Multiple guests (Larry Adam, Stacy Rasgon) emphasized that earnings and demand data for AI/data-center names have not meaningfully deteriorated — what changed overnight was investor willingness to extrapolate political risk into future cash flows, a classic pattern seen previously during the tariff turmoil.
2. Financing Structure, Not Demand, Is the Real Pressure Point
John Ford's "sushi" framing and the private-credit data from Leslie Picker both point to the same mechanism: leveraged data-center projects need fast utilization to service debt, so any delay in construction or customer commitment disproportionately threatens the financing layer (NeoClouds, private equity backers) rather than underlying AI demand.
3. Divergence Between AI Infrastructure "Haves" and "Have-Nots" Widens
Consistent with other reports today, guests distinguished diversified, cash-generative players (hyperscalers, Nvidia relative to peers) from narrowly focused, leveraged players (NeoClouds, private-credit-backed data centers) — the latter bearing disproportionate risk from any buildout slowdown.
4. Policy Risk Reshaping AI-Sector IPO Economics in Real Time
Anthropic's valuation compression from $2 trillion to $1.5 trillion illustrates how quickly new-issue pricing adjusts to sector sentiment and regulatory headlines, a dynamic likely to extend to other AI-related IPOs. ---
Sentiment Analysis
Overall Market Sentiment: Shaken, with a Contrarian Buy-the-Dip Undercurrent
Despite the severity of the decline, several guests actively framed this as a buying opportunity rather than the start of a deeper bear market, while flagging specific pockets (NeoClouds, private credit, consumer-exposed data-center financiers) as genuinely impaired.
Risk Factors Highlighted
Durable political overhang on data-center permitting: A Democratic sweep, including in Texas, raises the odds of slower approvals and new guardrails.
NeoCloud and private-credit financing stress: Leveraged data-center backers (NeoClouds, Blue Owl, KKR, Apollo) face higher financing costs and delayed repayment if projects slow.
Customer concentration risk: Microsoft's reported pullback from an Enscale deal illustrates how dependent NeoClouds are on a small number of anchor hyperscaler customers.
Anthropic IPO valuation and timing risk: Compressed valuation expectations reflect broader uncertainty about AI labs' ability to forecast revenue amid policy change.
Overextended healthcare valuations: Larry Adam warns healthcare pricing already assumes an unlikely full reversal of ACA/Medicaid cuts.
Deteriorating market breadth pre-dating the sell-off: Carter Worth's data shows a third-plus of S&P names were already down more than 25% from highs before today's headline decline.
Memory supply constraints spilling into consumer tech: AI-driven memory shortages are weakening PC and smartphone demand, a secondary effect of the AI buildout.
Energy/geopolitical uncertainty tied to Iran: Adam's bullish oil-price call depends on an anticipated détente that is not yet confirmed.
This episode was covered in today's [The Market Signal — 2026-11-06](https://marketsignal.beehiiv.com/p/the-market-signal-2026-11-06), a cross-source synthesis of multiple podcast reports.