Prof G Markets: JPMorgan’s Playbook for a 10-15% Correction (or Worse) — ft. Michael Cembalest

Podcast: Prof G Markets (Vox Media Podcast Network)
Hosts: Scott Galloway, Ed Elson
Guest: Michael Cembalest, Chairman of Market and Investment Strategy, J.P. Morgan Asset & Wealth Management
Date: November 21, 2025

Episode Overview

In this episode, Scott Galloway and Ed Elson tackle the mounting anxiety in financial markets regarding a potential correction—especially amid fears of a tech and artificial intelligence (AI) investment bubble. They are joined by Michael Cembalest, a top strategist at J.P. Morgan, to discuss whether today's market exuberance, centered around the "Mag 10" (top 10 tech and AI stocks), is heading for a significant pullback. The conversation ranges from comparisons to past bubbles, portfolio defense strategies, and the geopolitics of tech, to the risks and realities underpinning the AI-driven rally.

Key Discussion Points & Insights

1. Bearish Market Concerns: Setting the Table

[06:44–10:56]

• The episode opens by referencing a recent highly bearish outlook from Professor Aswath Damodaran.
  • Damodaran’s fear: “There’s no place to hide in stocks. …if the Mag 10 go down by 40%, it’s going to ripple through stocks.” ([07:19])
  • He suggests even considering collectibles/cash, unprecedented for him.
• Cembalest’s response: Skeptical about academic forecasts without real-world track records. Instead points to indicators like Berkshire Hathaway’s cash stockpile as more meaningful than discourse from those not managing actual portfolios ([08:24]).
  • “Professors are basically running fantasy baseball teams by coming out intermittently and telling you what their trades are. …It’s not real money, not real life.” —Michael Cembalest ([08:24])
• Current J.P. Morgan portfolios already hold 30–40% in cash and equivalents, suggesting a “defensive” stance but not outright panic ([10:41]).

2. Is There an AI Bubble? Tech’s Moonshot Spending

[11:49–14:18]

• 75% of post-2022 revenue/capex growth is from ~40 AI-related stocks.
• Cembalest is particularly unsettled by Meta’s commitment to spend 65–70% of revenues on capex/R&D for AI, paralleling their earlier metaverse ambitions.
  • “As an investor, that doesn’t fill me with a lot of good feeling about where we’re going to be two years from now.” ([11:59])
• Tech capex relative to U.S. GDP now matches major historic undertakings: “the moon landing, the Manhattan Project, the highway system… and the Hoover Dam combined.” ([13:10])
  • Memorable moment: Galloway’s amazement at that scale — “Wow.” ([13:13])
• Positives: Unlike early-2000s bubbles, current tech spending is mostly internally financed (except Oracle), not reliant on debt, so it could last longer but will still require eventual profits ([13:40]).

3. Market Fragility: Can Mega Tech Drag Down the Market?

[14:18–18:35]

• Galloway outlines a “house of cards” scenario: If just a few major companies roll back on AI, or if OpenAI/Nvidia experience a severe correction, $3 trillion could vanish from markets, triggering flat GDP and systemic fragility.
  • “If Nvidia… goes down 60%... we have a $3 trillion destruction in the capital markets… America has become so fragile because it is now a giant bet on AI.” —Scott Galloway ([15:10])
• Cembalest agrees in part: Yesterday, tech investment was a rounding error for GDP; now, it’s a third of all recent GDP growth. Much current economic strength is “bailed out” by generative AI investment ([16:49]).
• Risks are partially contained because most tech firms (sans Oracle) are funding spending with cash, not debt.

4. AI Capex and Financial Engineering — Blue Owl, Meta, SPVs

[19:22–22:02]

• Discussion of how some firms (like Meta) are using “off-balance sheet” SPVs to shift capex risk onto private lenders (e.g. Blue Owl), rather than absorbing the debt themselves.
  • “The Meta partnership with Blue Owl… [is] not consolidated onto Meta’s balance sheet… It’s worse. They basically have walkaway rights every four years… So Blue Owl’s holding the bag.” —Michael Cembalest ([20:16])
• While there is “exuberance and hype” around AI, most major players are not yet addicted to debt, which limits systemic danger compared to past bubbles.

