CNBC The Exchange
2026-11-02 · Hosted by Kelly Evans · CNBC
Executive Summary
Broadcasting live from the independent "Shadow Fed" conference in Manhattan, CNBC's The Exchange combined deep monetary-policy analysis with real-time coverage of a brutal tech selloff. The Nasdaq fell nearly 3% to session lows as Intel's almost-30% drop — its worst day since 1974 — rippled through chip names, while Amazon, Meta and Alphabet together lost $750 billion in combined market cap since Monday. Oaktree's Howard Marks delivered the episode's intellectual centerpiece, arguing AI is "the least knowable, least predictable thing" he's witnessed in his career, and that if investors want AI exposure they should buy stocks rather than debt, since debt offers no upside beyond repayment.
Key Stories & Changes
1. Tech Selloff Deepens Live From the Shadow Fed Conference
Nasdaq down nearly 3%, at session lows during the broadcast
S&P 500 down 1.8%
Intel down almost 30%, worst day since 1974, driven by memory costs, supply constraints and weak PC demand
Neo-cloud/memory-adjacent names also fell sharply: down 11-13% across the group
Amazon, Meta and Alphabet lost a combined $750 billion in market cap since Monday
Four Dow components (Chevron, Visa, Amgen, JNJ) hit record highs even as tech fell
2. Howard Marks (Oaktree): AI Is the Least Predictable Thing He's Seen
Said he cannot determine how any single AI company's product will be used, priced, or monetized five years out — "unusual uncertainty," though he noted others might frame it as "unusual upside potential"
Advised that investors wanting AI exposure should buy equities, not credit, since debt offers no upside beyond interest and principal repayment
Oaktree, now part of Brookfield, is focused on AI infrastructure (data centers, power) rather than cloud business models directly, prioritizing less-crowded markets
Flagged power as "going to be a real bottleneck" as data-center infrastructure needs to scale three to ten times
Estimated the AI investment cycle is in roughly the "third inning"
Said he doesn't believe the Fed should try to prevent irrational exuberance, calling "occasional hiccups" (like this week's selloff) healthy
3. Fed Speakers Turn Hawkish at the Shadow Fed Conference
Dallas Fed President Lorie Logan: said the Fed needs at least two more quarter-point hikes before pausing, calling current policy not yet "modestly restrictive"; cited inflation that would "continue to run above target" without further tightening
St. Louis Fed President Alberto Musalem: said "additional firming of monetary policy will be necessary and soon," describing financial conditions as still accommodative despite recent tightening
Former Richmond Fed President Jeffrey Lacker: said he was "surprised at the lack of urgency" given inflation near 3.5% for an extended period, and floated nominal GDP targeting as a framework that could gain more prominence
Former St. Louis Fed President James Bullard: agreed nominal GDP targeting deserves more attention, but cautioned forecasting remains inherently imperfect
The odds of a third rate hike next year fell below 50% following the tech selloff, even as Fed speakers leaned hawkish
4. Lou Crandall: Treasury Debt Issuance in an Unusually Unpredictable Period
Rates & I-CAP chief economist said he has "no strong view" on where yields top out, focusing instead on how much the Treasury will sell
Flagged the Treasury's August decision to triple long-end buybacks outside its normal announcement cadence as "extraordinarily out of character," damaging predictability investors rely on
Said it's "a virtual certainty" coupon offering sizes will stay flat near-term, but noted increasing room to raise shorter-duration (2/3/5-year) auction sizes given sustained high deficits
Said he was "frankly terrified" by chatter about a new Fed-Treasury Accord, warning against anything that blurs monetary-fiscal independence
5. Jason Trenner (Baird): Anti-Data-Center Politics a Real, Underappreciated Risk
Said data centers poll roughly 80-20 unfavorable and AI overall 70-30 unfavorable with the American public, a dynamic he observed firsthand speaking with officials in Texas and Ohio
Called this political backlash "throwing sand in the gears" of the AI buildout — potentially a long-term positive by slowing excess CapEx, but a real near-term headwind
Said free cash flow is "deteriorating, if not negative in many places," forcing hyperscalers increasingly to the capital markets to sustain spending
