by AdvisorAnalyst Editorial Staff, November 21, 2026
The next move in equities will be decided by two numbers, and neither of them is a share price. The first is the Federal Reserve's decision in December, where a hike is now widely expected. The second is the 10-year Treasury yield, which spent most of the autumn near 5 percent and closed Friday, November 20, at 4.94 percent. Until the cost of capital stops rising, the market cannot settle on what the AI buildout is worth. Until the buildout shows profit beyond the companies selling the shovels, the market will not pay up for it. Those two conditions, not the headlines, set the path from here.
The headlines were severe enough. Bloomberg's Jeran Wittenstein and Ryan Vlastelica report that the Nasdaq 100 fell 10 percent this week, its worst week since March 2020, and ended Friday on the brink of a bear market. The S&P 500 is in correction, down 12 percent from its October record. Nvidia fell 23 percent for the week and sits 48 percent below its October 6 high, and the Philadelphia Semiconductor Index has lost 41 percent in the second half of the year.¹ Anthropic's record initial public offering was supposed to steady the mood. Instead, the stock finished Friday less than $6 above its offer price.¹
To be clear, a month of research published on AdvisorAnalyst.com anticipated most of the pressure points this week exposed. What follows sets the short-term shock against the longer-term evidence, and separates what has changed from what has only been repriced.
What Did This Week Actually Break?
Three blows landed within days of each other. Nvidia's board ousted Jensen Huang on November 12, and even a raised revenue forecast and a $150 billion buyback could not stop the selling while the question of his permanent successor stays open. The Democratic midterm sweep raised the odds that data centre construction slows. And the Anthropic offering, which, with a concurrent private placement, raised more money than any new issue in history, gave investors a clearer look at AI economics without giving them a reason to buy.¹
Anna Wong of Bloomberg Economics argues the Anthropic deal could mark the end of the AI trade, because the company's financials show weakening pricing power for frontier models.¹ Paul Nolte of Murphy & Sylvest Wealth Management is blunter about the mood: "The market is in limbo-land, especially when it comes to tech and AI. You need to gird your loins and expect a period of volatility and downside risk."¹
What the week did not break is demand. Brian Smedley of Cynosure Strategies argues "The market has overshot to the downside, just as it overshot to the upside earlier this year."¹ Marta Norton of Empower sees the Anthropic listing as a source of clarity rather than a verdict: "It has been hard to get a real feel for AI demand, and Anthropic is the answer to that question."¹
Is This the End of the AI Trade, or the End of the Easy Part?
The institutional research points to the second answer. Goldman Sachs Global Investment Research, in a 2027 outlook led by Kamakshya Trivedi and Dominic Wilson, finds that AI accounts for about 92 percent of U.S. market value gains this year and 62 percent of S&P 500 earnings growth, but only about 21 percent of GDP growth. In the team's words, "the AI impulse has been a much larger contributor to growth in earnings and market value than to growth in GDP." With S&P 500 earnings growth expected to slow from roughly 36 percent in 2026 to 11 percent in 2027, the team warns that "the market is running into more conventional late-cycle headwinds."²
That distinction matters. A slowdown in AI spending is a larger threat to equity indexes than to the economy, which is exactly what this week's price action reflected. Kimberly Woody of Globalt Investments frames the new test plainly: "A good story is no longer enough." She adds, "It is not enough to own a business simply because it has AI exposure."³ The investable question has moved from whether AI demand is real to which companies keep the profits.
Brian Levitt of Invesco reads the midterm backlash in the same light. Voters backed restrictions in virtually all of 41 data centre ballot measures, yet he argues the demand case is unchanged: "What has changed is the timeline for building the infrastructure required to meet that demand." In his view, "The incentives to build remain enormous."⁴ A longer buildout is a different investment than a cancelled one.
Who Sets the Price of Capital Now?
This is the connective tissue running through almost every piece of the past month. Malcolm Polley of Stratos Investment Management puts it simply: "Until we know the cost of capital, we can't adequately value the asset, and if the cost of capital is going higher, then valuations need to come down."¹
Karen Ward of J.P. Morgan Asset Management explains why the bond market now holds that power. "In stark contrast to the decade pre-pandemic, the defining feature of the current economic cycle is a desire to spend," she writes. AI-related debt accounts for more than 20 percent of U.S. dollar investment-grade issuance, and net bond supply is approaching global financial crisis highs. Her conclusion: "Bond investors find themselves in the powerful position of being the ultimate allocator of capital."⁵
The inflation backdrop keeps that price high. Hubert Marleau of Palos Management notes that "Over the past 3 months, prices have now increased at a 5.7% annualised rate," with 10-year yields pushing through 5.15 percent earlier in the month and futures treating a December hike as close to certain.⁶ Yet Komson Silapachai and Thomas Urano of Sage Advisory see the bond market beginning to look past inflation. Treasury yields sit about 50 basis points below their early-October highs, and they argue "markets may be beginning to price slower growth rather than persistent inflation and capital-supply pressures."⁷ Friday's close at 4.94 percent fits that read. The bond market, in other words, is now weighing growth risk as heavily as price risk.