5. Probability and Severity of Correction

[22:02–23:26]

• Cembalest’s forecast: A 10–15% correction in 2026 is likely; much larger (40%) drawdowns would require extreme circumstances.
  • “I’d be really surprised not to see [a 10-15% correction].” —Michael Cembalest ([22:45])
• The severity of the correction (12% vs. 40%) is the key strategic question for investors and pension funds.

6. The Role of Government and Industrial Policy in the Bubble

[26:27–30:33]

• Galloway posits that because AI is propping up the economy, the government might intervene to prop up these tech giants, inflating the bubble further (echoing socialism).
• Cembalest counters: So far, interventions (e.g., with MP Materials and Intel) are mostly industrial policy to shore up critical infrastructure, not blanket support for big tech. Expects more (targeted) interventions, which he largely supports given the threat from China and supply chain vulnerabilities ([28:31]).

7. Portfolio Defense: How JPMorgan Is Positioning

[31:07–32:39]

• Rather than radically altering asset allocations, J.P. Morgan encourages clients to shift down the risk spectrum (e.g. from “growth” to “balanced”), rather than change portfolio composition entirely.
• Key features of a “defensive” portfolio: 30-40% in cash/short-term paper, heavily diversified hedge funds, municipals, preferreds, with selective equity overweight in underpriced sectors like healthcare ([32:51]).

8. Tech Valuation: Context in History and ‘Real’ Risk Premia

[33:48–36:29]

• Tech’s high PE ratios are justified—at least in part—due to exceptional margins and capital efficiency compared to other sectors, even pre-AI boom ([33:53]).
• Cembalest is dismissive of using the equity risk premium as a blunt-edged tool; sector nuances matter more.

Memorable Quote on Tech Valuation:

“Just to look at PE differentials is really missing the point… before this generative AI boom… tech had double the margins of the rest of the market.” —Michael Cembalest ([33:53])

9. Comparisons to 2001 & 2008—How This Cycle Is Different

[36:29–39:49]

• 2001: Obvious bubble—companies had no earnings, discipline was easy for managers.
• 2008: Deep financial system crisis, hidden risks everywhere.
• Today: Big tech is genuinely profitable and innovative, but overextension looms—especially as resource constraints (like power) and capital needs for AI buildout become daunting.
  • “We’re getting closer to a power wall that will prevent OpenAI from getting anywhere near [its goals]… announcements require 30 gigawatts of power, equivalent to 16 Hoover Dams. It’s just not going to happen.” —Michael Cembalest ([38:41])

10. Labor Market Disruption or Market Repricing?

[41:52–46:36]

• Galloway’s forked path: Either AI’s promised cost savings lead to chaotic labor market layoffs, or the valuations are halved as revenue doesn’t materialize.
  • “Either… chaos in the labor markets for the next three years, or these companies adjust down their expectations and valuations.” —Scott Galloway ([43:33])
• Cembalest: Agrees there’s little hard data connecting AI to actual profitability/productivity at scale. Most labor impact evidence is anecdotal or preliminary.

11. Geopolitical Risks: China, Taiwan, and the US Tech Moat

[49:06–52:40]

• Cembalest lists four looming risks for 2026:

  1. Power constraints on data center/AI growth.
  2. China achieving AI parity and “dumping” LLMs cheaply on the global market.
  3. Diminished dependence on Taiwan, increasing geopolitical instability.
  4. The risk that investors en masse say, "we can't see the ROI," prompting widespread profit-taking.
  • “China is on a really long-term journey and I would agree that that is very much explicitly their goal [to catch and undercut US AI].” —Michael Cembalest ([51:50])

12. Global Fragmentation, Inflation, and Fed Policy

[54:53–55:12]

• Fed faces a “fork in the road”: Input prices rising, but wage/labor data softening; yet, the Fed appears poised to cut rates anyway—an unusual and risky move.
  • “The gamble is that the inflationary increase is Tariff related… but that’s a big bet.” ([55:12])
• Immigration and labor supply: Restrictive policy may worsen inflation; US needs more workers to keep wage pressures at bay.

