CNBC Halftime Report

2026-11-06 · Hosted by Scott Wapner · CNBC

Executive Summary

The NASDAQ suffered its worst day since April 2025 as the AI trade unraveled, with the SOX semiconductor index having its worst day since the March 2020 COVID crash — down almost 10% from highs, with Micron down 45%, Marvell down 57%, and Qualcomm down 42% from their respective highs. Bespoke noted semiconductors have fallen at least 17.5% in five days only during the Asian financial crisis, the Russian debt default, the dot-com bust, the Global Financial Crisis, and the COVID crash — placing today's move in rare historical company. The panel broadly agreed the midterm election results materially raised the difficulty of building new data centers, citing Meta president Dan McCormick's comments that midterm results "will make building data centers harder," Satori Research's call that the infrastructure trade "as we know it is ending," and Dan Ives tempering his bullishness to say 10% of planned data centers "will not happen."

Key Stories & Changes

1. Historic Semiconductor Sell-Off

  • SOX down almost 10% from highs — worst day since March 2020 (COVID)

  • Micron down 45%, Marvell down 57%, Qualcomm down 42% from their respective highs

  • Bespoke: only five prior periods since 1994 saw semis fall 17.5%+ in five days — Asian financial crisis, Russian debt default, dot-com bust, GFC, and COVID

  • Josh Brown: pain concentrated specifically in AI hardware/infrastructure names, not broad market-wide; Berkshire Hathaway hit a record high the same day

2. Political Commentary Shifts the AI Buildout Narrative

  • Meta president Dean "Powell" McCormick (Wall Street Journal tech conference): midterm results "will make building data centers harder"

  • Satori Research: the infrastructure trade "as we know it is ending," arguing peak capex growth has passed and risk is now to the downside

  • Dan Ives tempered his bullishness, saying 10% of planned data centers "will not happen" and will be "wiped off the roadmap," moving his capex-growth peak estimate from 2029 to 2027

  • Rob Sechan: "nobody voted to make compute less scarce" — argues this is a timing issue, with existing capacity becoming more valuable as new capacity gets harder to build

3. Debate: Is This a Timing Issue or a Demand Collapse?

  • Malcolm Etheridge: roughly 40% of the S&P's value and about 75% of 2026 GDP growth are tied to hyperscaler AI spending; growth concentrated in just three companies (Nvidia, Broadcom, Micron) tied to OpenAI/Anthropic compute commitments

  • Brad Gerstner (cited from the All-In podcast): a pullback in AI infrastructure spend could cost at least 50 basis points of GDP growth next year

  • Sarat Sethi: valuations already reflect skepticism — Nvidia trades at 14x forward earnings and Sandisk at 3.5x, implying the market never fully believed in peak-pace estimates

  • Josh Brown: current environment gives hyperscalers political cover to pull back capex guidance they may have wanted to trim anyway (citing reports that Microsoft's Amy Hood was uncomfortable with aggressive AI spending)

4. Qualcomm CEO: Data Center Revenue Target Is "Supply Bound, Not Demand Bound"

  • Qualcomm shares fell 11% after earnings; CEO Cristiano Amon says the company's $5 billion fiscal 2027 and $15 billion fiscal 2029 data-center revenue targets are unaffected by the broader sell-off

  • Amon: "If I have more supply, I'll probably do better than five billion dollars" — framing demand as essentially unconstrained

  • Qualcomm positions its low-power-consumption chip design (targeting 0.3 picojoules of energy per bit, better than its advertised 0.5) as a differentiator for data-center total cost of operations

  • Qualcomm also building out cellular technology for Starlink's planned nationwide terrestrial wireless network, expected to show up in revenue starting fiscal 2028

  • QCOM: Qualcomm — -11% (-42% from highs) — CEO says data-center targets are supply-, not demand-, constrained

  • NVDA: Nvidia — -8.5% — Trading at just 14x forward earnings; next earnings Nov. 17 seen as key catalyst