Why Did Diversification Fail to Show Up?
Many clients learned this month that a diversified-looking portfolio can behave like a concentrated one. Michael Clarfeld and Corey Hardie of ClearBridge Investments show that a blend of 50 percent S&P 500, 30 percent Russell 1000 Growth and 20 percent Russell 1000 Value puts about 31 percent of the portfolio in five AI-linked companies. The top 10 S&P 500 holdings now make up 40.6 percent of the index, and the index diversifies like a portfolio of just 41 stocks, down from 138 in 2016.⁸
Mike Green of Tier1 Alpha explains why concentration keeps feeding itself: passive flows, not fundamentals, set prices at the margin. He describes active managers as "the parasitic load on the passive environment" and notes that "it's easier to get a trillion-dollar company public than it is to get a 500 million-dollar company public." The structure, he argues, functions as "a liquidity management exercise, effectively crowd control with water cans."⁹ When the flows reverse, the same plumbing works in the other direction, which helps explain the speed of this week's decline.
The traditional hedges have not filled the gap. Russ Koesterich of BlackRock finds that since late February the energy sector has moved against the S&P 500 with a correlation of about negative 0.6, while rising rates have blunted bonds and gold has offered little protection.¹⁰ That may be starting to change. Sage's team argues that "A shift in the balance of risks could be enough to turn the bond-equity correlation negative again," which would restore Treasuries as a hedge if growth, not inflation, becomes the dominant fear.⁷
What Is the Market Beneath the Index Saying?
The damage has been far narrower than the index suggests. Jurrien Timmer of Fidelity observes that AI names have fallen more than 30 percent while the S&P 500 excluding AI has held near its August high, and he reads the selloff as a rebalancing rather than a break.¹¹ Chris Galipeau of Franklin Templeton Institute sticks with the broadening call: the equal-weight S&P 500 trades at 17.3 times earnings against 20.6 times for the cap-weighted index. "The economy remains resilient," he writes, and "Volatility is your friend."¹²
The economic risk is that this resilience rests on one crop. Northern Trust's Carl Tannenbaum, Vaibhav Tandon and Ryan Boyle find that business investment supplied 1.8 percentage points of the third quarter's 3.4 percent GDP gain. Their warning is direct: "A tech downturn could spur a broad recession if it is the only source of growth; better diversity is needed."¹³ The good news in their review is early evidence of breadth, with 522 projects worth $383 billion spread across metals, energy, pharmaceuticals and defence.¹³
Where Does the Income Case Stand?
For the first time in years, fixed income is paying investors to wait. AllianceBernstein Research notes that "Since the global financial crisis, the US high-yield index has crossed above the 8% threshold only eight times," with returns in the following year averaging about 12 percent.¹⁴ Goldman's team adds that "Bond yields are now providing a good running yield and a decent spread to cash," and forecasts a 4.6 percent 10-year Treasury yield by the end of 2027.²
The caution sits in the issuance. Goldman expects "Further widening in index-level credit spreads in 2027," especially in AI-related debt.² Galipeau's preference for short-duration mandates, corporate credit and municipal bonds reflects the same discipline: collect the income, but keep a close eye on who is borrowing and why.¹²
What Does a Market With More Than One Engine Look Like?
The arrival state the research describes is not a return to AI-led leadership. It is a market where returns come from AI adopters, productivity beneficiaries, non-AI cyclicals and income, rather than from a handful of hardware suppliers. Goldman frames the 2027 question as whether that handoff goes smoothly.² Woody points to semiconductors as "the clearest near-term beneficiaries because they are paid as capacity is built," but expects the next stage to reward companies that turn AI into durable free cash flow.³ Ward sees the market moving "less on the build-out, and more on the rollout."⁵
Mark Luschini of Janney Montgomery Scott captures the near-term stance most strategists share: "We're oversold, but not deeply so. We could plumb lower still. I want to see stabilization before I make a strong commitment."¹ Stabilization, in this framework, is a checkable condition rather than a feeling.