Still constructive on equities generally, favoring the equal-weight S&P 500 over market-cap-weighted exposure given tech concentration risk
Flagged the next trade opportunity as companies that best *implement* AI (healthcare, financial services, industrials) rather than the infrastructure providers themselves
6. Fed Dissents and Communication Strategy Debate
Two Fed officials — Cleveland's Beth Hammack and Governor Lisa Cook (her first dissent since 2022) — dissented in favor of a quarter-point hike at the latest meeting
Former Fed presidents Jeffrey Lacker and James Bullard debated Chair Kevin Warsh's preference for less forward guidance, with Lacker saying the September meeting's dot plot was effectively "the most forward guidance of any meeting"
Both agreed the Fed should shift focus toward communicating how policy will react to data rather than pre-committing to a path
7. Binance $6.3 Billion Hack Referenced in News Update
CNBC's news update confirmed Binance's disclosure of a $6.3 billion crypto hack, the largest on record, with blockchain intelligence firm TRM Labs pointing to North Korean hackers
Binance pledged to cover losses via its user protection fund established in 2018
Trends Identified
1. Power, Not Capital or Demand, Is the AI Buildout's True Bottleneck
Howard Marks's framing — that demand for AI infrastructure vastly outstrips supply, and that the scarce input is power generation capacity rather than money — reinforces a theme echoed elsewhere this week: the next phase of AI infrastructure risk is physical and regulatory, not financial.
2. Credit Markets Increasingly Skeptical of AI-Linked Debt Quality
Marks's warning against lending into AI uncertainty, combined with broader market chatter about Anthropic's credit rating prospects, suggests fixed-income investors are beginning to demand a risk premium for AI-adjacent debt that equity investors have been slower to price in.
3. Fed Hawkishness Collides With a Fragile AI-Led Equity Market
Multiple Fed speakers (Logan, Musalem) pushing for further tightening just as AI-related equities sold off sharply illustrates a growing policy tension: the Fed's inflation fight is being conducted against a backdrop where AI CapEx is simultaneously driving growth and threatening financial-market stability.
4. Political Backlash Against Data Centers Becomes a Market-Relevant Risk
Jason Trenner's on-the-ground reporting of 70-30/80-20 public opposition to AI and data centers signals that electoral politics — not just earnings or Fed policy — could materially constrain the AI buildout's pace, a risk investors had largely discounted until now. ---
Sentiment Analysis
Overall Market Sentiment: Hawkish Caution
The combination of a sharp tech selloff and uniformly hawkish Fed commentary from multiple conference speakers created a notably more cautious tone than typical market coverage.
Risk Factors Highlighted
AI infrastructure's unknowable long-term economics: Howard Marks explicitly said no one can reliably forecast product use, pricing, or margins five years out.
Power supply bottleneck: Data-center expansion plans of three-to-ten times current capacity face a real constraint in available power generation.
AI-linked credit risk: Marks warned that lending into AI infrastructure without high confidence of repayment is unwise; Anthropic's expected sub-investment-grade rating reinforces this.
Hawkish Fed policy path: Multiple Fed speakers calling for additional tightening threatens to compound pressure on an already-selling-off tech sector.
Treasury issuance unpredictability: The Treasury's unusual August move to triple long-end buybacks outside normal cadence has damaged market confidence in debt-management predictability.
Political backlash against data centers: 70-30/80-20 public opposition to AI/data centers could translate into policy headwinds after the midterms.
Deteriorating hyperscaler free cash flow: Broad-based FCF deterioration is forcing continued reliance on capital markets to fund AI buildout.
Fed-Treasury coordination concerns: Crandall's explicit fear of a new "accord" that would blur monetary-fiscal independence.
Record crypto hack: The $6.3 billion Binance theft, the largest in crypto history, underscores unresolved cybersecurity risk in the asset class.
This episode was covered in today's [The Market Signal — 2026-11-02](https://marketsignal.beehiiv.com/p/the-market-signal-2026-11-02), a cross-source synthesis of multiple podcast reports.