  • AMD: AMD — ~-15% (2-day) — Market punishing name further after a deal it already disliked

  • CRWV: CoreWeave — New low — Leverage and customer-concentration concerns intensifying

5. Anthropic IPO Valuation Expectations Slide

  • Kate Rooney: investor expectations for Anthropic's IPO have fallen from roughly $2 trillion (floated about a month ago) toward $1.5 trillion, which one investor called "a stretch" given the current tech discount

  • Brad Gersner, an early private investor who plans to be a large IPO buyer, said he expects pricing at the lower end of the range

  • Midterm-driven data-center backlash and new regulatory scrutiny (explicitly flagged as a risk factor in Anthropic's S-1 filing) complicate the AI labs' ability to forecast future revenue

  • OpenAI released a new "policy blueprint" promising community input, cost support, and scrapping NDAs — seen as damage control after the midterms, though OpenAI says the effort predates the election

6. Panel Stock Moves

  • Malcolm Etheridge bought Qualcomm, citing 60% projected growth in non-handset revenue and 89% automotive growth, viewing it as a "show-me story" similar to UPS's successful pivot away from Amazon dependence

  • Malcolm Etheridge sold American Express, citing unwillingness to hold consumer-exposed names in the current environment despite strong card-member spending growth

  • Sarat Sethi added Medline (healthcare/medical supplies distributor, 10% organic growth, 30% margins) and trimmed UnitedHealth after a strong run

  • Josh Brown highlighted MSCI as a "best stock in the market" idea, calling $600 a key technical level to watch

1. Market Narrows From a Broad AI Rally to a Selective "Personal Pie"

Josh Brown described the shift vividly: the AI trade moved from a "huge pie" where many participants (semis, power names, NeoClouds) could win, to a "personal pie" where only a handful of the most financially resilient names will benefit as compute becomes scarcer and more politically contested to build.

2. Incumbents Benefit From Scarcity; Leveraged Suppliers Suffer

A consistent theme across the panel: companies with existing data-center capacity and diversified cash flows (hyperscalers) stand to benefit from reduced new supply, while suppliers and NeoClouds dependent on continuous new construction face rising funding costs and demand uncertainty.

3. Rotation Into Healthcare and Energy as a Pre-Existing, Not Just Post-Election, Trend

Panelists noted the healthcare rally began weeks before the midterms, suggesting the market had already started pricing in skepticism about AI-driven earnings sustainability — the election just accelerated and made visible a rotation that was already underway.

4. AI IPO Valuations Recalibrating in Real Time

Anthropic's narrowing valuation range illustrates how quickly new-issue pricing can move with sector sentiment — a dynamic that will likely extend to other AI-adjacent IPOs in the pipeline. ---

Sentiment Analysis

Overall Market Sentiment: Divided and Uncertain

The panel was notably split — some treating the sell-off as a healthy, overdue repricing of an overextended trade, others warning that the breadth of economic dependence on AI capex makes this a much bigger problem than a sector correction.

Risk Factors Highlighted

Durable regulatory overhang post-midterms: New Democratic majorities in Congress and key state governments raise the odds of legislation slowing data-center construction.

Concentration of GDP and earnings growth in AI capex: An estimated 75% of 2026 GDP growth and much of S&P earnings growth trace to a handful of AI-spending companies, per Brad Gerstner's estimate.

NeoCloud leverage and funding access: Highly levered NeoClouds without diversified businesses face existential risk if capacity delays persist.

Decelerating AI revenue growth rates: Growth rates are compressing sharply from triple-digit sequential gains, pressuring valuations across the sector.

Anthropic IPO valuation and forecasting risk: Regulatory uncertainty makes it difficult for AI labs to issue credible forward revenue guidance as public companies.

Hyperscaler capex guidance cuts: Companies may use the political environment as cover to pull back previously aggressive spending commitments.

No confirmed technical bottom: Lack of capitulation signals (VIX levels, correlation spikes) suggests further downside risk before stabilization.

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.

Keep Reading