When Is the Right Moment to Reduce Concentration?
The case for less concentration is strong. The case for cutting it this week is much weaker. Portfolios that carry 30 percent or more in a handful of AI names should come down over time, but a forced sale into a washout locks in the damage and gives up the rebound that Smedley expects when the overshoot corrects.¹ The better moment to make these moves is likely to come in the period ahead, once prices stabilize and the cost of capital settles.
That argues for a plan now and a graduated execution later. Clarfeld and Hardie's prescription is to look through the labels and set position limits deliberately rather than by default.⁸ In practice, that means deciding the target weight for AI exposure today, then closing the gap in stages: trimming into strength rather than weakness, directing new contributions, dividends and coupons toward the underweight areas, and using rebalancing bands rather than calendar dates. Tax lots and capital gains belong in the plan from the start, since the cheapest trims are often in lots bought this year at higher prices.
The most durable reason to trim is rooted in the client's plan rather than a market call. David Rosenberg of Rosenberg Research points to baby boomers holding about 55 percent of their wealth in equities at age 70, with household balance sheets roughly 73 percent in stocks and 7 percent in bonds.¹⁵ For many clients, the right concentration level is a function of age, income needs and time horizon, not of where Nvidia trades next week. Framing the reduction as "this is the right size for your plan" holds up far better than "we are calling the top," and it eases the pull of anchoring to October's highs. The two ideas work together. The life-stage case sets the destination; market conditions set the pace. Acting on Rosenberg's point does not mean selling into this week's weakness. It means using the rebounds that follow, when the overshoot Smedley describes begins to correct, to trim concentrated positions into strength. That is also when the plan matters most, because a recovering market is exactly when a client is most tempted to call off the reduction.
The receiving side matters as much as the trimming side. Galipeau's discipline on volatility is a useful timing rule here, since he treats a VIX close above 30 as the point to start adding equity exposure and a close above 50 as the signal to buy aggressively.¹² Equal-weight, value, energy and quality income are the destinations the research names most often. To be clear, the aim is not to abandon AI. It is to own it at a size the client chooses, rather than at the size the index assigns.
Where Does This Leave Allocators?
The current position is clear enough. Mega-cap AI has repriced sharply. The Nasdaq 100 is near a bear market and the S&P 500 is in correction. The equal-weight index, energy and income assets have held up far better. Inflation is running hot and the Fed is expected to hike in December, yet the 10-year yield closed Friday at 4.94 percent, below the 5 percent line it held for much of the autumn. That dip is days old. Whether it holds depends on whether yields bounce off these short-term lows in the coming week and the weeks that follow. The bond market is acting as the gatekeeper for AI funding. Demand for AI has not disappeared. The market is no longer paying for it in advance.
The trigger for change is the cost of capital. If the 10-year yield holds below 5 percent through the December decision and the next inflation prints while credit spreads stop widening, the case for a broader, more balanced recovery strengthens, and the AI leaders can begin to trade on earnings rather than funding fears. If yields climb through 5 percent and spreads keep widening, the funding squeeze tightens, and the burden falls on the companies that need capital markets most. A single close below 5 percent is a start, not a confirmation, and a bounce off the short-term lows would put the line straight back in play. It is simply too early to tell which path wins. The discipline of the allocation process remains unchanged: set concentration targets now, reach them in stages as conditions improve, collect the income on offer, and let the facts set the pace.
5 Key Takeaways
- Watch the 10-year against 5 percent through the December Fed meeting. It closed November 20 at 4.94 percent, but one close is not a trend. Watch whether it bounces off those short-term lows in the week ahead. Holding below 5 percent through the meeting eases the funding pressure on AI capex; a move above 5.15 percent extends it.
- Measure true concentration now, reduce it in stages. Run a look-through on every equity sleeve. If five AI-linked names exceed roughly 30 percent of the total, as ClearBridge finds in a common blend, set a target weight today, then close the gap gradually: trim into rallies, redirect new cash and income, and avoid selling into a washout.
- Track credit spreads as the leading signal. With AI debt above 20 percent of investment-grade issuance, widening spreads will show funding stress before earnings do.
- Use the 8 percent high-yield level as an entry discipline. Historically, crossings above 8 percent have preceded average 12-month returns near 12 percent, but favour quality and shorter maturities while spreads widen.