13. Advice for ‘Everyday’ Investors for 2026

[60:43–62:43]

• Cembalest’s advice: Accumulate dry powder (cash) slowly—the next correction will be an opportunity. Many drawdowns are V-shaped, so having cash on hand is valuable for nimble investors.
  • “Having some spare cash and credit to be able to draw on when these things happen can be helpful… individual investors have an advantage over some of the large institutional investors.” ([61:23])
• His bull case: Real, revenue-driven AI productivity emerges, China cuts back on overproduction, Europe engages in deficit spending for defense. US inflation softens, housing and energy prices moderate. ([62:43–64:29])

Notable Quotes & Memorable Moments

• On the defensiveness of today’s portfolios:
  “Our normal balanced and conservative portfolios already have 30-40% in cash, cash equivalents, gold, very diversified hedge funds, municipals…”
  —Michael Cembalest ([10:41])

• On AI capex scale:
  “The tech capital spending in 2025 is equal relative to GDP to the moon landing, the Manhattan Project… and the Hoover Dam combined.”
  —Michael Cembalest ([13:10])

• On systemic fragility:
  “There’s the S&P 490 and there’s the S&P 10. And right now everything is banking on these ten companies living up to these extraordinary expectations.”
  —Scott Galloway ([15:10])

• On Meta’s off-balance-sheet debt:
  “They have walkaway rights every four years… Blue Owl’s holding the bag. So Blue Owl is at risk. …It’s essentially releasing risk the likes of which you’ve seen forever in commercial real estate.”
  —Michael Cembalest ([20:16])

• On plausible corrections:
  “It would be kind of shocking if you didn’t have some kind of profit taking correction in 2026… on the order of 10-15%.”
  —Michael Cembalest ([22:45])

• On AI’s actual economic impact:
  “There’s only a handful [of studies]… on the productivity consequences of doing this stuff. And what you can’t find at all is… explicit pathways to profitability for generative AI adoption for the hyperscalers.”
  —Michael Cembalest ([45:38])

• On China’s ambitions:
  “China’s on a really long-term journey and I would agree that that is very much explicitly their goal [to match US AI dominance].”
  —Michael Cembalest ([51:50])

Timestamps for Key Segments

• [06:44–10:56] — Debate over Aswath Damodaran’s bearishness and real-world vs. theoretical investment advice
• [11:49–14:18] — Discussion of AI boom: moonshot capex, Meta’s spending, sustainability of investment
• [14:18–18:35] — Market fragility if a top AI stock collapses; “house of cards” scenario
• [19:22–22:02] — Financial engineering: how Big Tech is moving capex risk off-balance-sheet
• [22:02–23:26] — Probability/scale of a major correction in 2026
• [26:27–30:33] — Would government intervention drive bigger excesses in Big Tech/AI?
• [31:07–32:39] — How institutional portfolios are adjusting (or not) for risk
• [33:48–36:29] — Valuing tech correctly: growth/margins vs. raw PEs, historic context
• [36:29–39:49] — 2001 vs. 2008 vs. today’s market; new risks (e.g. data center power)
• [41:52–46:36] — Labor market risks and the productivity “pyramid” of AI adoption
• [49:06–52:40] — China’s AI parity/“dumping,” Taiwan/semis risk, power wall as 2026 hazards
• [54:53–55:12] — Global fragmentation, inflation, and the Fed’s “big bet”
• [60:43–62:43] — Practical investor advice: accumulate cash, V-shaped recoveries, bull-case primer

Tone & Closing Impressions

The discussion is nuanced, data-driven, and candid, with a touch of Prof G’s signature irreverence. Cembalest is “measured, not panicked,” recognizing risks but distinguishing between 10–15% healthy market corrections and catastrophic systemic collapses. Galloway and Elson wrestle with headline risks, systemic overconcentration in mega-cap tech, and the uncertainty of whether today’s “AI moment” will look like the dot-com or something different.

As Cembalest puts it:

“Right now, with a little bit of Fed easing and some steady momentum, I’m more inclined to think of the 12 to 15 and the 40.” ([23:26])

For investors and market-watchers, this episode offers a roadmap to defense, a refresher on economic history, and a reality check on the new AI-fueled market paradigm.