- Separate the AI builders from the AI profit-takers. Hardware is paid first; software and adopters must prove cash flow. Pair selective AI exposure with equal-weight, energy and value holdings so the portfolio is not relying on one engine.
Notes
- Wittenstein, Jeran, and Ryan Vlastelica. "Wall Street Asks What Can Save AI Trade After Anthropic Wipeout." Bloomberg, 21 Nov. 2026.
- Trivedi, Kamakshya, and Dominic Wilson. "The Late-Cycle Reckoning: Goldman Sachs Maps the Growing Pains of 2027." AdvisorAnalyst.com, Goldman Sachs Global Investment Research, 18 Nov. 2026, https://advisoranalyst.com/2026/11/18/the-late-cycle-reckoning-goldman-sachs-maps-the-growing-pains-of-2027.html/.
- Woody, Kimberly. "The AI Trade Grows Up: Infrastructure Still Leads, but the Proof-of-Profit Era Has Begun." AdvisorAnalyst.com, Globalt Investments, 13 Nov. 2026, https://advisoranalyst.com/2026/11/13/the-ai-trade-grows-up-infrastructure-still-leads-but-the-proof-of-profit-era-has-begun.html/.
- Levitt, Brian. "What a Slower AI Buildout May Mean for Markets." AdvisorAnalyst.com, Invesco, 15 Nov. 2026, https://advisoranalyst.com/2026/11/15/what-a-slower-ai-buildout-may-mean-for-markets.html/.
- Ward, Karen, et al. "The Bond Market Becomes the Gatekeeper." AdvisorAnalyst.com, J.P. Morgan Asset Management, 19 Nov. 2026, https://advisoranalyst.com/2026/11/19/the-bond-market-becomes-the-gatekeeper.html/.
- Marleau, Hubert. "The Inflation Reckoning." AdvisorAnalyst.com, Palos Management, 15 Nov. 2026, https://advisoranalyst.com/2026/11/15/the-inflation-reckoning.html/.
- Silapachai, Komson, and Thomas Urano. "The Inflection Point: Why the Risk Calculus Is Shifting." AdvisorAnalyst.com, Sage Advisory, 12 Nov. 2026, https://advisoranalyst.com/2026/11/12/the-inflection-point-why-the-risk-calculus-is-shifting.html/.
- Clarfeld, Michael, and Corey Hardie. "Diversification Now an Active Decision." AdvisorAnalyst.com, ClearBridge Investments, 15 Nov. 2026, https://advisoranalyst.com/2026/11/15/diversification-now-an-active-decision.html/.
- AdvisorAnalyst Editorial Staff. "Crowd Control With Water Cans: Why Passive Flows, Not Fundamentals, Are Steering Markets." AdvisorAnalyst.com, 10 Nov. 2026, https://advisoranalyst.com/2026/11/10/crowd-control-with-water-cans-why-passive-flows-not-fundamentals-are-steering-markets.html/.
- Koesterich, Russ. "Energy Stocks Still a Portfolio Hedge." AdvisorAnalyst.com, BlackRock, 10 Nov. 2026, https://advisoranalyst.com/2026/11/10/energy-stocks-still-a-portfolio-hedge.html/.
- Timmer, Jurrien. "Jurrien Timmer: Balance of Power." AdvisorAnalyst.com, Fidelity Investments, 11 Nov. 2026, https://advisoranalyst.com/2026/11/11/jurrien-timmer-balance-of-power.html/.
- Galipeau, Chris. "Sticking With the Broadening Call." AdvisorAnalyst.com, Franklin Templeton Institute, 17 Nov. 2026, https://advisoranalyst.com/2026/11/17/sticking-with-the-broadening-call.html/.
- Tannenbaum, Carl, Vaibhav Tandon, and Ryan Boyle. "Concentrating on U.S. Business Investment." AdvisorAnalyst.com, Northern Trust, 18 Nov. 2026, https://advisoranalyst.com/2026/11/18/concentrating-on-u-s-business-investment.html/.
- AllianceBernstein Research. "Is 8% the Magic Number? Signs Point to Yes." AdvisorAnalyst.com, AllianceBernstein, 17 Nov. 2026, https://advisoranalyst.com/2026/11/17/is-8-the-magic-number-signs-point-to-yes.html/.
- Rosenberg, David. "David Rosenberg: Bubbles, Bonds, & Boomers." AdvisorAnalyst.com, Insight Is Capital, Rosenberg Research, 23 Oct. 2026, https://advisoranalyst.com/2026/10/23/david-rosenberg-bubbles-bonds-boomers.html/.